Autostart AS-1535 SH Technical data

Autostart AS-1535 SH Technical data

Oracle® TimesTen In-Memory Database

Troubleshooting Procedures Guide

Release 11.2.1

E13075-05

April 2010

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide, Release 11.2.1

E13075-05

Copyright © 2006, 2010, Oracle and/or its affiliates. All rights reserved.

This software and related documentation are provided under a license agreement containing restrictions on use and disclosure and are protected by intellectual property laws. Except as expressly permitted in your license agreement or allowed by law, you may not use, copy, reproduce, translate, broadcast, modify, license, transmit, distribute, exhibit, perform, publish, or display any part, in any form, or by any means.

Reverse engineering, disassembly, or decompilation of this software, unless required by law for interoperability, is prohibited.

The information contained herein is subject to change without notice and is not warranted to be error-free. If you find any errors, please report them to us in writing.

If this software or related documentation is delivered to the U.S. Government or anyone licensing it on behalf of the U.S. Government, the following notice is applicable:

U.S. GOVERNMENT RIGHTS Programs, software, databases, and related documentation and technical data delivered to U.S. Government customers are "commercial computer software" or "commercial technical data" pursuant to the applicable Federal Acquisition Regulation and agency-specific supplemental regulations. As such, the use, duplication, disclosure, modification, and adaptation shall be subject to the restrictions and license terms set forth in the applicable Government contract, and, to the extent applicable by the terms of the Government contract, the additional rights set forth in FAR 52.227-19, Commercial

Computer Software License (December 2007). Oracle USA, Inc., 500 Oracle Parkway, Redwood City, CA

94065.

This software is developed for general use in a variety of information management applications. It is not developed or intended for use in any inherently dangerous applications, including applications which may create a risk of personal injury. If you use this software in dangerous applications, then you shall be responsible to take all appropriate fail-safe, backup, redundancy, and other measures to ensure the safe use of this software. Oracle Corporation and its affiliates disclaim any liability for any damages caused by use of this software in dangerous applications.

Oracle is a registered trademark of Oracle Corporation and/or its affiliates. Other names may be trademarks of their respective owners.

This software and documentation may provide access to or information on content, products, and services from third parties. Oracle Corporation and its affiliates are not responsible for and expressly disclaim all warranties of any kind with respect to third-party content, products, and services. Oracle Corporation and its affiliates will not be responsible for any loss, costs, or damages incurred due to your access to or use of third-party content, products, or services.

Contents

Preface

................................................................................................................................................................. ix

Audience.......................................................................................................................................................

ix

Related documents......................................................................................................................................

ix

Conventions .................................................................................................................................................

ix

Documentation Accessibility..................................................................................................................... x

Technical support........................................................................................................................................

xi

What’s New

...................................................................................................................................................... xiii

New features for release 11.2.1.6............................................................................................................. xiii

New features for release 11.2.1................................................................................................................ xiii

1 Tools for Troubleshooting TimesTen

Using the ttIsql utility

.............................................................................................................................

1-1

Using the ttStatus utility

.........................................................................................................................

1-2

Using the ttCapture utility

......................................................................................................................

1-5

Using the logs generated by the TimesTen daemon

.........................................................................

1-6

Using the ttTraceMon utility

..................................................................................................................

1-6

Starting a trace and reading the trace buffer..................................................................................

1-7

SQL tracing..........................................................................................................................................

1-9

API tracing .......................................................................................................................................

1-10

DEADLOCK tracing .......................................................................................................................

1-10

LOCK tracing...................................................................................................................................

1-12

ERR tracing.......................................................................................................................................

1-13

AGING tracing ................................................................................................................................

1-14

AUTOREFRESH tracing.................................................................................................................

1-16

Using the ttXactAdmin utility

.............................................................................................................

1-19

Using ODBC tracing

.............................................................................................................................

1-20

Using SNMP traps to detect events

...................................................................................................

1-20

Monitoring the TimesTen system tables

..........................................................................................

1-20

Using the query optimizer

...................................................................................................................

1-21

2 Troubleshooting TimesTen Applications and Data Stores

Unable to start or stop TimesTen daemon

..........................................................................................

2-1

No response from TimesTen daemon or subdaemon

.......................................................................

2-2

Check the TimesTen user error log..................................................................................................

2-2 iii

iv

Extract a stack trace from the core file ............................................................................................

2-2

Unable to create shared segment

........................................................................................................... 2-3

Application unable to connect to data store in direct mode

............................................................ 2-3

Upgrading your data store ...............................................................................................................

2-4

Privileges to connect to data store ...................................................................................................

2-4

Check file system permissions to access data store.......................................................................

2-4

Check that the TimesTen daemon is running ................................................................................

2-4

Check DSN definition........................................................................................................................

2-4

Check DSN attributes .................................................................................................................

2-4

Check path name to data store and transaction log directories...........................................

2-4

Manage semaphores and shared memory segments....................................................................

2-5

Check available swap space (virtual memory) ..............................................................................

2-5

Increase the number of available file descriptors..........................................................................

2-6

Troubleshooting Client/Server problems

............................................................................................ 2-6

Cannot connect to the TimesTen Server .........................................................................................

2-6

TimesTen Server failed .....................................................................................................................

2-7

Cannot find Server DSN....................................................................................................................

2-7

TimesTen Server failed to load DRIVER ........................................................................................

2-7

Application times out when accessing TimesTen Server.............................................................

2-8

TimesTen Client loses connection with TimesTen Server ...........................................................

2-8

Failed to attach to shared memory segment for IPC.....................................................................

2-8

Increasing the maximum server connections on Windows XP...................................................

2-8

Thread stack overflow when using multiple client connections.................................................

2-8

Out of space when DSN specifies new data store .........................................................................

2-9

Application connects or disconnects are slow

.................................................................................... 2-9

Check if data store is being recovered ............................................................................................

2-9

Check ODBC tracing..........................................................................................................................

2-9

Application becomes disconnected unexpectedly

............................................................................. 2-9

Check for ODBC or JDBC errors...................................................................................................

2-10

Check the user error log .................................................................................................................

2-10

Application is slow

............................................................................................................................... 2-10

Consider connection mode ............................................................................................................

2-11

Update statistics for your tables....................................................................................................

2-11

Verify lock and isolation levels .....................................................................................................

2-12

Check trace settings ........................................................................................................................

2-12

Check partition counts for the tables............................................................................................

2-12

Application unresponsive, appears hung

......................................................................................... 2-13

Check logs and gather trace information.....................................................................................

2-13

Check for ODBC errors ..................................................................................................................

2-13

Check for deadlocks and timeouts ...............................................................................................

2-14

Application unable to find previously created objects

..................................................................

2-14

Specify object owner .......................................................................................................................

2-15

Check privilege to access tables ....................................................................................................

2-15

Check temporary DSN attribute ...................................................................................................

2-15

Check Overwrite DSN attribute....................................................................................................

2-15

Check path name to data store......................................................................................................

2-15

Troubleshooting OCI and Pro*C/C++ applications

....................................................................... 2-16

Running out of a resource

.................................................................................................................... 2-16

Operating system tools and shared memory ..............................................................................

2-16

Check the amount of memory allocated to the data store ........................................................

2-16

Permanent segment filling up................................................................................................

2-17

Temporary segment filling up ...............................................................................................

2-17

Update query optimizer statistics.................................................................................................

2-18

Check memory used by queries....................................................................................................

2-18

Check available swap space (virtual memory)...........................................................................

2-18

Check transaction log file use of disk space................................................................................

2-18

Check the semaphore limit ............................................................................................................

2-20

Duplicate results from a SELECT statement

.................................................................................... 2-20

Cannot attach PL/SQL shared memory

............................................................................................. 2-20

3 Troubleshooting Installation, Upgrades and Downgrades

Installing 32-bit TimesTen on 64-bit Windows

.................................................................................. 3-1

Downgrading a data store with Oracle data types to TimesTen 6.0

............................................... 3-1

4 Troubleshooting Oracle In-Memory Database Cache

Unable to create a cache group

.............................................................................................................. 4-1

Unable to start or stop the cache agent

................................................................................................. 4-2

Check status of the cache agent........................................................................................................

4-3

Check ORACLE_HOME environment variable ............................................................................

4-3

Check NLS environment variables..................................................................................................

4-3

Recovering cache grid after unexpected system shutdown

............................................................. 4-3

A portion of the cache grid nodes are still running ......................................................................

4-3

All cache grid nodes exited unexpectedly......................................................................................

4-3

Unable to resolve Oracle Service Name

............................................................................................... 4-4

Unable to resolve connect identifier

.....................................................................................................

4-4

Incompatible Oracle Server and Client versions

............................................................................... 4-5

Unable to validate Oracle username and password

.......................................................................... 4-5

Check library path environment variable ......................................................................................

4-6

Check status of TNS listener and Oracle Server ............................................................................

4-6

Check Oracle privileges.....................................................................................................................

4-6

Check DSN definition........................................................................................................................

4-6

Reboot TimesTen machine................................................................................................................

4-7

Set the cache administration user id and password .....................................................................

4-7

Check user and system environment ..............................................................................................

4-7

Verify the loaded dynamic libraries................................................................................................

4-7

OCI initialization failed

.......................................................................................................................... 4-8

Unsupported data type mapping

.......................................................................................................... 4-9

Null constraint does not match Oracle

................................................................................................. 4-9

DDL operations on cached Oracle tables may cause cache group operations to fail

.................. 4-9

Changes not visible after updating object in cache group

........................................................... 4-10

Loading or refreshing fails

.................................................................................................................. 4-10

Monitoring autorefresh cache groups

............................................................................................... 4-10

Using the ttCacheAutorefreshStatsGet procedure.....................................................................

4-11 v

Displaying information from the change log tables...................................................................

4-13

Understanding messages about autorefresh in the support log ..............................................

4-13

Diagnosing autorefresh failure .....................................................................................................

4-14

Diagnosing autorefresh performance problems.........................................................................

4-15

Using SNMP traps for alerts about autorefresh problems........................................................

4-15

Optimize Performance for IMDB Cache

.......................................................................................... 4-15

Autorefresh not refreshing cache at the specified interval

........................................................... 4-16

Reset autorefresh state....................................................................................................................

4-17

Recover and reset autorefresh Oracle objects .............................................................................

4-17

Incremental autorefresh not progressing

......................................................................................... 4-18

Validate autorefresh Oracle objects..............................................................................................

4-18

Incremental autorefresh becomes full autorefresh

......................................................................... 4-19

Detecting when incremental autorefresh becomes full .............................................................

4-19

Understanding the cache administration user tablespace ........................................................

4-19

Diagnosing a full cache administration user tablespace ...........................................................

4-20

Monitoring the usage of the cache administration user's tablespace ......................................

4-21

Considerations when the cache administration user's tablespace is full................................

4-21

Poor autorefresh performance

............................................................................................................ 4-22

Unresponsive or dead TimesTen database degrades autorefresh performance ...................

4-23

Setting cached TimesTen database timeout ........................................................................

4-23

Configuring recovery method for certain cache groups ....................................................

4-23

Excessive deadlocks, buffer busy and row lock waits during autorefresh cache group refresh....

............................................................................................................................................... 4-26

Abnormally large log and base tables degrade autorefresh performance..............................

4-27

Performance degrades when autorefresh interval is small.......................................................

4-28

5 Troubleshooting AWT Cache Groups

Unable to start or stop replication agent

.............................................................................................. 5-1

Replication does not work

...................................................................................................................... 5-2

Using SNMP traps for notification of replication events

................................................................. 5-2

Monitoring AWT performance

.............................................................................................................. 5-2

Possible causes of poor AWT performance

......................................................................................... 5-3

Permanent Oracle errors reported by TimesTen

................................................................................ 5-3

Transient Oracle errors reported by TimesTen

.................................................................................. 5-4

6 Troubleshooting Replication

Unable to create a replication scheme

.................................................................................................. 6-1

Unable to alter a replication scheme

.................................................................................................... 6-2

Unable to start or stop replication agent

.............................................................................................. 6-2

Using SNMP traps for notification of replication events

................................................................. 6-3

Replication does not work

...................................................................................................................... 6-3

Check status of TimesTen daemon and replication agents..........................................................

6-3

Check that replication agents are communicating ........................................................................

6-5

Check replication state ......................................................................................................................

6-5

Check replication scheme configuration ........................................................................................

6-6

Check ttRepAdmin -showconfig ..............................................................................................

6-6

Check the TTREP.TTSTORES table ..........................................................................................

6-7 vi

Check host names .......................................................................................................................

6-7

Check owner names...........................................................................................................................

6-8

Checking replication owner ......................................................................................................

6-8

Checking table owner.................................................................................................................

6-8

Check consistency between replicated tables .............................................................................

6-10

Replication unresponsive, appears hung

......................................................................................... 6-10

Check replication state ...................................................................................................................

6-10

Check return receipt timeout setting............................................................................................

6-11

Poor replication or XLA performance

................................................................................................ 6-11

Check network bandwidth ............................................................................................................

6-11

Check use of return receipt blocking............................................................................................

6-11

Check replication configuration....................................................................................................

6-12

Check size of log buffer..................................................................................................................

6-12

Check durability settings ...............................................................................................................

6-12

Check for reads from transaction log files...................................................................................

6-12

Problems using ttRepAdmin

.............................................................................................................. 6-14

Problems using ttRepAdmin -duplicate ......................................................................................

6-15

Returns 'Must specify -scheme' error ...........................................................................................

6-15

Problems with conflict checking

........................................................................................................ 6-15

Column cannot be used for replication timestamp....................................................................

6-16

Timestamp does not exist ..............................................................................................................

6-16

Conflict reporting slows down replication..................................................................................

6-16

Index

vii

viii

Preface

Oracle TimesTen In-Memory Database is a memory-optimized relational database.

Deployed in the application tier, Oracle TimesTen In-Memory Database operates on databases that fit entirely in physical memory using standard SQL interfaces. High availability for the in-memory database is provided through real-time transactional replication.

Audience

This guide describes how to troubleshoot some of the problems users encounter when using the Oracle TimesTen In-Memory Database.

To work with this guide, you should understand how database systems work and have some knowledge of SQL (Structured Query Language).

Related documents

TimesTen documentation is available on the product distribution media and on the

Oracle Technology Network: http://www.oracle.com/technology/documentation/timesten_doc.html

Conventions

TimesTen supports multiple platforms. Unless otherwise indicated, the information in this guide applies to all supported platforms. The term Windows refers to Windows

2000, Windows XP and Windows Server 2003. The term UNIX refers to Solaris, Linux,

HP-UX and AIX.

This document uses the following text conventions:

Convention boldface

italic

monospace

Meaning

Boldface type indicates graphical user interface elements associated with an action, or terms defined in text or the glossary.

Italic type indicates book titles, emphasis, or placeholder variables for which you supply particular values.

Monospace type indicates commands within a paragraph, URLs, code in examples, text that appears on the screen, or text that you enter.

ix

x

%

#

[ ]

{ }

Convention

italic monospace

Meaning

Italic monospace type indicates a variable in a code example that you must replace. For example:

|

. . .

Driver=

install_dir

/lib/libtten.sl

Replace

install_dir

with the path of your TimesTen installation directory.

Square brackets indicate that an item in a command line is optional.

Curly braces indicated that you must choose one of the items separated by a vertical bar ( | ) in a command line.

A vertical bar (or pipe) separates alternative arguments.

An ellipsis (. . .) after an argument indicates that you may use more than one argument on a single command line.

The percent sign indicates the UNIX shell prompt.

The number (or pound) sign indicates the UNIX root prompt.

TimesTen documentation uses these variables to identify path, file and user names:

Convention

install_dir

TTinstance bits

or

bb release

or

rr jdk_version timesten

DSN

Meaning

The path that represents the directory where the current release of

TimesTen is installed.

The instance name for your specific installation of TimesTen. Each installation of TimesTen must be identified at install time with a unique alphanumeric instance name. This name appears in the install path.

Two digits, either 32 or 64, that represent either the 32-bit or 64-bit operating system.

Three numbers that represent the first three numbers of the TimesTen release number, with or without a dot. For example, 1121 or 11.2.1 represents TimesTen Release 11.2.1.

Two digits that represent the version number of the major JDK release.

Specifically, 14 represent JDK 1.4; 5 represents JDK 5.

A sample name for the TimesTen instance administrator. You can use any legal user name as the TimesTen administrator. On Windows, the

TimesTen instance administrator must be a member of the

Administrators group. Each TimesTen instance can have a unique instance administrator name.

The data source name.

Documentation Accessibility

Our goal is to make Oracle products, services, and supporting documentation accessible to all users, including users that are disabled. To that end, our documentation includes features that make information available to users of assistive technology. This documentation is available in HTML format, and contains markup to facilitate access by the disabled community. Accessibility standards will continue to evolve over time, and Oracle is actively engaged with other market-leading technology vendors to address technical obstacles so that our documentation can be accessible to all of our customers. For more information, visit the Oracle Accessibility

Program Web site at http://www.oracle.com/accessibility/

.

Accessibility of Code Examples in Documentation

Screen readers may not always correctly read the code examples in this document. The conventions for writing code require that closing braces should appear on an otherwise empty line; however, some screen readers may not always read a line of text that consists solely of a bracket or brace.

Accessibility of Links to External Web Sites in Documentation

This documentation may contain links to Web sites of other companies or organizations that Oracle does not own or control. Oracle neither evaluates nor makes any representations regarding the accessibility of these Web sites.

Access to Oracle Support

Oracle customers have access to electronic support through My Oracle Support. For information, visit http://www.oracle.com/support/contact.html

or visit http://www.oracle.com/accessibility/support.html

if you are hearing impaired.

Technical support

For information about obtaining technical support for TimesTen products, go to the following Web address: http://www.oracle.com/support/contact.html

xi

xii

What’s New

This section summarizes the new features and functionality of Oracle TimesTen

In-Memory Database Release 11.2.1 that are documented in this guide, providing links into the guide for more information.

New features for release 11.2.1.6

This guide now has more information about improving performance for the IMDB cache, which are described in the following sections:

Optimize Performance for IMDB Cache

Excessive deadlocks, buffer busy and row lock waits during autorefresh cache group refresh

Abnormally large log and base tables degrade autorefresh performance

Performance degrades when autorefresh interval is small

New features for release 11.2.1

This guide has information about the following new features:

Issues with the NLS_LANG environment variable

Cannot attach PL/SQL memory

Changes not visible after updating object in cache group

Monitoring the usage of the cache administration user's tablespace

Considerations when the cache administration user's tablespace is full

Poor autorefresh performance

Poor replication or XLA performance

Issues with the NLS_LANG environment variable

On Windows, if the NLS_LANG environment variable is set to an unsupported value,

such as NA, you could experience problems connecting. See "Troubleshooting OCI and Pro*C/C++ applications" on page 2-16.

Cannot attach PL/SQL memory

See

"Cannot attach PL/SQL shared memory" on page 2-20 on how to recover if you

receive error 8517 "Cannot attach PL/SQL shared memory;

PLSQL_MEMORY_ADDRESS not valid or already in use." xiii

xiv

Changes not visible after updating object in cache group

If you modify an object in a cache group and then the changes do not appear on a subsequent SQL statement, then see

"Changes not visible after updating object in cache group" on page 4-10.

Monitoring the usage of the cache administration user's tablespace

To monitor the cache administration user tablespace, you can use either Oracle

Enterprise Manager alerts or set the TimesTen tablespace threshold parameter. See

"Monitoring the usage of the cache administration user's tablespace" on page 4-21 for

details.

Considerations when the cache administration user's tablespace is full

With Oracle tables that are cached in a TimesTen database, you can configure them to

use incremental automatic refresh. See "Considerations when the cache administration user's tablespace is full" on page 4-21 on how to specify what is to occur when the

cache administration user's tablespace is full.

Poor autorefresh performance

There is a new method for improving autorefresh performance:

"Unresponsive or dead TimesTen database degrades autorefresh performance" on page 4-23.

Poor replication or XLA performance

Additional methods for improving replication or XLA performance were added to

"Poor replication or XLA performance" on page 6-11.

1

1

Tools for Troubleshooting TimesTen

This chapter describes how to use the TimesTen utilities and other tools that are used to diagnose problems with the TimesTen data store. This chapter includes the following topics:

Using the ttIsql utility

Using the ttStatus utility

Using the ttCapture utility

Using the logs generated by the TimesTen daemon

Using the ttTraceMon utility

Using the ttXactAdmin utility

Using ODBC tracing

Using SNMP traps to detect events

Monitoring the TimesTen system tables

Using the query optimizer

Using the ttIsql utility

The ttIsql

utility allows you to interactively execute SQL statements and report status information on your data store.

All TimesTen SQL operations can be executed from a ttIsql Command>

prompt.

Example 1–1 Using the ttIsql utility

To start the ttIsql

utility for the demo data store, enter:

% ttIsql demo

You should see output similar to the following:

Copyright (c) 1996-2007, Oracle. All rights reserved.

Type ? or "help" for help, type "exit" to quit ttIsql.

connect "DSN=demo";

Connection successful: DSN=demo;UID=ttuser;DataStore=c:\temp\demo;

DatabaseCharacterSet=US7ASCII;ConnectionCharacterSet=US7ASCII;

DRIVER=C:\WINDOWS\system32\ttdv70.dll;Authenticate=0;PermSize=20;TypeMode=0;

(Default setting AutoCommit=1)

Command>

Tools for Troubleshooting TimesTen

1-1

Using the ttStatus utility

You can then execute SQL statements or ttIsql

commands at the

Command>

prompt.

"Using the ttIsql Utility" in the

Oracle TimesTen In-Memory Database Operations Guide

describes how to use the most common ttIsql

commands. The following ttIsql commands are commonly used when troubleshooting:

■ monitor

formats the contents of the SYS.MONITOR table.

See "Displaying data store information" in the

Oracle TimesTen In-Memory Database

Operations Guide

.

dssize

prints data store size information.

See "Displaying data store information" in the

Oracle TimesTen In-Memory Database

Operations Guide

.

showplan

prints the optimizer execution plans for selects/updates/deletes in this transaction.

See "Viewing and changing query optimizer plans" in the

Oracle TimesTen

In-Memory Database Operations Guide

.

isolation

sets or displays the isolation level.

See "Working with transactions" in the

Oracle TimesTen In-Memory Database

Operations Guide

.

timing

prints query timing.

See "Timing ODBC function calls" in the

Oracle TimesTen In-Memory Database

Operations Guide

.

optprofile

prints the current optimizer flag settings and join order.

See "Viewing and changing query optimizer plans" in the

Oracle TimesTen

In-Memory Database Operations Guide

.

For the full list of ttIsql

features, see the lists of options and commands under the description of the ttIsql

utility in the

Oracle TimesTen In-Memory Database Reference

.

Using the ttStatus utility

Use the ttStatus

utility to check the status of the TimesTen daemon and the state of all TimesTen -connections.

Example 1–2 ttStatus shows TimesTen daemon is not running

In this example, the output from ttStatus

indicates that no TimesTen daemon is running. If the daemon has stopped unexpectedly, see

"No response from TimesTen daemon or subdaemon" on page 2-2 for troubleshooting information.

On Windows:

C:\>ttStatus ttStatus: Could not connect to the TimesTen service.

If the TimesTen service is not running, please start it by running "ttDaemonAdmin

-start".

On UNIX platforms:

$ ttStatus ttStatus: Could not connect to the TimesTen daemon.

If the TimesTen daemon is not running, please start it by running "ttDaemonAdmin -start".

1-2

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Using the ttStatus utility

Example 1–3 ttStatus shows TimesTen daemon is running

In this example, the output from ttStatus

indicates that the TimesTen daemon is running. It recognizes one data store named demo

.

The first line indicates that the TimesTen daemon is running as process 884 on port

17000 for the TimesTen instance

MYINSTANCE

. The second line indicates the TimesTen

Server is running as process 2308 on port 17002.

There are currently seven connections to the data store: one user and six subdaemon connections. You may see up to 2047 connections.

The restart policies for the cache agent and the replication agent in the data store are set to manual

.

Note:

This example was produced on Windows. The results are the same on UNIX platforms except for the formats of the data store path and the shared memory key.

C:\>ttStatus

TimesTen status report as of Thu Jan 25 15:45:11 2007

Daemon pid 884 port 17000 instance MYINSTANCE

TimesTen server pid 2308 started on port 17002

TimesTen webserver pid 2188 started on port 17004

------------------------------------------------------------------------

Data store c:\temp\demo

There are 7 connections to the data store

Data store is in shared mode

Shared Memory KEY Global\DBI45b94095.1.SHM.4 HANDLE 0x278

Type PID Context Connection Name ConnID

Process 4616 0x00d08820 demo 1

Subdaemon 2136 0x00526768 Worker 2042

Subdaemon 2136 0x0072e750 Flusher 2043

Subdaemon 2136 0x007348b8 Checkpoint 2044

Subdaemon 2136 0x067b0068 Aging 2045

Subdaemon 2136 0x067c0040 Monitor 2047

Subdaemon 2136 0x068404c8 HistGC 2046

Replication policy : Manual

Cache agent policy : Manual

------------------------------------------------------------------------

End of report

Example 1–4 ttStatus shows replication information

In this example, the output from ttStatus

indicates that the TimesTen daemon is running. It recognizes three data stores: demo

, subscriber1ds

, and masterds

. The subscriber1ds

and masterds

data stores are replicated data stores. In this example, the output from ttStatus

indicates that the replication agents for the replicated data stores have been started. Bidirectional replication has been configured between masterds

and subscriber1ds

. Each replication agent has five connections to the data store.

C:\>ttStatus

TimesTen status report as of Thu Jan 25 16:23:33 2007

Daemon pid 5088 port 17000 instance MYINSTANCE

TimesTen server pid 4344 started on port 17002

Tools for Troubleshooting TimesTen

1-3

Using the ttStatus utility

TimesTen webserver pid 4216 started on port 17004

------------------------------------------------------------------------

Data store c:\temp\subscriber1ds

There are 12 connections to the data store

Data store is in shared mode

Shared Memory KEY Global\DBI45b9471c.2.SHM.2 HANDLE 0x280

Type PID Context Connection Name ConnID

Process 1244 0x00d08fb0 subscriber1ds 1

Replication 4548 0x00aed2f8 LOGFORCE 4

Replication 4548 0x00b03470 TRANSMITTER 5

Replication 4548 0x00b725a8 RECEIVER 6

Replication 4548 0x00b82808 REPHOLD 2

Replication 4548 0x00b98980 REPLISTENER 3

Subdaemon 2752 0x00526768 Worker 2042

Subdaemon 2752 0x0072a758 Flusher 2043

Subdaemon 2752 0x007308c0 Checkpoint 2044

Subdaemon 2752 0x00736a28 HistGC 2046

Subdaemon 2752 0x067f02f8 Aging 2045

Subdaemon 2752 0x068364a0 Monitor 2047

Replication policy : Manual

Replication agent is running.

Cache agent policy : Manual

------------------------------------------------------------------------

Data store c:\temp\masterds

There are 12 connections to the data store

Data store is in shared mode

Shared Memory KEY Global\DBI45b945d0.0.SHM.6 HANDLE 0x2bc

Type PID Context Connection Name ConnID

Process 5880 0x00d09008 masterds 1

Replication 3728 0x00aed570 LOGFORCE 4

Replication 3728 0x00b036e8 TRANSMITTER 5

Replication 3728 0x00b168b8 REPHOLD 3

Replication 3728 0x00b1ca20 REPLISTENER 2

Replication 3728 0x00b22b88 RECEIVER 6

Subdaemon 3220 0x00526768 Worker 2042

Subdaemon 3220 0x0072e768 Flusher 2043

Subdaemon 3220 0x007348d0 Checkpoint 2044

Subdaemon 3220 0x067b0068 Aging 2045

Subdaemon 3220 0x067c0040 Monitor 2047

Subdaemon 3220 0x068404c8 HistGC 2046

Replication policy : Manual

Replication agent is running.

Cache agent policy : Manual

------------------------------------------------------------------------

Data store c:\temp\demo

There are no connections to the data store

Replication policy : Manual

Cache agent policy : Manual

------------------------------------------------------------------------

End of report

Example 1–5 ttStatus shows cache group information

This example shows the cache agent running on rep1

data store. There is one cache group in the data store. The cache agent has five connections to the data store.

C:\>ttStatus

TimesTen status report as of Mon Mar 19 10:47:46 2007

Daemon pid 1012 port 17000 instance MYINSTANCE

No TimesTen server running

1-4

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Using the ttCapture utility

TimesTen webserver pid 1708 started on port 17004

----------------------------------------------------------------

Data store c:\data\rep1

There are 12 connections to the data store

Data store is in shared mode

Shared Memory KEY Global\DBI45ef98ac.1.SHM.56 HANDLE 0x260

Type PID Context Connection Name ConnID

Cache Agent 3380 0x00bbddf0 Handler 2

Cache Agent 3380 0x00c3f318 Aging 3

Cache Agent 3380 0x07380398 Timer 4

Cache Agent 3380 0x073cfa18 ttora70 6

Cache Agent 3380 0x073ff010 ttora70 7

Process 2084 0x00c48ee8 rep1 1

Subdaemon 1632 0x006bc430 Worker 2042

Subdaemon 1632 0x06630458 Flusher 2045

Subdaemon 1632 0x0664f978 Checkpoint 2044

Subdaemon 1632 0x0665ee60 HistGC 2043

Subdaemon 1632 0x066de720 Aging 2046

Subdaemon 1632 0x0670dc78 Monitor 2047

Replication policy : Manual

Cache agent policy : Manual

TimesTen's Cache agent is running for this data store

-----------------------------------------------------------------

End of report

Example 1–6 ttStatus shows connection to old instance

This example shows a connection to an old instance of a data store. This can occur when a data store is invalidated, but some users have not disconnected from the invalidated copy of the data store still in memory. After all users disconnect, the memory can be freed.

C:\>ttStatus

TimesTen status report as of Thu Jan 25 16:44:49 2007

Daemon pid 5088 port 17000 instance MYINSTANCE

TimesTen server pid 4344 started on port 17002

TimesTen webserver pid 4216 started on port 17004

-----------------------------------------------------------------

Data store c:\temp\sample

There are no connections to the data store

Obsolete or not yet active connection(s):

Process 4696 context 0xd08800 name sample connid 1, obsolete connection, shmKey

'Global\DBI45b94c6f.3.SHM.4'

Replication policy : Manual

Cache agent policy : Manual

-----------------------------------------------------------------

End of report

Using the ttCapture utility

The ttCapture

utility captures information about the configuration and state of your

TimesTen system into a file that provides

Technical support with a snapshot of your

system at the time the ttCapture

utility is running. The ttCapture

utility generates a file named ttcapture.

date.time.

log

. By default, the file is written to the directory from which you invoke the ttCapture

utility. Use the ttCapture -dest option to direct the output file to be written to another directory.

If you run ttCapture

again, it writes the information to a new file.

Tools for Troubleshooting TimesTen

1-5

Using the logs generated by the TimesTen daemon

On Windows platforms, running ttCapture

also produces an XML file named ttcapture.

date.time

.nfo

that contains output from the msinfo32

utility.

When you experience a problem with a TimesTen data store, run ttCapture

with the

DSN

option for the data store as soon as possible, either when you are encountering the problem or immediately afterward.

Note:

Always double-quote directory and file names in case there are spaces in the names.

When you contact

Technical support , upload the

ttcapture.

date.number.

log

file to the Service Request. Windows users should also upload the ttcapture.

date.time

.nfo

file.

See "ttCapture" in the

Oracle TimesTen In-Memory Database Reference

for information about additional options.

Using the logs generated by the TimesTen daemon

TimesTen uses a TimesTen daemon to manage access to the data stores. As the daemon operates, it generates error, warning and informational messages. These messages may be useful for TimesTen system administration and for debugging applications.

By default, informational messages are stored in:

A user error log that contains information you may need to see. Generally, these messages contain information about actions you may need to take.

A support log containing everything in the user error log plus information for use

by Technical support .

See "Modifying informational messages" in the

Oracle TimesTen In-Memory Database

Operations Guide

for information about configuring the logs, including their location and size.

Using the ttTraceMon utility

Use the ttTraceMon

utility to log various trace information on a number of TimesTen components. Each TimesTen component can be traced at different levels of detail. You can list all of the traceable TimesTen components and their current tracing level by specifying ttTraceMon

with the show

subcommand. The full list of options for ttTraceMon

is described in the "ttTraceMon" section in the

Oracle TimesTen

In-Memory Database Reference

.

TimesTen tracing severely impacts application performance and consumes a great deal of disk space if trace output is directed to a file. In addition, when using AWT cache groups, you must restart the replication agent when trying to trace the ORACON component with ttTraceMon

. Use the ttTraceMon

utility only when diagnosing problems. When you are finished, reset tracing to the default values.

Example 1–7 Default trace levels for components

This example shows that the levels for most tracing components are set to level 0 (off) for the demo data store. Both the ERR and DEADLOCK components are set to 1 for

tracing by default. See "ERR tracing" on page 1-13.

% ttTraceMon -e show demo

1-6

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Using the ttTraceMon utility

AGING ... 0

API ... 0

ASYNCMV ... 0

AUTOREFRESH ... 0

CG ... 0

CGRID ... 0

CGRIDC ... 0

CKPT ... 0

DEADLOCK ... 1

EE ... 0

ERR ... 1

FLOW ... 0

HEAP ... 0

LATCH ... 0

LOCK ... 0

LOG ... 0

LOGF ... 0

OPT ... 0

ORACON ... 0

PLOAD ... 0

PT ... 0

REPL ... 0

SM ... 0

SQL ... 0

TEST ... 0

TRACE ... 0

XA ... 0

XACT ... 0

The output for most TimesTen components is of interest only to

Technical support

.

However, the output for the SQL, API, LOCK, ERR, AGING and AUTOREFRESH components may be useful to you when you are troubleshooting application problems.

The rest of this section includes the following topics:

Starting a trace and reading the trace buffer

SQL tracing

API tracing

DEADLOCK tracing

LOCK tracing

ERR tracing

AGING tracing

AUTOREFRESH tracing

Starting a trace and reading the trace buffer

Start a new trace by specifying ttTraceMon

datastore

. For example, to start a trace on the demo data store, enter:

% ttTraceMon demo

Trace monitor; empty line to exit

Trace >

At the Trace prompt, specify the type of trace and its level. For example, to start tracing the SQL component at level 3, enter:

Tools for Troubleshooting TimesTen

1-7

Using the ttTraceMon utility

Trace > level sql 3

At this point you can run your application and the TimesTen trace information is written to a trace buffer. There are two ways to read the contents of the trace buffer:

From the Trace prompt, use the outfile

command to direct the trace buffer data to a file. (You must do this before running your application.) When writing tracing information to a file, new trace information is concatenated to the existing contents of the file.

From the Trace prompt, use the dump

command to display the trace buffer data to your screen.

Note:

The contents of the trace buffer accumulate with each new trace. To clear the trace buffer, use the flush

command from a ttTraceMon

prompt. Clear the buffered trace records for a specific component by specifying the component with the flush

command.

Each record from the trace buffer has the following format:

timestamp sequence component level connection processid operation

The fields in the records are defined as follows:

timestamp

is the time at which the operation was executed.

sequence

is the incremental number that identifies the trace line.

component

is the TimesTen component being traced (such as SQL, API, LOCK, or

ERR).

level

is the trace level associated with the trace line. The range of trace levels differs by component, but for all components, the lowest trace level generates the least verbose output and highest trace level generates the most verbose output.

For example, if you are tracing SQL at level 4, your output includes lines for levels

2 (prepare), 3 (execute), and 4 (open, close, fetch).

Note:

Trace levels for some components are not a continuous range of numbers. If you enter a number that does not correspond to a supported level for a component, tracing occurs at the highest supported level that is less than the number you entered. For example, if tracing levels for a component are 1, 2, 3, 4, and 6 and you enter 5, tracing events for level 1, 2, 3, and 4 are generated.

connection

is the internal connection ID identifying the connection that generated the trace. This number corresponds to the ConnID shown in ttStatus output. The connection ID is also used as the first element of the transaction ID.

processid

is the operating system process ID for the process that generated the trace.

operation

is the operation that occurred (such as SQL statement, API operation, or error message).

For example, a line from the trace buffer after a SQL trace at level 3 might look like this:

10:39:50.231 5281 SQL 2L 1C 3914P Preparing: select cust_num from customer

1-8

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Using the ttTraceMon utility

SQL tracing

Using ttTraceMon

with the SQL component provides information about the SQL being prepared or executed by the TimesTen engine.

Table 1–1 describes the levels for

SQL tracing. Each level with a ’+’ sign includes the trace information described for that level, plus all levels preceding it.

Table 1–1

Level

2

3

4

5

SQL tracing levels

Output

SQL commands being prepared.

+ SQL commands being executed

+ The effect of command pooling (prepares not being done because the prepared command already exists in the pool), the need for reprepares (for example, because an index was created), and commands being destroyed.

At this level, ttTraceMon

also shows when a query command is being opened, fetched, and closed.

+ Some internal data, such as command numbers, which are not generally useful for customer-level debugging.

Note:

TimesTen recommends tracing SQL at level 3 or 4. SQL tracing does not show any information about the optimizer. Optimizer tracing is managed by a separate component (OPT) at level 4 only, and is not designed for customer use.

Example 1–8 SQL trace

In this example, we execute ttTraceMon

to do a SQL trace at level 4 on the demo

data store. We direct the output from the SQL trace to the

SQLtrace.txt

file. We then flush the buffer so that the trace does not report past SQL statements.

% ttTraceMon demo

Trace monitor; empty line to exit

Trace > outfile SQLtrace.txt

Trace > level sql 4

Trace > flush

At this point, we execute an application that includes the following SQL statement:

SELECT * FROM departments WHERE department_id = 10;

The trace information is written to the

SQLtrace.txt

file:

12:19:36.582 269 SQL 2L 3C 29570P Preparing: select * from departments where department_id = 10

12:19:36.583 270 SQL 4L 3C 29570P sbSqlCmdCompile ()(E): (Found already compiled version: refCount:01, bucket:28) cmdType:100, cmdNum:1000146.

12:19:36.583 271 SQL 4L 3C 29570P Opening: select * from departments where department_id = 10;

12:19:36.583 272 SQL 4L 3C 29570P Fetching: select * from departments where department_id = 10;

12:19:36.583 273 SQL 4L 3C 29570P Closing: select * from departments where department_id = 10;

5 records dumped

When the application has completed, we turn off SQL tracing and exit ttTraceMon

.

Trace > level sql 0

Tools for Troubleshooting TimesTen

1-9

Using the ttTraceMon utility

Trace > {press ENTER – blank line}

API tracing

API traces are generated for database operations such as connecting to a data store, changing a connection attribute, and committing a transaction.

Table 1–2 describes the

levels for API tracing. Each level with a ’+’ sign includes the trace information described for that level, plus all levels preceding it.

2

3

Table 1–2

Level

1

4

API tracing levels

Output

All rollback attempts by the subdaemon. This occurs if an application exits abruptly and the subdaemon recovers its connection.

+ Some low-on-space conditions.

+ Create, connect, disconnect, checkpoint, backup, and compact operations for the data store, as well as commit and rollback for each connection, and a few other operations.

+ Most other operations conducted at TimesTen's internal API level. It does not show numerous operations on the storage manager and indexes that are done internally.

Note:

TimesTen recommends tracing at level 3.

Example 1–9 API trace

In this example, we execute ttTraceMon

to do a API trace at level 3 on the demo

data store. The output from the API trace is written to the

APItrace.txt

file. Before saving the API trace to the buffer, we use the flush

command to empty the buffer.

% ttTraceMon demo

Trace monitor; empty line to exit

Trace> outfile APItrace.txt

Trace> level api 3

Trace > flush

At this point, we execute the application. When the application has completed, we turn off API tracing and exit ttTraceMon

:

Trace > level api 0

Trace > {press ENTER – blank line}

The contents of

APItrace.txt

are similar to the sample output shown below. The output shows connection to the data store, setting the connection character set, setting the isolation level, and committing a transaction.

11:54:26.796 1016 API 3L 2C 4848P sb_dbConnect()(X)

11:54:26.796 1017 API 3L 2C 4848P sb_dbConnCharsetSet()(E)

11:54:26.796 1018 API 3L 2C 4848P sb_dbConnSetIsoLevel()(E)

11:54:39.795 1019 API 3L 2C 4848P sb_dbConnSetIsoLevel()(E)

11:54:45.253 1020 API 3L 2C 4848P sb_xactCommitQ()(E)

DEADLOCK tracing

Use the DEADLOCK component to trace the occurances of all deadlocks for all applications.

1-10

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Using the ttTraceMon utility

Table 1–3 describes the DEADLOCK tracing levels. Each level with a ’+’ sign includes

the trace information described for that level, plus all levels preceding it.

Table 1–3

Level

1

4, 6

DEADLOCK tracing levels

Output

Logs deadlock cycles as they are discovered.

+ Provides detail information about how the deadlock is detected.

Example 1–10 DEADLOCK trace

In this example, we execute ttTraceMon

to do a DEADLOCK trace at level 1, which is the default, on myDSN

data store.

We make two connections to myDSN

. For the first connection, autocommit is on. We create table test

and insert two rows. Then, we set autocommit off and update the x1=1

row of table test. Because autocommit is off, the row is not inserted into the table until we commit. A lock is held until we commit or roll back the transaction.

Command> create table test (x1 int unique, y1 int);

Command> insert into test values (1,1);

1 row inserted.

Command> insert into test values (2,2);

1 row inserted.

Command> autocommit 0;

Command> update test set y1=y1 where x1=1;

1 row updated.

For the second connection to myDSN

, autocommit is set to off. We update the x1=2 row of table test.

Command> autocommit 0;

Command> update test set y1=y1 where x1=2;

1 row updated.

Now, we create a deadlock situation by executing update statements in both connections for rows that are locked by each other. The first connection executes an update against the row where x1=2

.

Command> update test set y1=y1 where x1=2;

6003: Lock request denied because of time-out

Details: Tran 2.1 (pid 32750) wants Un lock on rowid BMUFVUAAAAaAAAAETk, table ME.TEST. But tran 3.2 (pid 32731) has it in Xn (request was Xn).

Holder SQL (update t1 set y1=y1 where x1=2)

The command failed.

The second connection executes an update against the row where x1=1

.

Command> update test set y1=y1 where x1=1;

6002: Lock request denied because of deadlock

Details: Tran 3.2 (pid 32731) wants Un lock on rowid BMUFVUAAAAaAAAADzk, table ME.TEST. But tran 2.1 (pid 32750) has it in Xn (request was Xn).

Holder SQL (update t1 set y1=y1 where x1=1)

The command failed.

We use the flush

command to empty the buffer.

% ttTraceMon myDSN

Trace monitor; empty line to exit

Trace> flush

Tools for Troubleshooting TimesTen

1-11

Using the ttTraceMon utility

The trace buffer contains the following information showing all level 1 deadlock traces, as evidenced by ’1L’.:

Trace> dump

09:50:26.444 13 DEADLOCK 1L 2036C 3484P edge 1: xid 3.2, cid 3,

<Row BMUFVUAAAAaAAAADzk,0x8c5

74(574836)> 0 cnt=1 , Tbl 'T1', SQL='update t1 set y1=y1 where x1=1'

09:50:26.455 14 DEADLOCK 1L 2036C 3484P edge 0: xid 2.1, cid 2,

<Row BMUFVUAAAAaAAAAETk,0x8c5

74(574836)> 0 cnt=1 , Tbl 'T1', SQL='update t1 set y1=y1 where x1=2'

09:50:26.455 15 DEADLOCK 1L 2036C 3484P Victim: xcb:3.2,

SQL='update t1 set y1=y1 where x1=1'

If you want more informatin, set DEADLOCK tracing to a higher value. For example, the following sets DEADLOCK tracing to level 4 in ttTraceMon

:

Trace > level deadlock 4

LOCK tracing

Use the LOCK component to trace the locking behavior of your application to detect trouble with deadlocks or lock waits. LOCK tracing generates a great deal of volume, so it is important to choose the appropriate level of tracing. Level 3, for example, begins to generate a large number of traces, as traces are written for fairly common events. In addition, the traces themselves may be somewhat hard to read in places. If you cannot discern enough information for your purposes, contact

Technical support

for more information.

Table 1–4 describes the LOCK tracing levels. Each level with a ’+’ sign includes the

trace information described for that level, plus all levels preceding it.

Table 1–4

Level

3

4

1

2

6

LOCK tracing levels

Output

Deadlock cycles as they are discovered.

+ Failures to grant locks for any reason.

+ Lock waits (which may or may not be granted).

+ All lock grants/releases, some routine calls, and the mechanism of the deadlock detector.

+ Each step in cycle traversal.

Example 1–11 LOCK trace

In this example, we execute ttTraceMon

to do a LOCK trace at level 4 on myDSN

data store.

We make two connections to myDSN

. For the first connection, we set autocommit on.

We create table test

and insert three rows. We create a materialized view using table test

.

We turn on tracing at level 4 and use the flush

command to empty the buffer.

% ttTraceMon myDSN

Trace monitor; empty line to exit

Trace> level lock 4

Trace> flush

1-12

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Using the ttTraceMon utility

For the second connection to myDSN

, we set autocommit off. We insert a row into table test. Because autocommit is off, the row is not inserted into the table until we commit.

A lock is held until we commit or roll back the transaction.

If we use the dump

command to display the contents of the trace buffer, we see that there are no records in the trace buffer:

Trace> dump

0 records dumped

From the first connection, we try to drop the materialized view. We cannot drop the view because there is a transaction that has not been committed or rolled back:

Command> drop materialized view v;

6003: Lock request denied because of time-out

Details: Tran 3.71 (pid 24524) wants Sn lock on table TTUSER.TEST. But tran 1.42

(pid 24263) has it in IXn (request was IXn). Holder SQL (insert into test values

(100);)

The command failed.

The trace buffer contains the following information:

Trace> dump

20:09:04.789 174759 LOCK 3L 3C 24524P ENQ: xcb:00003 <Tbl 0x9b894,0x0>

0+Sn=>SL activity 0 Sn cnt=0; Holder xcb:00001 IXn

20:09:04.789 174760 LOCK 3L 3C 24524P Connection 3 scheduled for sleep

20:09:04.789 174761 LOCK 3L 3C 24524P Connection 3 sleeping

20:09:14.871 174762 LOCK 3L 2047C 24237P Connection 3 timed out

20:09:14.871 174763 LOCK 3L 2047C 24237P Connection 3 woken up

20:09:14.871 174764 LOCK 3L 3C 24524P Connection 3 awake

20:09:14.871 174765 LOCK 2L 3C 24524P ENQ: xcb:00003 <Tbl 0x9b894,0x0>

0+Sn=>TM activity 0 Sn cnt=1; Holder xcb:00001 IXn

7 records dumped

When finished with the trace, we set LOCK tracing back to its default setting (0) and exit ttTraceMon

:

Trace > level lock 0

Trace > {press ENTER – blank line}

ERR tracing

It may be useful to trace the ERR component. For example, an ERR trace at level 4 shows all of the error messages that are pushed in the TimesTen direct driver (not all errors are output to the user because they are handled internally). ERR tracing at level

1 is the default. No output is written for ERR tracing at level 2 and 3.

Table 1–5 describes ERR tracing levels. Each level with a ’+’ sign includes the trace

information described for that level, plus all levels preceding it.

Table 1–5

Level

ERR tracing levels

1 (set by default)

4

Output

Fatal errors

+ All other error messages, many of which are handled internally by

TimesTen.

Example 1–12 ERR trace

In this example, we execute ttTraceMon

to do a ERR trace at level 4 on myDSN

data store.

Tools for Troubleshooting TimesTen

1-13

Using the ttTraceMon utility

First we create a table:

Command> create table test (id tt_integer);

Next we turn on tracing at level 4. Rather than direct the trace output to a file as in the previous examples, we read it directly from the trace buffer. Before saving the ERR trace to the buffer, we use the flush

command to empty the buffer.

% ttTraceMon myDSN

Trace monitor; empty line to exit

Trace> level err 4

Trace> flush

Now we execute a SQL script with three errors in it. The errors are:

Creating a table with the same name as an existing table

Using incorrect syntax to insert values into the table

Inserting CHAR data into a TT_INTEGER column

Command> create table test (id tt_integer);

2207: Table TEST already exists

The command failed.

Command> insert into test values 'abcd');

1001: Syntax error in SQL statement before or at: "'abcd'", character position:

25 insert into test values 'abcd');

^^^^^^

The command failed.

Command> insert into test values ('abcd');

2609: Incompatible types found in expression

The command failed.

The trace information is written to the trace buffer. We display it by using the dump command.

Trace> dump

19:28:40.465 174227 ERR 4L 1C 24263P TT2207: Table TEST already exists

-- file "eeDDL.c", lineno 2930, procedure "sbEeCrDtblEval()"

19:28:51.399 174228 ERR 4L 1C 24263P TT1001: Syntax error in SQL statement before or at: "'abcd'", character position: 25 insert into test values 'abcd');

^^^^^^

-- file "ptSqlY.y", lineno 6273, procedure "reserved_word_or_syntax_error"

19:29:00.725 174229 ERR 4L 1C 24263P TT2609: Incompatible types found in expression -- file "saCanon.c", lineno 12618, procedure "sbPtAdjustType()"

3 records dumped

Set ERR tracing back to its default setting (1) and exit ttTraceMon

:

Trace > level err 1

Trace > {press ENTER – blank line}

AGING tracing

Use the ttTraceMon

utility to obtain the following information:

When aging starts and ends

How many rows have been deleted by the aging subdaemon

See "Implementing aging in your tables" in the

Oracle TimesTen In-Memory Database

Operations Guide

.

1-14

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Using the ttTraceMon utility

Table 1–6 describes the AGING tracing levels. Each level with a ’+’ sign includes the

trace information described for that level, plus all levels preceding it.

3

4

Table 1–6

Level

1

2

AGING tracing levels

Description

Displays messages about the following events:

The aging subdaemon starts least recently used (LRU) or time-based aging.

The aging subdaemon repeats LRU aging because the LRU threshold was not met.

The aging subdaemon ends LRU or time-based aging.

+ Displays messages about the following events

for each table

:

Aging has started.

Aging has ended. The message includes the reason for ending and the total number of rows deleted.

+ Detailed report on how many rows were deleted during each aging cycle.

+ Message every time the aging subdaemon wakes up.

Example 1–13 AGING trace

In this example, we execute ttTraceMon

to do an AGING trace on myDSN

data store.

The data store contains

TTUSER.MYTAB

table, which has a time-based aging policy.

The table is described as follows:

Command> describe TTUSER.MYTAB;

Table TTUSER.MYTAB:

Columns:

*ID TT_INTEGER NOT NULL

TS TIMESTAMP (6) NOT NULL

Aging use TS lifetime 3 minutes cycle 1 minute on

1 table found.

(primary key columns are indicated with *)

The table contains the following rows before the aging cycle begins:

Command> select * from TTUSER.MYTAB;

< 1, 2007-03-21 12:54:06.000000 >

< 3, 2010-03-17 08:00:00.000000 >

< 4, 2007-03-21 12:59:40.000000 >

< 5, 2007-03-21 13:00:10.000000 >

< 6, 2007-03-21 13:01:22.000000 >

5 rows found.

We execute ttTraceMon

to do an AGING trace at level 3. Rather than direct the trace output to a file, we read it directly from the trace buffer. Before saving the AGING trace to the buffer, we use the flush

command to empty the buffer.

% ttTraceMon myDSN

Trace monitor; empty line to exit

Trace> level aging 3

Trace> flush

Display the trace information in the buffer by using the dump

command.

Trace> dump

13:16:56.802 1247 AGING 1L 2045C 17373P Entering sbAgingTB(): curTime=78

Tools for Troubleshooting TimesTen

1-15

Using the ttTraceMon utility

13:16:56.803 1248 AGING 2L 2045C 17373P Entering sbAgingOneTable(): curTime=78, ltblid= 637140

13:16:56.804 1249 AGING 3L 2045C 17373P curTime=78, 4 deleted, 1 remaining, tbl = TTUSER.MYTAB

13:16:56.804 1250 AGING 2L 2045C 17373P Exiting sbAgingOneTable(): curTime=78, reason = 'no more rows', 4 deleted, 1 remaining, tbl = TTUSER.MYTAB

13:16:56.804 1251 AGING 1L 2045C 17373P Exiting sbAgingTB(): curTime=78

5 records dumped

We set AGING tracing back to its default setting (0) and exit ttTraceMon

:

Trace > level aging 0

Trace > {press ENTER – blank line}

AUTOREFRESH tracing

Use the ttTraceMon

utility to obtain information about the progress of autorefresh operations for cache groups with the AUTOREFRESH cache group attribute.

See "AUTOREFRESH cache group attribute" in the

Oracle In-Memory Database Cache

User's Guide

.

Table 1–7 describes AUTOREFRESH tracing levels. Each level with a ’+’ sign includes

the trace information described for that level, plus all levels preceding it.

Table 1–7

Level

1

2

AUTOREFRESH tracing levels

Description

Autorefresh summary for the interval:

The time that autorefresh started

Number of autorefreshed rows for the interval

Number of autorefreshed root table rows for interval

Total number of autorefreshed rows since the cache agent started

Total number of autorefreshed rows in the root table since the cache agent started

The time that autorefresh ended

Note

: Times and information about root table rows are reported for full autorefresh.

+ Autorefresh summary at the cache group level:

The time that autorefresh started for each cache group

Number of autorefreshed rows for each cache group

Number of autorefreshed root table rows for each cache group

Total number of autorefreshed rows for each cache group since the cache agent started

Total number of autorefreshed rows in the root table for each cache group since the cache agent started

The time that autorefresh ended for each cache group

Note

: Times and information about root table rows are reported for full autorefresh.

1-16

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Using the ttTraceMon utility

Table 1–7 (Cont.) AUTOREFRESH tracing levels

Level

3

4

Description

+ Autorefresh summary at the table level:

The time that autorefresh started

Number of autorefreshed rows

Total number of autorefreshed rows since the cache agent started

The time that autorefresh ended

+ Autorefresh details for each table:

The time that autorefresh started for each table

The autorefresh query

Query execute time in milliseconds on the Oracle database

Query fetch time in milliseconds on the Oracle database

Query apply time in milliseconds on TimesTen

Query execute time in milliseconds on the Oracle database for child tables

Query fetch time in milliseconds on the Oracle database for child tables

Query apply time in milliseconds on TimesTen for child tables

The time that autorefresh ended for each table

The autorefresh bookmark ( logseq

) to which autorefresh was completed

Example 1–14 AUTOREFRESH trace

In this example, we use the ttTraceMon

utility to trace autorefresh operations on the cgDSN

data store. When we set the trace level to 1, we see information that is similar to the output of the ttCacheAutorefreshStatsGet

built-in procedure.

% tttracemon cgDSN

Trace monitor; empty line to exit

Trace> level autorefresh 1

Trace> dump

08:56:57.445 19398 AUTOREFRESH 1L 5C 32246P Autorefresh number 1415 started for interval 60000

08:56:57.883 19419 AUTOREFRESH 1L 5C 32246P Duration For Interval 60000ms: 420

08:56:57.883 19420 AUTOREFRESH 1L 5C 32246P Num Rows For Interval 60000ms: 0

08:56:57.883 19421 AUTOREFRESH 1L 5C 32246P Num Root Rows For Interval 60000ms: 0

08:56:57.883 19422 AUTOREFRESH 1L 5C 32246P Cumulative Rows for Interval 60000ms:

11587

08:56:57.883 19423 AUTOREFRESH 1L 5C 32246P Cumulative Root Rows for Interval

60000ms: 1697

08:56:57.883 19424 AUTOREFRESH 1L 5C 32246P Autorefresh number 1415 ended for interval 60000ms successfully.

7 records dumped

Tracing at level 4 produces additional information about autorefresh operation 1415.

Information about autorefresh is provided for the testuser.readcache

cache group, the testuser.readtab

root table and the autorefresh interval.

Trace> level autorefresh 4

Trace> dump

08:56:57.445 19398 AUTOREFRESH 1L 5C 32246P Autorefresh number 1415 started for interval 60000

08:56:57.471 19399 AUTOREFRESH 2L 5C 32246P Autorefresh started for cachegroup

TESTUSER.READCACHE

Tools for Troubleshooting TimesTen

1-17

Using the ttTraceMon utility

08:56:57.471 19400 AUTOREFRESH 3L 5C 32246P Incremental autorefresh started for table TESTUSER.READTAB

08:56:57.471 19401 AUTOREFRESH 4L 5C 32246P Autorefresh Query: SELECT L."COL_10",

X."COL_20", X.ft$NotDelete, Z.rowid FROM (SELECT DISTINCT "COL_10" FROM

"TESTUSER"."TT_03_454854_L" WHERE logseq >:logseq AND ft_cacheGroup <>

100000000000*1844259679+-299200618) L,(SELECT "TESTUSER"."READTAB"."COL_10",

"TESTUSER"."READTAB"."COL_20", 1 AS ft$NotDelete FROM "TESTUSER"."READTAB",

"TESTUSER"."T1" WHERE "TESTUSER"."READTAB"."COL_10" = "TESTUSER"."T1"."COL_10")

X, "TESTUSER"."READTAB" Z WHERE L ."COL_10" = X."COL_10" (+) AND X."COL_10" =

Z."COL_10" (+), logseq: 7

08:56:57.870 19402 AUTOREFRESH 3L 5C 32246P Duration for table TESTUSER.READTAB:

70

08:56:57.870 19403 AUTOREFRESH 3L 5C 32246P Num Rows for table TESTUSER.READTAB: 1

08:56:57.870 19404 AUTOREFRESH 3L 5C 32246P Cumulative rows for table

TESTUSER.READTAB: 1559

08:56:57.870 19405 AUTOREFRESH 4L 5C 32246P Autorefresh Query Execute duration for table TESTUSER.READTAB: 60

08:56:57.870 19406 AUTOREFRESH 4L 5C 32246P Autorefresh Query Fetch duration for table TESTUSER.READTAB: 0

08:56:57.870 19407 AUTOREFRESH 4L 5C 32246P Autorefresh Query Apply duration for table TESTUSER.READTAB: 0

08:56:57.870 19408 AUTOREFRESH 4L 5C 32246P Max logseq applied for table

TESTUSER.READTAB: 8

08:56:57.870 19409 AUTOREFRESH 4L 5C 32246P Autorefresh Query Execute duration for

7 child(ren) table(s): 32

08:56:57.870 19410 AUTOREFRESH 4L 5C 32246P Autorefresh Query Fetch duration for 7 child(ren) table(s): 0

08:56:57.870 19411 AUTOREFRESH 4L 5C 32246P Autorefresh Query Apply duration for 7 child(ren) table(s): 0

08:56:57.870 19412 AUTOREFRESH 3L 5C 32246P Incremental autorefresh ended for table TESTUSER.READTAB

08:56:57.872 19413 AUTOREFRESH 2L 5C 32246P Duration For Cache Group

TESTUSER.READCACHE: 1020

08:56:57.872 19414 AUTOREFRESH 2L 5C 32246P Num Rows For Cache Group

TESTUSER.READCACHE: 1

08:56:57.872 19415 AUTOREFRESH 2L 5C 32246P Num Root Rows For Cache Group

TESTUSER.READCACHE: 0

08:56:57.872 19416 AUTOREFRESH 2L 5C 32246P Cumulative Rows for Cache Group

TESTUSER.READCACHE: 11776

08:56:57.872 19417 AUTOREFRESH 2L 5C 32246P Cumulative Root Rows for Cache Group

TESTUSER.READCACHE: 1697

08:56:57.872 19418 AUTOREFRESH 2L 5C 32246P Autorefresh ended for cache group

TESTUSER.READCACHE

08:56:57.883 19419 AUTOREFRESH 1L 5C 32246P Duration For Interval 60000ms: 420

08:56:57.883 19420 AUTOREFRESH 1L 5C 32246P Num Rows For Interval 60000ms: 0

08:56:57.883 19421 AUTOREFRESH 1L 5C 32246P Num Root Rows For Interval 60000ms: 0

08:56:57.883 19422 AUTOREFRESH 1L 5C 32246P Cumulative Rows for Interval 60000ms:

11587

08:56:57.883 19423 AUTOREFRESH 1L 5C 32246P Cumulative Root Rows for Interval

60000ms: 1697

08:56:57.883 19424 AUTOREFRESH 1L 5C 32246P Autorefresh number 1415 ended for interval 60000ms successfully.

27 records dumped

We set AUTOREFRESH tracing back to its default setting (0) and exit ttTraceMon

:

Trace > level autorefresh 0

Trace > {press ENTER – blank line}

1-18

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Using the ttXactAdmin utility

Using the ttXactAdmin utility

The ttXactAdmin

utility displays ownership, status, log and lock information for each outstanding transaction. You can also use it to show all current connections to a data store. The ttXactAdmin

utility is useful for troubleshooting problems with replication, XLA, and asynchronous writethrough cache groups.

Example 1–15 Using ttXactAdmin to diagnose a lock timeout

Use ttXactAdmin

to diagnose a lock timeout. Consider two connections that are trying to update the same row. The following transaction by Connection 1 is in progress:

UPDATE table1 SET c1 = 2 WHERE c1 = 1;

Connection 2 attempts to make the following update:

UPDATE table1 SET c1 = 3 WHERE c1 = 1;

Connection 2 receives the following error:

6003: Lock request denied because of time-out

Details: Tran 2.3 (pid 2880) wants Un lock on rowid 0x00156bbc, table

TTUSER.TABLE1.

But tran 1.21 (pid 2564) has it in Xn (request was Xn). Holder SQL (update table1 set c1 = 2 where c1 = 1;)

The command failed.

The details of the error indicate that transaction 1.21 has a lock on row 0x00156bbc, the row that transaction 2.3 wants to update. ttXactAdmin

displays this information in output that pertains to actions in the entire data store:

$ ttXactAdmin myDSN

2007-03-23 11:26:01.643

c:\datastore\myDSN

TimesTen Release 7.0.2.0.0

Outstanding locks

PID Context TransID TransStatus

Program File Name: ttIsql

Resource ResourceID Mode Name

2564 0xeeb9a8 1.21

Active Database

Row

Table

0x01312d00

0x00156bbc

1910868

IX

Xn

IXn

TTUSER.TABLE1

TTUSER.TABLE1

Program File Name: ttIsql

2880 0xeeb9a8 2.3

Active Database

Table

Command

0x01312d00

1910868

19972120

IX

IXn

S

TTUSER.TABLE1

Awaiting locks

PID Context

2880 0xeeb9a8

TransID Resource ResourceID

2.3

Row 0x00156bbc

RMode HolderTransID HMode Name

Un 1.21

Xn TTUSER.TABLE1

2 outstanding transactions found

See "ttXactAdmin" in the

Oracle TimesTen In-Memory Database Reference

.

Tools for Troubleshooting TimesTen

1-19

Using ODBC tracing

Using ODBC tracing

On Windows, use the ODBC trace facility to verify the sequence and content of your commands. The ODBC trace facility works only if you have linked your application with the ODBC Driver Manager. Enable tracing by double-clicking

ODBC

in the

Control Panel. This opens the ODBC Data Source Administrator. Choose the

Tracing

tab.

On UNIX platforms, ODBC tracing is available only when using a driver manager. To turn on tracing, set the Trace and TraceFile attributes.

Using SNMP traps to detect events

Network management software uses SNMP (Simple Network Management Protocol) to query or control the state of network devices such as routers and switches. These devices can generate alerts called

traps

to inform the network management systems of problems.

TimesTen sends SNMP traps for particular critical events to help facilitate user recovery mechanisms. These events are also recorded in the support log. Exposing them through SNMP traps allows network management software to take immediate action.

How to configure TimesTen to generate SNMP traps as well as how to receive the traps is described in "Diagnostics through SNMP Traps" in the

Oracle TimesTen

In-Memory Database Error Messages and SNMP Traps

.

To understand how network software might be used to detect SNMP traps, use the snmptrapd

program provided in your TimesTen directory:

/

install_ dir

/demo/snmp

. This demo listens on a designated port for SNMP trap messages and either prints the traps to stdout

or logs them to syslogd

. See the

/

install_ dir

/demo/snmp/README.txt

file for details.

Monitoring the TimesTen system tables

Each TimesTen data store contains a group of system tables that store metadata about the current state of the data store. The system tables are described in "System Tables" in the

Oracle TimesTen In-Memory Database System Tables and Limits Reference

.

Note:

You can execute SELECT statements on a system table, but you cannot execute a statement such as INSERT, UPDATE or DELETE on these tables.

Of particular interest when troubleshooting is the SYS.MONITOR table, which contains statistics about certain events that have occurred since the first connection to the data store. For example, the SYS.MONITOR table contains information about the number of connections to the data store; the number of checkpoints taken; the size of the data store; and the amount of memory currently in use. Check the contents of the

SYS.MONITOR table by executing SELECT statements on the columns or by using the ttIsql monitor

command. For an example of how to use the ttIsql monitor command, see "Using the ttIsql Utility" in the

Oracle TimesTen In-Memory Database

Operations Guide

.

The SYS.MONITOR table is useful for troubleshooting performance problems. See

"Reading query plan from the PLAN table" in the

Oracle TimesTen In-Memory Database

Operations Guide

for details. Check the contents of the SYS.MONITOR table by

1-20

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Using the query optimizer executing SELECT statements on the columns or by using the ttIsql showplan command, as described in "Viewing and changing query optimizer plans" in the

Oracle

TimesTen In-Memory Database Operations Guide

.

Using the query optimizer

The query optimizer is an important tool for performance tuning.

For details about using the query optimizer, see:

"The TimesTen Query Optimizer" in the

Oracle TimesTen In-Memory Database

Operations Guide

"Viewing and changing query optimizer plans" in the

Oracle TimesTen In-Memory

Database Operations Guide

If you find that a given query runs more slowly than expected, confirm that the query

optimizer has the latest statistics for the tables in your query, as described in "Update query optimizer statistics" on page 2-18. If, after updating your statistics, your query

still runs too slowly, it is possible that the TimesTen optimizer is not choosing the optimal query plan to answer that query. Under these circumstances, you can adjust how the optimizer generates a plan by using the ttOpt

* procedures described in

"Modifying plan generation" in the

Oracle TimesTen In-Memory Database Operations

Guide

.

Tools for Troubleshooting TimesTen

1-21

Using the query optimizer

1-22

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

2

2

Troubleshooting TimesTen Applications and

Data Stores

This chapter helps you diagnose and remedy some of the problems encountered while using a TimesTen data store.

If you are still having problems with your data store after following the

troubleshooting recommendations in this chapter, please contact Technical support .

This chapter includes the following topics:

Unable to start or stop TimesTen daemon

No response from TimesTen daemon or subdaemon

Unable to create shared segment

Application unable to connect to data store in direct mode

Troubleshooting Client/Server problems

Application connects or disconnects are slow

Application becomes disconnected unexpectedly

Application is slow

Application unresponsive, appears hung

Application unable to find previously created objects

Troubleshooting OCI and Pro*C/C++ applications

Running out of a resource

Duplicate results from a SELECT statement

Cannot attach PL/SQL shared memory

Unable to start or stop TimesTen daemon

This section describes what to check if you are unable to start or stop the TimesTen main daemon.

Possible cause

Incorrect privilege

What to do

You need the ADMIN privilege to start or stop the

TimesTen daemon. Ensure that you are using the ttDaemonAdmin

utility to start the daemon. The output from ttDaemonAdmin

shows whether you have the correct privilege.

Troubleshooting TimesTen Applications and Data Stores

2-1

No response from TimesTen daemon or subdaemon

Possible cause

Another process is using the

TimesTen daemon port.

TimesTen daemon is already running.

Other problems

What to do

Use the ttVersion

utility to verify what port number the

TimesTen daemon is expected to use. Use an OS command like netstat

to check whether another process is listening on the port. If there is a conflict, either change the port number used by the other process or use ttmodinstall

to change the port used by TimesTen.

Ensure that you are using the ttDaemonAdmin

utility to start the daemon. The output from ttDaemonAdmin shows whether the daemon is already running.

Inspect the user error log produced by the daemon. See

"Using the logs generated by the TimesTen daemon" on page 1-6.

No response from TimesTen daemon or subdaemon

This section describes what to do if one or more of the TimesTen processes appears to be unavailable:

Check the TimesTen user error log

Extract a stack trace from the core file

Check the TimesTen user error log

If you receive an error that indicates the TimesTen subdaemon has stopped, inspect

the user error log, as described in "Using the logs generated by the TimesTen daemon" on page 1-6.

If the TimesTen daemon crashes, it cannot send anything to the user error log, but the subdaemons send a 'main daemon vanished' message to the log before exiting:

09:24:13 Err : 4375 ------------------: Main daemon has vanished

Restart the daemon. The next connection to each data store causes TimesTen to recover from the checkpoint and transaction log files. See "Working with the Oracle TimesTen

Data Manager Daemon" in the

Oracle TimesTen In-Memory Database Operations Guide

.

Extract a stack trace from the core file

If you experience a crash by one of the TimesTen processes on a UNIX system and have exhausted all of the diagnostic options, check to see if TimesTen has generated a core file. Use the ttVersion

utility to find the core file. Look for a line in the output that shows a path for the daemon home directory:

TimesTen Release (ttuser:40732)

2007-04-04T17:53:04Z

Instance admin: ttuser

Instance home directory:

/node1/ttuser/ttcur/TTBuild/linux86_dbg/install

Daemon home directory:

/node1/ttuser/ttcur/TTBuild/linux86_dbg/install/info

After locating the core file, attach to the debugger on the system and extract the stack

trace from the core file and send the trace results to Technical support .

On Windows systems you can obtain diagnostic information for a service failure by enabling the 'allow service to interact with desktop' option in the properties dialog for the TimesTen data manager in the Service menu. If a fatal fault occurs in the TimesTen

2-2

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Application unable to connect to data store in direct mode data manager service, a pop-up asks if you would like to start the debugger. Contact

Technical support and provide the stack trace.

Unable to create shared segment

You may receive an error that indicates that a shared segment could not be created:

4671: TT14000: TimesTen daemon internal error: Error 28 creating shared segment,

KEY 0x0201f7eb

4671: -- OS reports too many shared segments in use

4671: -- Confirm using 'ipcs' and take appropriate action

4671: 18538 ------------------: subdaemon process exited

Using the Linux ipcs

command may display information like this:

------ Shared Memory Segments --------

key shmid owner perms bytes nattch status

0x00000000 1098350592 user1 777 10624 2 dest

0x00000000 1084817409 user1 777 2439680 2 dest

0x911fc211 1098383362 user2 666 67108864 1

0x2814afba 170721285 root 666 1048576 1

A status of dest

means the memory segment is marked to be destroyed. nattch shows the number of processes still attached to the memory segment. The ipcrm command cannot free the shared memory until the processes detach from the segment or exit. If an application connects to TimesTen and then becomes inactive, nothing can free the shared memory until the user exits or stops the application.

Application unable to connect to data store in direct mode

This section describes what to check if your application is unable to connect to a data store in direct mode.

Possible cause See...

Mismatch between the release of

TimesTen and data store

User does not have the CREATE

SESSION privilege.

Incorrect file permissions

Incompatible connection attributes or incorrect path name for data store set in the DSN

No available shared memory segment or maximum size of shared memory segment too small

"Upgrading your data store" on page 2-4

"Privileges to connect to data store" on page 2-4

TimesTen daemon or Data Manager service not running

"Check file system permissions to access data store" on page 2-4

"Check that the TimesTen daemon is running" on page 2-4

"Check DSN definition" on page 2-4

"Manage semaphores and shared memory segments" page 2-5

on

Not enough swap space

Inadequate number of file descriptors

Other possible causes

"Check available swap space (virtual memory)" on page 2-5

"Increase the number of available file descriptors" on page 2-6

"Using the logs generated by the TimesTen daemon" on page 1-6

Troubleshooting TimesTen Applications and Data Stores

2-3

Application unable to connect to data store in direct mode

Upgrading your data store

A data store is only guaranteed to be accessible by the same minor release of TimesTen that was used to create the data store. When you upgrade the TimesTen software and you would like to use the new release to access a data store that was previously created, create a data store with the new release. Then use the ttMigrate

utility to copy the tables, indexes, and table data from the old data store to the new one.

See "Data Store Upgrades" in the

Oracle TimesTen In-Memory Database Installation Guide

for details.

Privileges to connect to data store

The user must have the CREATE SESSION privilege to connect to the data store. If you do not have access, the administrator must use the GRANT statement to grant you the

CREATE SESSION privilege. See "Granting privileges to connect to the database" in the

Oracle TimesTen In-Memory Database Operations Guide

.

Check file system permissions to access data store

A "permission denied" error is generated if you attempt to connect to a data store and you do not have the proper permissions to access the checkpoint or transaction log files or the directory where those files reside. Check the file system permissions on the files located in the directory specified in the DataStore attribute in your DSN.

Check that the TimesTen daemon is running

If the TimesTen daemon or Data Manager service is not running, an attempt to connect to a data store generates TimesTen error 799 (Unable to connect to daemon; check daemon status).

Use the ttStatus

utility as described in

"Check the TimesTen user error log" on page 2-2 to check the status of the TimesTen daemon.

Check DSN definition

In your DSN description:

Check DSN attributes

Check path name to data store and transaction log directories

Check DSN attributes

Certain connection options or DSN attribute settings combinations are not compatible.

In cases where incompatible settings are used, an error is returned to the application when it attempts to connect to a data store.

Check path name to data store and transaction log directories

Confirm that you have specified the correct path names in the DataStore and LogDir attributes in your DSN. Also confirm that the path names are absolute path names, rather than relative. Otherwise, the path name will be relative to the directory where the application was started.

On Windows, be careful to distinguish between User and System DSNs in the ODBC

Data Source Administrator. Do not create user DSNs because they are visible only to the user who defines them. System DSNs are visible to all users. In particular, if you run a TimesTen application as a Windows service, it runs as the user "SYSTEM" by

2-4

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Application unable to connect to data store in direct mode default and does not see any User DSNs. Make sure that you are not using a mapped drive in the data store path name.

Manage semaphores and shared memory segments

An error is generated if you attempt to connect to or create a shared data store whose size is larger than the maximum size of shared memory segments configured on your system. Also, an error is generated if the system cannot allocate any more shared memory segments.

On UNIX systems, use commands similar to the following:

■ ipcs -ma

to check if you have other shared memory segments using up memory, such as Oracle instances or other instances of TimesTen.

■ ipcrm

to remove a message queue, semaphore set or shared memory segment identifier. Use ipcrm

to clean up semaphores or shared memory segments after a faulty TimesTen shutdown, instance crash, daemon crash or other application issues that use shared memory segments and semaphores. Use

-m

to remove a shared memory segment. Use

-s

to remove a semaphore.

■ ps -eafl

to see how much memory is being used by running processes. ulimit -a

to see if there are any limits on the maximum amount of memory one process can address, maximum file size, and the maximum number of open files.

If a shared memory segment is available but is too small to hold your data store, use the ttSize

utility to estimate the amount of memory required for your tables and then check the values of the PermSize and TempSize attributes to verify the amount of memory established for your data store. "Monitoring PermSize and TempSize attributes" in the

Oracle TimesTen In-Memory Database Operations Guide

describes guidelines for setting the size of your permanent and temporary data partitions. If the amount of memory established for your data store is too large, reset PermSize and

TempSize to smaller values. See "Check the amount of memory allocated to the data store" on page 2-16 for more information. Another option is to increase the maximum

size of the shared memory segment, as described below.

If a data store becomes invalidated because of a system or application failure, a subsequent connection recovers the data store. If recovery fails because you have run out of data store space, then reconnect to the data store with a larger PermSize and

TempSize value than the ones that are currently in effect. If recovery fails because you do not have enough shared memory, then you should increase the maximum size of the shared memory segments for the system.

For more information on how to configure shared memory for TimesTen, see

"Installation prerequisites" in the

Oracle TimesTen In-Memory Database Installation Guide

.

Check available swap space (virtual memory)

There must be enough swap space to back up shared memory.

On UNIX systems, use the swap command to check and add virtual memory to your system.

On Windows systems, check and reset the size of your virtual memory from the

Advanced tab in your Computer Management Properties dialog window.

Troubleshooting TimesTen Applications and Data Stores

2-5

Troubleshooting Client/Server problems

Increase the number of available file descriptors

Each process connected to a TimesTen data store keeps at least one operating system file descriptor open. Additional file descriptors may be opened for each connection if checkpoints are issued, and transactions are committed or rolled back. If you receive an error that all file descriptors are in use when attempting to connect to a data store, then increase the allowable number of file descriptors. See your operating system documentation for limits on file descriptors and information about changing the number of file descriptors.

Troubleshooting Client/Server problems

This section includes the following topics:

Cannot connect to the TimesTen Server

TimesTen Server failed

Cannot find Server DSN

TimesTen Server failed to load DRIVER

Application times out when accessing TimesTen Server

TimesTen Client loses connection with TimesTen Server

Failed to attach to shared memory segment for IPC

Increasing the maximum server connections on Windows XP

Thread stack overflow when using multiple client connections

Out of space when DSN specifies new data store

Also consider the topics described in "Application unable to connect to data store in direct mode" on page 2-3.

Cannot connect to the TimesTen Server

You have not correctly identified the system where the TimesTen Server is running.

On a Windows client machine, select the TimesTen Server in the TimesTen Data

Source Setup dialog that is displayed as part of the ODBC Data Source Administrator.

To verify the TimesTen Server:

1.

2.

On the Windows Desktop, choose

Start

>

Settings

>

Control Panel

.

Double click the

ODBC

icon. This opens the ODBC Data Source -Administrator.

3.

4.

5.

6.

7.

Click the

System DSN

tab. This displays the System Data Sources list.

Select the TimesTen Client data source. This opens the TimesTen Client DSN

Setup dialog.

Click

Servers

. This opens the TimesTen Logical Server List.

Select the TimesTen Server from the list. This opens the TimesTen Logical Server

Name Setup dialog.

Verify that the values for the

Network Address

and

Port Number

are correct. If necessary, change the values.

2-6

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Troubleshooting Client/Server problems

Note:

If you typed the hostname or network address directly into the

Server Name field of the TimesTen Client DSN Setup, the Client tries to connect to the TimesTen Server using the default port.

If the Network Address and Port Number values are correct, the TimesTen Server may not be running. See "Starting and stopping the Oracle TimesTen Data Manager service on Windows" in the

Oracle TimesTen In-Memory Database Operations Guide

for information about starting the server manually. See "Testing connections" in the

Oracle

TimesTen In-Memory Database Operations Guide

for more information about identifying this problem.

On UNIX, specify the TimesTen Server with the TTC_Server connection attribute in the odbc.ini

file on the client machine. If the value specified for TTC_Server is an actual hostname or IP address, the client tries to connect to the TimesTen Server using the default port. In TimesTen, the default port is associated with the TimesTen release number. If the value specified for TTC_Server is a logical ServerName, this logical

ServerName must be defined in the ttconnect.ini

file. The ttconnect.ini

entry for this ServerName needs to correctly define the hostname/IP address and port number on which the TimesTen Server is listening.

If the Network Address and Port Number values are correct, the TimesTen Server may not be running or did not start. See "Starting and stopping the daemon on UNIX" in the

Oracle TimesTen In-Memory Database Operations Guide

for information about starting the server manually. See "Testing connections" in the

Oracle TimesTen In-Memory

Database Operations Guide

for more information about identifying this problem.

TimesTen Server failed

Check the server's log file. Server log messages are stored in the files specified by the

-userlog

and

-supportlog

options in the ttendaemon.options

file. See

"Creating and configuring Client DSNs on UNIX" and "Managing TimesTen daemon options" in the

Oracle TimesTen In-Memory Database Operations Guide

.

The maximum number of concurrent IPC connections to the Server of a particular

TimesTen instance is 24,999. However, TimesTen has a limit of 2043 connections

(direct or client/server) to a single DSN.

Client/server users can change the file descriptor limit to support a large number of connections. For an example, see "Installation prerequisites" in the

Oracle TimesTen

In-Memory Database Installation Guide

.

Cannot find Server DSN

On UNIX, verify that the Server DSN is defined in the sys.odbc.ini

file on the machine running the TimesTen Server.

On Windows, verify that the Server DSN is defined as a System DSN in the ODBC

Data Source Administrator on the machine running the TimesTen Server. See

"Creating and configuring a logical server name on Windows" in the

Oracle TimesTen

In-Memory Database Operations Guide

.

TimesTen Server failed to load DRIVER

This error only occurs on UNIX platforms. Open the sys.odbc.ini

file on the machine running the TimesTen Server and locate the Server DSN you are trying to

Troubleshooting TimesTen Applications and Data Stores

2-7

Troubleshooting Client/Server problems connect. Verify that the dynamic library specified in the DRIVER attribute for the

Server DSN exists and is executable.

Application times out when accessing TimesTen Server

The default TimeOut interval is 60 seconds.

To increase this interval on UNIX, change the value of the TTC_Timeout attribute in the odbc.ini

file.

To set the timeout interval on Windows, see the instructions in "Setting the timeout interval and authentication" in the

Oracle TimesTen In-Memory Database Operations

Guide

.

TimesTen Client loses connection with TimesTen Server

Check to see if the error was due to the Client timing out. Check the TimesTen Server's log to see why the Server may have severed connection with the Client. Use ping to determine if your network is up or try using telnet

to connect to the TimesTen

Server port number.

Failed to attach to shared memory segment for IPC

While using shared memory segment (SHM) as IPC, the application may see the following error message from the TimesTen Client ODBC Driver if the application reaches the system-defined per-process file-descriptor-limit.

SQLState = S1000,

Native Error = 0,

Message = [TimesTen][TimesTen 11.2.1 CLIENT]Failed to attach to shared memory segment for IPC. System error: 24

This may happen during a connect operation to the Client DSN when the shmat system call fails because the application has more open file descriptors than the system-defined per-process file descriptor limit. To correct this problem, you must increase your system-defined per-process file descriptor limit. For more information about file descriptor limits, see "System Limits" in the

Oracle TimesTen In-Memory

Database Reference

.

Increasing the maximum server connections on Windows XP

On Windows XP, by default, there can be approximately 47 child server processes.

You can increase the number of connections by setting the MaxConnsPerServer connection attribute in the ttendaemon.options

file or in the DSN. This increases the number of connections to 47 times the MaxConnsPerServer value.

Thread stack overflow when using multiple client connections

On Solaris, you may receive messages in the user error log about thread stack overflow. On other platforms, you may receive messages about a segmentation fault that mention a possible thread stack overflow.

If these messages occur, increase the server stack size by one of the following methods:

Specify the

-ServerStackSize

option in the ttendaemon.options

file. The ttendaemon.options

file applies to all DSNs in the TimesTen instance.

Specify the ServerStackSize connection attribute for a specific DSN. This takes precedence over the value in the ttendaemon.options

file.

2-8

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Application becomes disconnected unexpectedly

Increasing the server stack size decreases the number of concurrent connections that can be made before running out of swap space.

See "Working with the TimesTen Client and Server" in the

Oracle TimesTen In-Memory

Database Operations Guide

.

Out of space when DSN specifies new data store

You may receive "out of space" messages if you change a DSN to specify a new data store while there are existing connections to the original data store in a system with multiple client connections. This can happen on 32-bit platforms if either data store is close to 2 GB.

Close all connections to the original data store. This causes a new server process to be created for connections to the data store that is now specified in the DSN. Use the ttStatus

utility to list the connections for the old data store. Alternatively, you can restart the server by using the ttDaemonAdmin

utility with the

-restartServer option, which resets all client connections on all DSNs in the instance.

Application connects or disconnects are slow

This section describes what to check if you encounter slow connects and disconnects to a data store.

Possible cause

Data store is being recovered.

ODBC tracing is enabled.

Other possible causes

See...

"Check if data store is being recovered" on page 2-9

"Check ODBC tracing" on page 2-9

"API tracing" on page 1-10

Check if data store is being recovered

A slow connect may indicate that a TimesTen data store is being recovered. This happens only for a first connect.

Check ODBC tracing

On Windows platforms, if ODBC tracing is enabled, it can slow connect and disconnect speeds. Double-click

ODBC

in the Control Panel to open the ODBC Data

Source Administrator. Select the

Tracing

tab and confirm tracing is disabled. See

"Using ODBC tracing" on page 1-20.

Application becomes disconnected unexpectedly

If an application becomes disconnected from a TimesTen data store, one of the following events occurs:

If there was no outstanding transaction, the connection is cleanly removed by the

TimesTen daemon. Other existing connections continue processing as if no problem had occurred.

If there was an outstanding transaction but the application was not in the middle of executing code in the TimesTen library, the transaction is rolled back and the connection is cleanly removed by the TimesTen daemon. Other existing connections continue processing as if no problem had occurred.

Troubleshooting TimesTen Applications and Data Stores

2-9

Application is slow

This section describes what to check if your application unexpectedly disconnects from the data store.

Possible cause

Internal application error.

Failure of a concurrent application thread.

If using a client/server connection, the client may have disconnected from the application.

An error in the TimesTen library

See...

"Check for ODBC or JDBC errors" on page 2-10

"Check for ODBC or JDBC errors" on page 2-10

"Check the user error log" on page 2-10

"Troubleshooting Client/Server problems" on page 2-6

Contact

Technical support

.

Check for ODBC or JDBC errors

Check for the following types of errors:

ODBC errors returned by the

SQLError

function

JDBC errors returned by the

SQLException

class

The application may have encountered a problem that caused it to exit prematurely, which in turn may have caused other connections to be forced to disconnect. Call

SQLError

after each ODBC call to identify error or warning conditions when they first happen. Examples of

SQLError

usage can be found in the demo programs and in

"Retrieving errors and warnings" in the

Oracle TimesTen In-Memory Database Error

Messages and SNMP Traps

.

In more extreme cases, it may be helpful to use ttTraceMon

to generate a level 4 ERR trace for the application and review all of the errors messages that are pushed in the

TimesTen direct driver. See

"ERR tracing" on page 1-13 for details.

Check the user error log

If a TimesTen application disconnects without returning an ODBC error or any other warning, look through the user error log. See

"Using the logs generated by the

TimesTen daemon" on page 1-6.

Application is slow

For details on how to maximize the performance of your application and TimesTen data store, see:

"TimesTen Database Performance Tuning" in the

Oracle TimesTen In-Memory

Database Operations Guide

"Application Tuning" in the

Oracle TimesTen In-Memory Database C Developer's

Guide

"Application Tuning" in the

Oracle TimesTen In-Memory Database Java Developer's

Guide

This section describes some of the issues that impair performance.

Possible cause

Using client/server mode

Outdated database statistics

See...

"Consider connection mode" on page 2-11

"Update statistics for your tables" on page 2-11

2-10

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Application is slow

Possible cause See...

Committing transactions too frequently

"Turn off autocommit mode" in the

Oracle TimesTen

In-Memory Database Operations Guide

DurableCommits attribute enabled "Use durable commits appropriately" in the

Oracle

TimesTen In-Memory Database Operations Guide

Not preparing SQL statements used more than once

"Prepare statements in advance" in the

In-Memory Database Operations Guide

Oracle TimesTen

Wrong kind of index, too many indexes, wrong size for hash index

"Select hash, range, or bitmap indexes appropriately" in the

Oracle TimesTen In-Memory Database Operations Guide

"Size hash indexes appropriately" in the

Oracle TimesTen

In-Memory Database Operations Guide

Inefficient use of locks

Improperly configured materialized view

"Verify lock and isolation levels" on page 2-12

"Performance implications of materialized views" and

"Materialized view tuning" in the

Oracle TimesTen

In-Memory Database Operations Guide

"Poor replication or XLA performance" on page 6-11

If replication is used, configuration of replication scheme or network environment may be impacting application.

If IMDB Cache is used, IMDB Cache configuration or environment may be impacting application.

"Poor autorefresh performance" on page 4-22

Too many table partitions

Tracing is unnecessarily enabled for one or more TimesTen components.

"Check partition counts for the tables" on page 2-12

"Check trace settings" on page 2-12

Consider connection mode

Client/server connections are slower than direct connections to TimesTen data stores.

Driver manager connections can also moderately impact performance. The performance overhead imposed by client/server connections can be significant because of the network latencies involved in all communication with the data store.

If your application must run on a different machine from the one hosting the data store, see "Client/Server tuning" in the

Oracle TimesTen In-Memory Database Operations

Guide

.

Update statistics for your tables

The TimesTen query optimizer in general is very good at choosing the most efficient query plan. However, it needs additional information about the tables involved in complex queries in order to choose the best plan. By knowing the number of rows and data distributions of column values for a table, the optimizer has a much better chance of choosing an efficient query plan to access that table.

Before preparing queries that will access a TimesTen table, use the ttOptUpdateStats

procedure to update the statistics for that table. When updating the statistics for a table, you get the best results if you update statistics on your tables after loading them with data, but before preparing your queries. For example, if you update statistics on a table before populating it with data, then your queries are optimized with the assumption that the tables contain no rows (or very few). If you later populate your tables with millions of rows and then execute the queries, the plans that worked well for the situation where your tables contained few rows may now be very slow.

Troubleshooting TimesTen Applications and Data Stores

2-11

Application is slow

For more information about updating statistics, see "The TimesTen Query Optimizer" in the

Oracle TimesTen In-Memory Database Operations Guide

.

Verify lock and isolation levels

The manner in which multiple applications concurrently access the data store can have a major impact on performance.

An application can acquire locks on the entire data store, individual tables, and individual rows. Additionally, applications can set an isolation level that determines whether they hold read and update locks until their transactions commit or roll back.

Check the SYS.MONITOR table or use the ttXactAdmin

utility to detect whether an application is spending time waiting for locks. See

"Check for deadlocks and timeouts" on page 2-14 and

"Using the ttXactAdmin utility" on page 1-19.

If lock contention is high, you may be able to improve the overall performance of your system by implementing the following:

Set the LockLevel configuration attribute or use the ttLockLevel

procedure to place locks on rows, rather than on the entire data store. (Row locking is the default.)

Use the ttOptSetFlag

procedure to prevent the query optimizer from placing locks on tables. (Table locks are sometimes the default, particularly for updates that affect many rows.)

Use read-committed isolation level (Isolation=1, the default) for those applications do not require serializable access to the transaction data.

If you see a lot of lock contention, but the above settings are all set to minimize contention, then the contention may be related to the application itself. For example, concurrent threads may be repeatedly accessing the same row. The ttXactAdmin utility can sometimes help you detect this sort of contention. Tracing can also be useful in this situation.

For more information about locks and isolation levels, see "Concurrency control through isolation and locking" in the

Oracle TimesTen In-Memory Database Operations

Guide

.

Check trace settings

Use ttTraceMon -e show

as described in

"Using the ttTraceMon utility" on page 1-6

to confirm tracing is off on all TimesTen components. ERR should be set to 1; all other components should be set to 0. Trace levels are preserved when a data store is reloaded.

On Windows platforms, confirm that ODBC tracing is disabled. Double-click

ODBC

in the Control Panel to open the ODBC Data Source Administrator. Select the

Tracing

tab

and confirm tracing is disabled. See "Using ODBC tracing" on page 1-20.

Check partition counts for the tables

When a table is created, it has one partition. When you use ALTER TABLE ... ADD

COLUMN to add new columns, a new partition is added to the table. Adding multiple columns with a single ALTER TABLE ... ADD COLUMN statement only adds one partition.

There is a limit of 255 partitions per table. Exceeding this number generates an 8204 error. An extra read for each new partition slightly degrades performance for each of

2-12

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Application unresponsive, appears hung the new partitions. A high partition count should be avoided. On replicated tables that have multiple partitions, additional space is used for each update on the subscriber side, proportional to the number of partitions. This can result in the subscribers using slightly more perm space than the master.

The partition value for each table is tracked in the SYS16 column of the system table,

SYS.TABLES. Obtain the partition counts for tables by using the following query:

SELECT tblname, sys16 FROM SYS.TABLES;

If you discover that a table has too many partitions, do

one

of the following:

Re-create the table

Save and restore the table. Use ttMigrate -c

to create a migration file. Then restore the table without additional partitions by using ttMigrate -r

-noRepUpgrade

.

ALTER TABLE ... DROP COLUMN does not remove partitions from a table. On replicated systems, all master and subscriber data stores must be migrated using the

-noRepUpgrade

option. Replication does not occur for tables that have different partition structures.

Application unresponsive, appears hung

This section describes what to check if your application is unresponsive and appears to be hung.

Possible cause

All causes

Internal application error

Inconsistent connection attributes set in

DSN

Excessive lock contention

See...

"Check logs and gather trace information" on page 2-13

"Check for ODBC errors" on page 2-13

"Consider connection mode" on page 2-11

"Check for deadlocks and timeouts" on page 2-14

Check logs and gather trace information

If your application hangs, check the transaction log by using the ttXactAdmin

utility.

See

"Using the ttXactAdmin utility" on page 1-19.

Also check the user error log for errors, as described in "Using the logs generated by the TimesTen daemon" on page 1-6 .

You can also generate a trace log to detect the activities on various TimesTen components as described in

"Using the ttTraceMon utility" on page 1-6 .

Check for ODBC errors

Check the ODBC errors returned by the

SQLError

function in all applications to determine whether one of them has encountered a problem that caused it to hang. Call

SQLError

after each ODBC call to identify error or warning conditions when they first happen. Examples of

SQLError

usage can be found in the demo programs and in

"Retrieving errors and warnings" in the

Oracle TimesTen In-Memory Database Error

Messages and SNMP Traps

.

Troubleshooting TimesTen Applications and Data Stores

2-13

Application unable to find previously created objects

If the problem is repeatable, use ttTraceMon

to generate a SQL trace to determine where the application is hanging. See

"SQL tracing" on page 1-9 for details. In more

extreme cases, it may be helpful to generate a level 4 ERR trace for the application and review all of the errors messages that are pushed in the TimesTen direct driver. See

"ERR tracing" on page 1-13 for details.

Check for deadlocks and timeouts

If there is no connect problem, a deadlock or timeout may be the problem. The

SYS.MONITOR table records information about deadlocks and timeouts. See

"Monitoring the TimesTen system tables" on page 1-20 for information on how view

the contents of this table. You can also use the ttXactAdmin

utility to detect the types of locks currently held by uncommitted transactions and the resources on which they are being held.

If a deadlock occurs, the TimesTen subdaemon negotiates the problem by having an application involved in the deadlock generate TimesTen error 6002 (Lock request denied because of deadlock). The error message contains the SQL that the lock holder is running, which can help you diagnose the cause of the deadlock. If your application encounters this error, it should roll back the transaction and then reissue the statements for that transaction. Deadlocks can be caused if your application issues statements in a particular order that results in a circular wait, and can sometimes be prevented by changing the order in which the statements are issued.

An application encounters TimesTen error 6003 (Lock request denied because of timeout) if it is unable to acquire a lock within the time period defined by the lock timeout interval set by the LockWait attribute in the DSN or by the ttLockWait procedure in your application. Upon encountering a timeout error, your application can reissue the statement. Keeping transactions short reduces the possibility of lock timeout errors.

System tables are a common source of lock contention. Reduce contention on the system tables by executing prepared statements, rather than executing the same statements directly each time.

In multithreaded applications, a thread that issues requests on different connection handles to the same data store may encounter lock conflict with itself. TimesTen resolves these conflicts with lock timeouts.

Application unable to find previously created objects

This section describes what to check if your application is unable to locate previously created tables, indexes, sequences or views in the data store.

Possible cause

No owner or incorrect owner specified

User does not have SELECT privileges to tables.

Data store is temporary.

Overwrite attribute is enabled.

Path name specified in DSN is relative.

See...

"Specify object owner" on page 2-15

"Check privilege to access tables" on page 2-15

"Check temporary DSN attribute" on page 2-15

"Check Overwrite DSN attribute" on page 2-15

"Check path name to data store" on page 2-15

2-14

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Application unable to find previously created objects

Specify object owner

Tables, indexes and sequences can be created either with a single name, such as

PARTS,

or with a qualified name incorporating an owner and table name, such as

STAN.PARTS

. When accessing a table or index, if no owner is specified, TimesTen first assumes that the owner is the login ID of the user (the value of the UID attribute). If

TimesTen cannot find the table or index under the user's login ID, it then assumes that the owner is user SYS.

If applications need to connect to a data store as different users and share objects, explicitly specify the owners of the objects when they are created and referenced.

Check privilege to access tables

All privileges for the user can be viewed in the SYS.USER_SYS_PRIVS table, which contains all of the system-level privileges for a given user, and the SYS.USER_TAB_

PRIVS table, which contains all of the object-level privileges for a given user. Check these tables to verify if you have 'SELECT' privilege for the tables. If you do not have

'SELECT' privilege for the tables, the privilege may be granted with the GRANT statement. The method for granting privileges is described in the "Managing Access

Control" chapter in the

Oracle TimesTen In-Memory Database Operations Guide

.

Check temporary DSN attribute

Temporary data stores (DSN attribute: Temporary=1) persist until all connections to the data store have been removed. When attempting to access a table in a temporary data store and the table does not exist, it is possible that the data store in which the table resided in has been dropped.

Check Overwrite DSN attribute

If the Overwrite and AutoCreate DSN attributes are enabled and the data store already exists, TimesTen drops that data store and creates a new one. Any tables that were created in the old data store are dropped.

Check path name to data store

To ensure that you are always accessing the same data store when connecting to a particular DSN, use an absolute data store path name instead of a relative one. For example, if the demo data store is in the datastore directory, specify:

DataStore=/datastore/demo rather than:

DataStore=demo

In the latter case, the data store path name is relative to the directory where the application was started. If you are unable to find a table and you are using a relative data store path name, it is possible that the data store in which the table resides in does exist but the data store (checkpoint and log) files are in a different directory than the one that you are accessing.

See "Specifying Data Source Names to identify TimesTen data stores" in the

Oracle

TimesTen In-Memory Database Operations Guide

.

Troubleshooting TimesTen Applications and Data Stores

2-15

Troubleshooting OCI and Pro*C/C++ applications

Troubleshooting OCI and Pro*C/C++ applications

On Windows, the

NLS_LANG

setting is taken from the registry if it is not in the environment. If

NLS_LANG

is set to an unsupported value, such as

NA

, an OCI connection failed error or an ORA-12705 error is thrown. If your OCI or Pro*C/C++ program has trouble connecting to TimesTen, verify that the setting of

HKEY_LOCAL_

MACHINE\Software\ORACLE\NLS_LANG

is valid and indicates a character set supported by TimesTen. This is likely only an issue on machines that previously had

Oracle9

i

or earlier Oracle versions installed.

Refer to the "Globalization support" section in the OCI chapter of the

Oracle TimesTen

In-Memory Database C Developer's Guide

for more information on

NLS_LANG

.

Running out of a resource

This section describes what to check if TimesTen runs out of resources such as memory space, disk space, file descriptors, and semaphores.

Symptom

Memory consumption seems high.

Running out of memory space

Running out of disk space

Running out of transaction log space

Running out of file descriptors

Running out of semaphores

Running out of CPU

See...

"Operating system tools and shared memory" on page 2-16

"Operating system tools and shared memory" on page 2-16

"Check the amount of memory allocated to the data store" on page 2-16

"Update query optimizer statistics" on page 2-18

"Check memory used by queries" on page 2-18

"Check available swap space (virtual memory)" on page 2-18

"Check transaction log file use of disk space" on page 2-18

"Check transaction log file use of disk space" on page 2-18

"Increase the number of available file descriptors" on page 2-6

"Check the semaphore limit" on page 2-20

Obtain a stack trace and contact

Technical support

.

Operating system tools and shared memory

Operating system tools such as top

, vmstat

, and sar

provide statistics about processes and memory usage. The output from these tools can be misleading as an indicator of TimesTen memory consumption because they report shared memory usage for each process but do not report total shared memory usage. Adding together various memory statistics for TimesTen processes overestimates the amount of memory used by TimesTen because shared memory is by definition shared.

Check the amount of memory allocated to the data store

TimesTen uses both permanent and temporary data partitions. The amount of memory allocated for these partitions is set by the PermSize and TempSize attributes in the DSN definition for the data store.

2-16

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Running out of a resource

When the TimesTen data store fills up, it is important to determine whether it is the permanent or the temporary segment that is filling up. Use the ttIsql dssize command to list allocated, in-use, and high water mark sizes for the permanent and temporary data partitions. The dssize

command selects the following values from

SYS.MONITOR:

PERM_ALLOCATED_SIZE

PERM_IN_USE_SIZE

PERM_IN_USE_HIGH_WATER

TEMP_ALLOCATED_SIZE

TEMP_IN_USE_SIZE

TEMP_IN_USE_HIGH_WATER

The permanent segment consists of table and index data, while the temporary segment consists of internal structures, such as locks, sorting areas, and compiled commands.

Keeping transactions short and making sure there is enough temporary space in the data store prevents locks from occupying all of the remaining temporary space. You can also use table locks if transactions are acquiring tens of thousands of row locks.

For tips on how to estimate the size of your data store, see "Size your data store correctly" in the

Oracle TimesTen In-Memory Database Operations Guide

.

Permanent segment filling up

Consider whether you can drop any indexes. You may want to look at query plans to see which indexes are actually used. See "Viewing and changing query optimizer plans" in the

Oracle TimesTen In-Memory Database Operations Guide

. You can also use the ttRedundantIndexCheck

procedure to discover redundant indexes. The procedure returns suggestions about which indexes to drop.

Use the ttSize

utility to estimate the amount of memory used by each table in the data store. If the amount of data you need to store is too big, you may need to reset the

PermSize attribute for the data store to increase the size of the permanent segment.

Alternatively, you may need to partition your data into several different data stores if, for example, you cannot shrink the temporary segment or create a bigger data store because of limits on the memory segment size.

Sometimes when the permanent segment fills up, copying the data out of the data store, deleting all the data, and copying it back in frees up space. This can be done more efficiently by using the ttMigrate

utility with the

-noRepUpgrade

option to migrate the data out, destroy and re-create the data store, and migrate the data back in. This operation is described in "Reducing data store size" in the

Oracle TimesTen

In-Memory Database Installation Guide

.

Finally, you may have to configure the operating system to allow a larger amount of shared memory to be allocated to a process. You may also have to allocate more swap space for virtual memory. See

"Check available swap space (virtual memory)" on page 2-18.

Temporary segment filling up

Some commands may be allocating too much space because of out-of-date statistics.

See

"Update query optimizer statistics" on page 2-18.

If updating the statistics does not reduce temporary segment memory usage, disconnect all connections and then reconnect them. Verify that all connections have

Troubleshooting TimesTen Applications and Data Stores

2-17

Running out of a resource been disconnected by using the ttStatus

utility. That frees up all temporary space, but you must reprepare commands.

Diagnose memory usage by queries. See "Check memory used by queries" on page 2-18.

If the problem is chronic, monitor the data store to try to identify the source of the problem. Use the ttWarnOnLowMemory

procedure to enable warnings in the user log that indicate that the data store is filling up.

Update query optimizer statistics

If the data store seems to have enough free space but runs out of data store space when executing a query, make sure you have updated the optimizer statistics with the ttOptUpdateStats

or ttOptEstimateStats

procedure. To execute some queries,

TimesTen needs to allocate temporary space. The amount of temporary space required is estimated from statistics about the tables used by the query. Without correct statistics, the temporary space required may be underestimated.

See

"Using the query optimizer" on page 1-21.

Check memory used by queries

You can check the memory that a query uses by observing the high water mark for temporary memory usage. The high water mark represents the largest amount of in-use temporary space used since the high water mark was initialized or reset.

Complete the following tasks:

1.

Use the ttIsql dssize

command to check TEMP_IN_USE_SIZE and TEMP_IN_

USE_HIGH_WATER. (Alternatively, you can query the SYS.MONITOR table for these values.)

2.

3.

4.

Call the ttMonitorHighWaterReset

procedure to reset the TEMP_IN_USE_

HIGH_WATER to the current value for TEMP_IN_USE_SIZE.

Execute a query.

Use dssize

to check TEMP_IN_USE_HIGH_WATER for peak memory usage for the query.

Check available swap space (virtual memory)

If you receive an error indicating that you have run out of swap space, you may need to increase the amount of available swap space (also referred to as "virtual memory").

On UNIX systems, use the swap

command to check and reset the amount of virtual memory currently established for your system.

On Windows systems, check and reset the size of your virtual memory by choosing

Control Panel

>

System

>

Advanced

.

Check transaction log file use of disk space

TimesTen saves a copy of the data store in one of two checkpoint files, which are stored in the directory specified by the DataStore attribute. Each checkpoint file can grow on disk to be equivalent to the size of the data store in shared memory. For each permanent data store, you must have enough disk space for the two checkpoint files and for transaction log files.

2-18

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Running out of a resource

Transaction log files accumulate in the directory specified by the LogDir attribute and are only deleted when checkpoints are performed. If the LogDir attribute is not specified in the DSN, transaction log files accumulate in the directory specified by the

DataStore attribute. The maximum size of your transaction log files is set by the

LogFileSize attribute.

When a disk fills up with TimesTen data, it is most often due to a build-up of transaction log files. Transaction log files are used for numerous purposes in

TimesTen, including checkpointing, backups, and replication. It is important to determine which operation is putting a "hold" on the transaction log files, so that appropriate action can be taken to allow the transaction log files to be purged. This can be done by using the ttLogHolds

built-in procedure. There are six types of log holds.

They are discussed in detail below.

Checkpoint

- If a TimesTen application crashes and the data store needs to be recovered, the checkpoint files and transaction log files are used to recover the data. The "most recent" transaction log files are used -- those written since the checkpoint was done. Transaction log files accumulate during the interval between checkpoints. Your application should periodically call the ttCkpt

or ttCkptBlocking

procedure to checkpoint the data and free up the space on the disk. If checkpoints are done very infrequently, a large number of transaction log files may accumulate, particularly if many changes are made to the data store during that interval. See "Checkpoint Operations" in the

Oracle TimesTen

In-Memory Database Operations Guide

.

Replication

-TimesTen replication transmits changes to one data store to one or more other data stores. It does this by reading the log and sending any relevant changes. If replication is paused, the transaction log files build up. To prevent log build-up, avoid pausing replication for too long. Delete subscriptions entirely, and reset replication where appropriate. See "Setting the replication state of subscribers" in

Oracle TimesTen In-Memory Database TimesTen to TimesTen

Replication Guide

for more information on pausing and restarting or resetting replication.

Backup

- TimesTen supports an incremental backup facility that uses transaction log files to augment a backup with changes made since the last backup.

Transaction log files accumulate during the interval between incremental backups.

To avoid a large log build-up, do incremental backups at relatively frequent intervals. If desired, disable incremental backups and do full backups instead. See

"Copying, migrating, backing up and restoring a data store" in the

Oracle TimesTen

In-Memory Database Operations Guide

.

XLA

- TimesTen's persistent XLA facility reports changes to the data store by using transaction log files. Transaction log files are kept until the corresponding transactions have been acknowledged using the ttXlaAcknowledge

C function.

Call ttXlaAcknowledge

frequently enough to prevent transaction log files building up. See "Retrieving update records from the transaction log" in the

Oracle

TimesTen In-Memory Database C Developer's Guide

.

XA

- TimesTen's XA support uses transaction log files to resolve distributed transactions. If these transactions are not resolved in a timely manner, transaction log files build up. See "Distributed Transaction Processing XA" in the

Oracle

TimesTen In-Memory Database C Developer's Guide

.

Long-running transactions

- TimesTen uses the transaction log to roll back transactions. A log hold is placed for the duration of a transaction. Transactions that are active for a long time result in log file building up if the transaction has written at least one log record. (That is, it is not a read-only transaction.) Commit write transactions with reasonable frequency to avoid significant log file build-up.

Troubleshooting TimesTen Applications and Data Stores

2-19

Duplicate results from a SELECT statement

See "Size transactions appropriately" in the

Oracle TimesTen In-Memory Database

Operations Guide

for more information on transaction length.

The following attributes are related to disk use:

The LogPurge attribute indicates whether transaction log files that no longer have a hold on them are purged (removed from the disk) or simply archived (renamed).

If the LogPurge attribute is set to the default value of 0, TimesTen renames transaction log files that it no longer needs by appending the string

.arch

to the name. Once renamed, you must delete the transaction log files manually when they are no longer needed. If transaction log files are not purged, they continue to accumulate space, even when no longer needed by TimesTen.

The Preallocate attribute indicates whether disk space should be reserved for checkpoint files at connect time. This is useful for big data stores, to ensure that the disk always has room for the checkpoint files as data is added to the data store.

Check the semaphore limit

When creating multiple client/server connections to a TimesTen data store configured to allow shared memory segment as IPC, you may encounter errors that indicate

TimesTen was unable to create a semaphore.

Semaphore limits are platform-dependent. See your operating system documentation and "Increase number of semaphores" in the

Oracle TimesTen In-Memory Database

Installation Guide

.

Duplicate results from a SELECT statement

Using read-committed isolation level can lead to duplicates in a result set. A SELECT statement selects more or fewer rows than the total number of rows in the table if some rows are added or removed and committed in the range in which the SELECT scan is occurring. This may happen when an UPDATE, INSERT or DELETE statement adds or deletes a value from an index and the SELECT scan is using this index. This can also happen when an INSERT or DELETE adds or deletes rows from the table and the SELECT operation is using an all-table scan.

Index values are ordered. An UPDATE of an index value may delete the old value and insert the new value into a different place. In other words it moves a row from one position in the index to another position. If an index scan sees the same row in both positions, it returns the row twice. This does not happen with a serial scan because table pages are unordered and rows do not need to be moved around for an UPDATE.

Hence once a scan passes a row, it will not see that same row again.

The only general way to avoid this problem is for the SELECT statement to use serializable isolation. This prevents a concurrent INSERT, DELETE or UPDATE operation. There is no reliable way to avoid this problem with INSERT or DELETE by forcing the use of an index because these operations affect all indexes. With UPDATE, this problem can be avoided by forcing the SELECT statement to use an index that is not being updated.

For more information about serializable isolation, see "Concurrency control through isolation and locking" in the

Oracle TimesTen In-Memory Database Operations Guide

.

Cannot attach PL/SQL shared memory

You can receive error 8517 "Cannot attach PL/SQL shared memory;

PLSQL_MEMORY_

ADDRESS

not valid or already in use" for any of the following reasons:

2-20

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Cannot attach PL/SQL shared memory

User allocated memory already uses that address.

Some shared memory already uses that address.

A shared library already uses that address.

To recover, specify a virtual address that is free in all processes that can connect to the database. If you have a program on a 32-bit operating system that allocates large amounts of memory before connecting to TimesTen, it may clash with the PL/SQL shared memory segment. In this case, either allocate memory after connecting to

TimesTen or use a 64-bit operating system.

Troubleshooting TimesTen Applications and Data Stores

2-21

Cannot attach PL/SQL shared memory

2-22

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

3

3

Troubleshooting Installation,

Upgrades and Downgrades

This chapter includes the following topics:

Installing 32-bit TimesTen on 64-bit Windows

Downgrading a data store with Oracle data types to TimesTen 6.0

Installing 32-bit TimesTen on 64-bit Windows

The default ODBC Data Source Administrator on 64-bit Windows does not show

TimesTen 32-bit drivers and DSNs. If Windows is installed in the default location

(

C:\WINDOWS

), use

C:\WINDOWS\SysWOW64\odbcad32.exe

for the ODBC Data

Source Administrator when you are installing 32-bit TimesTen on a 64-bit Windows machine.

Downgrading a data store with Oracle data types to TimesTen 6.0

In rare situations, after upgrading a data store from TimesTen 6.0, you may find that you need to downgrade a TimesTen 7.0 or later data store back to TimesTen 6.0 after the data types are already converted to Oracle types. However, the ttMigrate

utility for TimesTen 6.0 does not understand Oracle data types, and this can lead to problems when downgrading data stores from TimesTen 7.0 or later. To avoid any pitfalls in the downgrade process, you should convert the Oracle data types back to TimesTen types using TimesTen 7.0 or later first, and only then downgrade the data store to TimesTen

6.0, using the following steps:

1.

Create a migration file using TimesTen 7.0 or later ttMigrate

.

ttMigrate -c datastore datastore.migrate

2.

Destroy the data store using TimesTen 7.0 or later ttDestroy

.

ttDestroy

datastore

3.

Convert the data types to TimesTen types using TimesTen 7.0 or later ttMigrate

.

ttMigrate -r -noRepUpgrade -convertTypesToTT datastore datastore.migrate

4.

Create a new migration file using TimesTen 7.0 or later ttMigrate

.

ttMigrate -c datastore datastore.migrate

5.

Destroy the data store using TimesTen 7.0 or later ttDestroy

.

ttDestroy

datastore

Troubleshooting Installation, Upgrades and Downgrades

3-1

Downgrading a data store with Oracle data types to TimesTen 6.0

6.

In another terminal, with the environment set correctly for TimesTen 6.0, restore the data store as a TimesTen 6.0 data store using TimesTen 6.0 ttMigrate

.

ttMigrate -r datastore

datastore

.migrate

Note:

Before restoring the data store with TimesTen 6.0 ttMigrate

, you must modify the DSN attributes appropriately for using with

TimesTen 6.0.

3-2

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

4

4

Troubleshooting Oracle In-Memory Database

Cache

This chapter describes how to troubleshoot some of the problems you may encounter when using Oracle In-Memory Database Cache (IMDB Cache). It includes the following topics:

Unable to create a cache group

Unable to start or stop the cache agent

Recovering cache grid after unexpected system shutdown

Unable to resolve Oracle Service Name

Unable to resolve connect identifier

Incompatible Oracle Server and Client versions

Unable to validate Oracle username and password

OCI initialization failed

Unsupported data type mapping

Null constraint does not match Oracle

DDL operations on cached Oracle tables may cause cache group operations to fail

Changes not visible after updating object in cache group

Loading or refreshing fails

Monitoring autorefresh cache groups

Optimize Performance for IMDB Cache

Autorefresh not refreshing cache at the specified interval

Incremental autorefresh not progressing

Incremental autorefresh becomes full autorefresh

Poor autorefresh performance

If you are having problems with an AWT cache group, see also Chapter 5,

"Troubleshooting AWT Cache Groups" .

Unable to create a cache group

This section describes some of the problems you might encounter when executing the

CREATE CACHE GROUP statement.

Troubleshooting Oracle In-Memory Database Cache

4-1

Unable to start or stop the cache agent

Possible cause What to do

User does not have the correct Oracle privileges to create the cache group type.

User has insufficient access to data store.

The internal/external user does not match the

Oracle user.

Cannot connect to Oracle

See "Check Oracle privileges" on page 4-6.

You must have CACHE_MANAGER privilege to create a cache group.

The TimesTen user name must be the same as the Oracle user name.

See:

"Unable to resolve Oracle Service Name" on page 4-4

"Unable to resolve connect identifier" on page 4-4

"Unable to validate Oracle username and password" on page 4-5

"Incompatible Oracle Server and Client versions" on page 4-5

Check whether Oracle needs to be restarted.

Check the network status.

See "Set the cache administration user id and password" on page 4-7.

Cache administration user ID or password not set (when trying to create AWT or autorefresh cache groups)

Unsupported data type mapping

Different nullability setting in Oracle

Failure to specify primary key in root table

See "Unsupported data type mapping" on page 4-9.

See "Null constraint does not match Oracle" on page 4-9.

The root table of a cache group must have a primary key. See "Defining cache groups" in the

Oracle In-Memory Database Cache User's

Guide

.

Unable to start or stop the cache agent

This section describes some of the problems you might encounter when starting or stopping the cache agent.

Possible cause

Cache agent already running

Unable to locate Oracle libraries

ORACLE_HOME is invalid.

Insufficient privileges

Wrong OracleID

What to do

See

"Check status of the cache agent" on page 4-3.

See

"Check status of TNS listener and Oracle

Server" on page 4-6.

Check the permissions on the libraries.

See "Check ORACLE_HOME environment variable" on page 4-3.

You must have CACHE_MANAGER privilege to start or stop the cache agent.

Ensure that the OracleID set in your DSN definition matches the Oracle Service Name for the Oracle instance that contains the tables to cache in TimesTen.

4-2

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Recovering cache grid after unexpected system shutdown

Check status of the cache agent

Check the status of the cache agent by using the ttStatus

utility as described in

"Using the ttStatus utility" on page 1-2 to check the status of the cache agent.

If the cache agent is not running, start it as described in "Starting the cache agent" in the

Oracle In-Memory Database Cache User's Guide

. If attempts to start the cache agent fail, then investigate the possible causes and reboot the machine before attempting to start the cache agent.

Check ORACLE_HOME environment variable

On UNIX or Linux platforms, check that the ORACLE_HOME environment variable is set correctly for the shell from which you are starting the cache agent and the

TimesTen daemon. Use the ttmodinstall

utility if you need to change the setting for ORACLE_HOME.

See "Environment variables" in

Oracle TimesTen In-Memory Database Installation Guide

.

Check NLS environment variables

NLS environment variables are set in the environment where the TimesTen application is running, even though TimesTen is not using the NLS environment variables. Unset the NLS environment variables and restart the TimesTen daemon, the cache agent, and the replication agent.

Recovering cache grid after unexpected system shutdown

The server may experience a system failure or an unexpected reboot, such as with a power outage. In this case, the cache grid exits unexpectedly without the normal shutdown procedure.

The following sections describe how to recover when the system unexpectedly shuts down for two scenarios:

A portion of the cache grid nodes are still running

All cache grid nodes exited unexpectedly

A portion of the cache grid nodes are still running

When the server shuts down, some of the cache grid notes exited unexpectedly, but others are still active. In this case, you must detach the dead nodes first by executing ttGridDetachList

from an attached node, as follows:

1.

2.

3.

4.

Connect to a surviving grid node and execute ttGridDetachList

to force a detach of all dead nodes from the grid.

Connect to the datastores on the rebooted server. Start the replication agent by executing ttRepStart

.

Attach the cache grid nodes by executing ttGridAttach

.

Resume normal database operations.

All cache grid nodes exited unexpectedly

If all cache grid nodes exited unexpectedly when the server shut down, perform the following tasks to recover the cache grid:

Troubleshooting Oracle In-Memory Database Cache

4-3

Unable to resolve Oracle Service Name

1.

2.

3.

4.

Log on to each grid node by connecting to the datastores on the rebooted server.

Start the replication agent by executing ttRepStart

. The replication agent will flush the existing log, even if the log is current.

Call ttGridAttach

on each node, which will fail with a communication error because it cannot communicate with other members. The failed attach cleans up the node information.

The last node on which you execute the ttGridAttach

should succeed. At this point you have cleaned up all nodes, so execute ttGridAttach

on all nodes again to attach each node to the grid.

Resume normal database operations.

Unable to resolve Oracle Service Name

If you receive error ORA-12514 indicating "could not resolve service name":

Use the Oracle TNSPING utility to verify that the service can be reached.

Ensure that the

OracleID

set in your DSN definition matches the Oracle Service

Name for the Oracle instance that contains the tables to cache in TimesTen.

Ensure that there is a service name defined. If it is a Windows Oracle client, use

Oracle Net Configuration Assistant to configure a service name. In Oracle Net

Configuration Assistant, navigate to Oracle Net Configuration -> Local -> Service

Naming, select your Oracle server and confirm that there is a service name or a

SID that identifies the Oracle server. If you add or modify a service name, you may need to reboot.

Check the cache administration user name and password on Oracle with SQL*Plus to make sure this service name works. For example:

%sqlplus

cache_admin_user

/

cache_admin_pwd

@

OracleHost cache_admin_user

is the cache administration user name,

cache_admin_pwd

is the cache administration user password, and

OracleHost

is the OracleID specified in your DSN definition.

Note:

Your cache administration user may be different from your regular Oracle user. See "Create Oracle users" in the

Oracle In-Memory

Database Cache User's Guide

.

Ensure that there is only one copy of tnsnames.ora

on your TimesTen machine.

Also check the permission on tnsnames.ora

.

If you are running TimesTen on a UNIX system, check that the ORACLE_HOME environment variable points to the correct Oracle installation directory. For example:

ORACLE_HOME=/products/oracle10g

Check the Oracle client and server versions. See

"Incompatible Oracle Server and

Client versions" on page 4-5.

Unable to resolve connect identifier

You may receive ORA-12154 "TNS:could not resolve the connect identifier specified" when you try to connect to a a data store.

4-4

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Unable to validate Oracle username and password

This can occur when you are trying to use IMDB Cache and Oracle on the same machine and the TNS_ADMIN environment variable does not point to the proper tnsnames.ora

file for Oracle. For example, you may have several instances of the

Oracle Database running on a laptop.

In a production environment, you typically have TimesTen and Oracle running on different machines. In this case, do not reset the TNS_ADMIN environment variable to point to a tnsnames.ora

file on the machine where TimesTen is running. The Oracle client uses the TNS_ADMIN setting to resolve the connection, but the TimesTen main daemon, the cache agent, the Web server, and the replication agent are unaware of the

TNS_ADMIN setting. IMDB Cache cannot operate properly when the Oracle client and TimesTen use different tnsnames.ora

files.

On Windows, set the TNS_ADMIN environment variable as follows:

1.

Right-click My Computer and choose Properties.

2.

3.

On the Advanced tab, choose Environment Variables.

Add or edit TNS_ADMIN as a system environment variable so that it points to the directory that contains the tnsnames.ora

file that you wish to use. You can include other tnsnames.ora

files with the INAME command inside the tnsnames.ora

file.

Incompatible Oracle Server and Client versions

If you receive connection timeout errors such as ORA-12170 or ORA-12535, or if you receive ORA-03134 (server version not supported), verify that you are using an Oracle client and Oracle server whose versions are compatible.

Metalink Documentation Note 207303.1, "Client/Server/Interoperability Support

Between Different Oracle Versions", lists the client/server combinations supported by

Oracle.

See "Oracle In-Memory Database Cache" in the

Oracle TimesTen In-Memory Database

Installation Guide

for information about Oracle clients and servers supported for use with TimesTen. Also check Oracle and TimesTen release notes for known problems with client/server versions.

Unable to validate Oracle username and password

This section describes some of the problems you might encounter when using the

Oracle username and password.

Possible cause

The library environment variable is not set correctly

Oracle processes not running

See...

"Check library path environment variable" on page 4-6.

"Check status of TNS listener and Oracle Server" on page 4-6.

"Check Oracle privileges" on page 4-6.

User does not have the correct Oracle privileges

Incorrectly configured DSN

Problems with cache administration user ID or password

Inconsistent user and system environments

"Check DSN definition" on page 4-6.

"Set the cache administration user id and password" on page 4-7.

"Check user and system environment" on page 4-7.

Troubleshooting Oracle In-Memory Database Cache

4-5

Unable to validate Oracle username and password

Possible cause

Dynamic libraries not loading

See...

"Verify the loaded dynamic libraries" on page 4-7.

Check library path environment variable

Check the library path environment variable on your platform.

On this platform...

UNIX except HP-UX

HP-UX

Windows

Check this variable...

LD_LIBRARY_PATH

On 64-bit platforms, LD_LIBRARY_PATH64 takes precedence over LD_LIBRARY_PATH. Make sure that the library path is specified in LD_LIBRARY_PATH64.

SHLIB_PATH

PATH

The library path environment variable must include the following information:

TimesTen and platform bit combination

64-bit TimesTen or 32-bit TimesTen on 32-bit platform

32-bit TimesTen on 64-bit platform

Setting

$ORACLE_HOME/LIB and $ORACLE_

HOME/NETWORK/LIB

$ORACLE_HOME/LIB32 and $ORACLE_

HOME/NETWORK/LIB32

Check status of TNS listener and Oracle Server

Try to connect to the Oracle database by using SQL*Plus or use Oracle Enterprise

Manager to verify the status.

Check Oracle privileges

From an Oracle SQL*Plus command prompt, list the current Oracle privileges granted to you by entering:

SELECT * FROM SESSION_ROLES;

SELECT * FROM SESSION_PRIVS;

Compare the privileges listed against the required privileges for the various IMDB

Cache operations that are specified in "Grant privileges to Oracle users" in the

Oracle

In-Memory Database Cache User's Guide

. Contact your Oracle Administrator if you require additional privileges.

Check DSN definition

Confirm you have correctly set the DSN attributes as described in "DSN for a

TimesTen database that caches data from an Oracle database" in the

Oracle

In-Memory Database Cache User's Guide

.

Confirm that the DSN definition for IMDB Cache is a system DSN.

Confirm that the DSN for IMDB Cache is defined only once.

4-6

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Unable to validate Oracle username and password

Confirm Oracle user name and password. Use SQLPlus and connect to Oracle using the same OracleID and OraclePWD used in your DSN definition to confirm they are correct.

Reboot TimesTen machine

If the Oracle client was installed and the machine has not been restarted, then the

TimesTen daemon is still running under the "old" environment before the Oracle client install. Reboot your machine so the TimesTen can start under the "new" environment.

Set the cache administration user id and password

From a ttIsql

session, connect to the data store and enter the following:

Command> call ttCacheUidPwdSet('scott','tiger');

If it returns an error, then check the Oracle ID, the cache administration user ID and cache administration password. Also check whether the Oracle instance is running.

Check user and system environment

Test to see if the problem is due to differences in user and system environment. This procedure requires two session windows (Command Prompt windows in Windows or shell windows in UNIX).

1.

2.

Stop the TimesTen daemon.

In one session window, start the Timesten daemon as a regular user.

On Windows:

% install_dir/srv/ttsrv1121.exe -d -verbose

On UNIX:

%

install_dir

/srv/timestend -d verbose

3.

4.

6.

7.

5.

Some messages will flash by, and then it goes into a wait state.

In another session window, try to restart the cache agent.

If Step 3 succeeds, then use Ctrl-C on Windows or the

kill

command on UNIX to

stop the TimesTen daemon you started for the other session in Step 2.

Compare the user environment and system environment. For example, do both user and system see the same copy of oci.dll

? Are there any differences in the path name to the oci.dll

library between the user and system environments?

If you detect differences, make the necessary modifications.

Reboot the system and restart the TimesTen daemon.

Verify the loaded dynamic libraries

If you are running on a Windows system with Visual C++ installed, verify the loaded dynamic libraries. This works only if you can start the cache agent without autorefresh:

1.

2.

Make sure TimesTen is started.

Start the cache agent without autorefresh.

Command> call ttCacheStart;

Troubleshooting Oracle In-Memory Database Cache

4-7

OCI initialization failed

Command> create cache group cg1 from t1(c1 int not null primary key);

3.

4.

Open the Windows Task Manager, find process ttora1121.exe

and highlight it.

Right-click on it and select Debug. This brings you into Visual C++ and you

should see the loaded DLL in the debug window, as described in "Unable to resolve Oracle Service Name" on page 4-4.

Load the cache group to force an cache connection from the cache agent:

Command> load cache group cg1 commit every 100 rows;

5.

Compare the loaded DLL in your debug window with the partial list shown in

Example 4–1

.

Example 4–1 List of loaded dlls

This partial list was created with the Oracle client.

Loaded 'E:\TimesTen\tt1121_32\bin\timestenorad1121.exe', no matching symbolic information found.

Loaded 'C:\WINDOWS\SYSTEM32\ntdll.dll', no matching symbolic information found.

Loaded 'C:\WINDOWS\SYSTEM32\kernel32.dll', no matching symbolic information found.

Loaded 'E:\TimesTen\tt1121_32\bin\tten1121.dll', no matching symbolic information found.

Loaded 'E:\TimesTen\tt1121_32\bin\ttcommon1121.dll', no matching symbolic information found.

Loaded 'C:\WINDOWS\SYSTEM32\wsock32.dll', no matching symbolic information found.

Loaded 'C:\WINDOWS\SYSTEM32\ws2_32.dll', no matching symbolic information found.

Loaded 'C:\WINDOWS\SYSTEM32\msvcrt.dll', no matching symbolic information found.

Loaded 'C:\WINDOWS\SYSTEM32\ws2help.dll', no matching symbolic information found.

Loaded 'C:\WINDOWS\SYSTEM32\advapi32.dll', no matching symbolic information found.

Loaded 'C:\WINDOWS\SYSTEM32\rpcrt4.dll', no matching symbolic information found.

...

OCI initialization failed

Error 5105, "OCI initialization failed," may occur when an operation requires contact with the Oracle database. For example, the error might occur in the following situations:

Starting the cache agent

Setting the cache administration user ID or password

Entering a SQL statement in TimesTen when autocommit=0 and PassThrough=3

Error 5105 contains additional information about its cause:

OCI is unable to find an Oracle library. See

"Check library path environment variable" on page 4-6 and check the permissions on the library specified in the

error message.

ORACLE_HOME is invalid. See "Check ORACLE_HOME environment variable" on page 4-3.

NLS environment variables are set in the environment where the TimesTen application is running, even though TimesTen is not using the NLS environment variables. Unset the NLS environment variables and restart the TimesTen daemon, the cache agent, and the replication agent.

4-8

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

DDL operations on cached Oracle tables may cause cache group operations to fail

Unsupported data type mapping

When you try to create a cache group, you may receive the following error:

5115: Unsupported type mapping for column

name

For example, table

tab

on Oracle can be described as follows:

COL1 NUMBER(38) NOT NULL

COL2 NUMBER(38)

Try to create the cache group as follows:

CREATE CACHE GROUP cg FROM tab(col1 CHAR(10) NOT NULL PRIMARY KEY);

Error 5119 is displayed and the cache group is not created because the statement attempts to map a column of NUMBER data type to a column of CHAR data type.

See "Data type mappings allowed for key columns" in the

Oracle In-Memory Database

Cache User's Guide

.

Null constraint does not match Oracle

When you try to create a cache group, you may receive the following warning:

Warning 5119: Column

name

has different nullability setting in Oracle

For example, table tab

on Oracle can be described as follows:

COL1 NUMBER(38) NOT NULL

COL2 NUMBER(38)

Try to create the cache group as follows:

CREATE CACHE GROUP cg

FROM tab(col1 INTEGER NOT NULL PRIMARY KEY, col2 INTEGER NOT NULL);

Warning 5119 is displayed because col2

on Oracle does not have a NULL constraint, but col2

in the cache group is defined as NOT NULL.

DDL operations on cached Oracle tables may cause cache group operations to fail

DDL operations that are performed on an Oracle table that is being cached in

TimesTen may cause a failure on the cache group. For example, the user drops a column on the Oracle table that is being cached in TimesTen. When the cache group is propagated or flushed, TimesTen will update the column that no longer exists in the

Oracle table. When the cache group loads or refreshes, then TimesTen attempts to retrieve data from the column that has been dropped.

The following cache group operations may fail:

Autorefresh does not occur.

AWT cache group operations are not propagated or refreshed to/from Oracle.

Cache group load or propagate fails.

If you suspect the cache group operations are not working properly because of a DDL operation on the Oracle base table, then use DDL tracking to diagnose the issue. DDL tracking saves the change history for all the cached Oracle tables. The SQL statement

Troubleshooting Oracle In-Memory Database Cache

4-9

Changes not visible after updating object in cache group and when it was executed are each written to a TimesTen table in the cache administrator user schema on Oracle.

For more information on how to create the DDL tracking objects and how to enable

DDL tracking for the base table within Oracle, see "Monitoring DDL operations on

Oracle tables" in the

Oracle In-Memory Database Cache User's Guide

. For details on the built-in procedures used for initializing and enabling DDL tracking, see the

Oracle

TimesTen In-Memory Database Reference

.

Changes not visible after updating object in cache group

If you modify an object in a cache group and then the changes do not appear on a subsequent SQL statement, then one of the following may have occurred:

The object was dropped from the Oracle database or was somehow damaged.

The Oracle database was restored or recovered to a time before the object was created.

The Oracle database was down.

The user modified the

OracleNetServiceName

DSN or connection attribute after creating the cache group, which points to an Oracle database other than the one that the cache group was created upon.

For example, if the user creates an AWT cache group. Then, the user added rows to a table. When the user performs a

SELECT * FROM

the table, the rows did not appear.

The ttmesg.log

error file does not display an error that Oracle is not available.

Instead, it displays the following messages:

12:09:02.10 Err : REP: 29934: CACHE1:meta.c(904): TT5221: TT5221: Oracle syntax error in OCIStmtExecute(): ORA-00942: table or view does not exist rc = -1 -- file "bdbStmt.c", lineno 1535, procedure "getOraOutTypesNLengths()"

12:09:02.27 Err : REP: 29934: CACHE1:receiver.c(1978): TT5250: Awt Initialization

Failure. Could not compile meta data sql.

12:09:02.27 Warn: REP: 29934: CACHE1:transmitter.c(6505): TT16060: Failed to read data from the network. select() timed out

To recover, perform the following:

1.

Stop all updates to the cache group.

2.

3.

If you are using an AWT cache group, then flush the cache group.

Recreate the cache group with the drop and create.

Loading or refreshing fails

If the LOAD CACHE GROUP or REFRESH CACHE GROUP statement fails when you specify COMMIT EVERY

n

ROWS and

n

is greater than 0, the contents of the target cache group could be in an inconsistent state. Some cache instances may be partially loaded.

Unload the cache group and then load it again. In some situations, it may be easier to drop and re-create the cache group.

Monitoring autorefresh cache groups

This section includes the following topics:

Using the ttCacheAutorefreshStatsGet procedure

4-10

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Monitoring autorefresh cache groups

Displaying information from the change log tables

Understanding messages about autorefresh in the support log

Diagnosing autorefresh failure

Diagnosing autorefresh performance problems

Using SNMP traps for alerts about autorefresh problems

Using the ttCacheAutorefreshStatsGet procedure

The ttCacheAutorefreshStatsGet

procedure returns information about the last ten autorefresh operations on a specified cache group.

The ttCacheAutorefreshStatsGet

procedure returns information only when the cache agent is running and the autorefresh state is ON or PAUSED. All of the return fields are set to 0 when the cache agent is restarted or the autorefresh state is changed to OFF.

Example 4–2 Calling ttCacheAutorefreshStatsGet

This example uses testcache

, which is a READONLY cache group with one table and an incremental autorefresh interval of 10 seconds.

Command> call ttcacheautorefreshstatsget('user1','testcache');

< 1164260, 2007-07-23 15:43:52.000000, 850280, 44, 0, 75464, 528255, 75464, 310, 110, 6800,

1890912, 12439795, 1890912, 160020, InProgress >

< 1164260, 2007-07-23 15:43:33.000000, 831700, 43, 13550, 108544, 759808, 108544, 1030, 230,

12290, 1815448, 11911540, 1815448, 160020, Complete >

< 1164260, 2007-07-23 15:43:12.000000, 810230, 42, 17040, 115712, 809984, 115712, 610, 330,

16090, 1706904, 11151732, 1706904, 146470, Complete >

< 1164260, 2007-07-23 15:42:52.000000, 790190, 41, 14300, 94208, 659456, 94208,560, 320,

13410, 1591192, 10341748, 1591192, 129430, Complete >

< 1164260, 2007-07-23 15:42:32.000000, 770180, 40, 12080, 99328, 695296, 99328,450, 290,

11340, 1496984, 9682292, 1496984, 115130, Complete >

< 1164260, 2007-07-23 15:42:12.000000, 750130, 39, 10380, 86016, 598368, 86016,430, 230,

9720, 1397656, 8986996, 1397656, 103050, Complete >

< 1164260, 2007-07-23 15:41:52.000000, 730130, 38, 13530, 112640, 700768, 112640, 530, 220,

12780, 1311640, 8388628, 1311640, 92670, Complete >

< 1164260, 2007-07-23 15:41:32.000000, 710120, 37, 9370, 56320, 326810, 56320, 310, 160,

8900, 1199000, 7687860, 1199000, 79140, Complete >

< 1164260, 2007-07-23 15:41:22.000000, 700120, 36, 2120, 10240, 50330, 10240, 50, 200, 1870,

1142680, 7361050, 1142680, 69770, Complete >

< 1164260, 2007-07-23 15:41:12.000000, 690110, 35, 0, 0, 0, 0, 0, 0, 0, 1132440, 7310720,

1132440, 67650, Complete >

10 rows found.

Table 4–1 describes the results from the first row of output.

Table 4–1

Result

1164260

2007-07-23

startTimestamp

15:43:52.0

00000

ttCacheAutorefreshStatsGet results from last autorefresh operation

Field name

cgId

Description

Cache group ID

Timestamp when autorefresh started for this interval

850280

cacheAgentUpTime

Number of cache agent clock ticks in milliseconds at the time the autorefresh transaction started for this interval. This value is cumulative and is reset when the cache agent process starts.

Troubleshooting Oracle In-Memory Database Cache

4-11

Monitoring autorefresh cache groups

Table 4–1 (Cont.) ttCacheAutorefreshStatsGet results from last autorefresh operation

Result

44

0

75464

528255

75464

310

110

6800

1890912

12439795

1890912

160020

InProgress

Field name

autorefNumber autorefDuration autorefNumRows numOracleBytes autorefNumRootTblRows autorefQueryExecDuration autorefQueryFetchDuration autorefTtApplyDuration totalNumRows totalNumOracleBytes totalNumRootTblRows totalDuration autorefreshStatus

Description

Autorefresh number

The number of milliseconds spent in this autorefresh operation. It is zero because the operations is in progress.

The number of rows autorefreshed in this autorefresh operation. This would include all rows in the root table and child tables if the cache group had child tables.

Note

: This information is not provided for full autorefresh.

The number of bytes transferred from Oracle in this autorefresh operation.

Note

: This information is not provided for full autorefresh.

The number of root table rows autorefreshed in this autorefresh operation.

The duration in milliseconds for the autorefresh query to execute on Oracle.

Note

: This information is not provided for full autorefresh.

The duration in milliseconds for the autorefresh query to fetch rows from Oracle.

Note

: This information is not provided for full autorefresh.

The duration in milliseconds for TimesTen to apply the updated rows to the cache group.

Note

: This information is not provided for full autorefresh.

The total number of rows autorefreshed since the cache agent started.

Note

: This information is not provided for full autorefresh.

The total number of bytes transferred from

Oracle since the cache agent started.

Note

: This information is not provided for full autorefresh.

Note

: This information is not provided for full autorefresh.

Note

: This information is not provided for full autorefresh.

The total number of root table rows autorefreshed since the cache agent started.

The total autorefresh duration in milliseconds since the cache agent started.

Status. The status can also be

Complete

or

Failed

.

4-12

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Monitoring autorefresh cache groups

Note that the total number of autorefreshed rows (1890912) is the same as the total number of autorefreshed root table rows in this example because there are no child tables.

The number of autorefreshed rows in TimesTen does not necessarily reflect the number of rows updated on Oracle. The Oracle updates may be applied in TimesTen more than once, or multiple Oracle updates on the same row may be applied as one update in TimesTen.

Displaying information from the change log tables

TimesTen provides a SQL script that gathers information from the change log tables that exist on the Oracle database for autorefresh cache groups. See "Oracle objects used to manage a caching environment" in the

Oracle In-Memory Database Cache User's Guide

for more information about change log tables.

The script displays the following information for each cached table:

****************************

* Host name: my-pc

* Timesten datastore name: c:\data\tt1121

* Cache table name: USER1.TESTCACHE

* Change log table name: tt_03_55555_L

* number of rows in change log table: 100000

* Maximum logseq on the change log table: 38

* Timesten has autorefreshed updates up to logseq: 38

* Number of updates waiting to be autorefreshed: 0

* Number of updates that has not been marked with a valid logseq: 0

****************************

The log sequence number ( logseq

) acts as a marker for the autorefresh operation.

Run the script as the cache administration user on the Oracle database using

SQL*Plus. If you run the script as a different user, it reports that the change log tables do not exist.

The script is in the following location:

install_dir

/oraclescripts/cacheInfo.sql

Understanding messages about autorefresh in the support log

The support log contains messages that show the progress of autorefresh. For example, testcache

is a READONLY cache group with an autorefresh interval of 10 seconds (10,000 milliseconds).

The support log shows when autorefresh starts:

15:43:33.96 Info: ORA: 5264: ora-5264-5676-refresh03918: Starting autorefresh number 43 for interval 10000ms

The message includes the following information:

Timestamp (

15:43:33.96

)

Cache agent process ID (

5264

)

Thread ID (

5676

)

The thread ID is important because autorefresh numbers are unique only for a specific interval. Always check both the thread ID and the autorefresh number when you are tracking a specific autorefresh operation.

Troubleshooting Oracle In-Memory Database Cache

4-13

Monitoring autorefresh cache groups

The support log also contains a longer message that reports information similar to the ttCacheAutorefreshStatsGet

procedure. 108544 rows were updated in this autorefresh interval, and 1815448 rows have been updated since the cache agent was started. Note that the total number of rows and the total number of root table rows are the same in this message because there is only one table in the cache group.

Number refers to the autorefresh number. All times are expressed in milliseconds.

15:43:51.81 Info: ORA: 5264: ora-5264-5676-refresh04387: Cache agent refreshed cache group USER1.TESTCACHE: Number - 43, Duration - 13550, NumRows - 108544,

NumRootTblRows - 108544, NumOracleBytes - 759808, queryExecDuration - 230, queryFetchDuration - 1030, ttApplyDuration - 12290, totalNumRows - 1815448, totalNumRootTblRows - 1815448, totalNumOracleBytes - 11911540, totalDuration -

160020

Additional messages show that the autorefresh operation completes successfully:

15:43:51.81 Info: ORA: 5264: ora-5264-5676-refresh04449: Autorefresh number 43 finished for interval 10000ms successfully

15:43:51.81 Info: ORA: 5264: ora-5264-5676-fresher01619: Autorefresh number 43 succeeded for interval 10000 milliseconds

Inspect the timestamps to determine whether autorefresh is progressing as expected.

See "Managing TimesTen daemon options" in the

Oracle TimesTen In-Memory Database

Operations Guide

for information about setting the support log location.

Diagnosing autorefresh failure

If ttCacheAutorefreshStatsGet

shows that the status of an autorefresh operation is

Failed

, check the support log for messages related to the autorefresh operation with number the number shown in the ttCacheAutorefreshStatsGet

output. Look for errors that occurred after the autorefresh operation started.

Example 4–3 ttCacheAutorefreshStatsGet output shows autorefresh failure

This row of output from ttCacheAutorefreshStatsGet

shows a failed autorefresh operation.

< 1164260, 2007-08-01 14:56:36.000000, 959350,

9

, 0, 0, 0, 0, 0, 0, 0, 1, 7, 1,

50, Failed >

The autorefresh number is 9.

The support log shows the start message for autorefresh number 9:

14:56:36.10 Info: ORA: 5988: ora-5988-

4724

-refresh03926: Starting autorefresh number

9

for interval 15000ms

The thread ID for autorefresh number 9 is 4724. Look for error messages with this thread ID.

The following messages appear in the support log:

14:56:36.10 Info: ORA: 5988: ora-5988-4724-refresh03953: Autorefresh thread for interval 15000ms is connected to instance inst1 on host host1. Server handle

231976252

14:56:36.12 Err : ORA: 5988: ora-5988-4724-refresh07567: TimesTen error code:5901, msg The Oracle refresh log table, "USER2"."TT_03_81799_L", for base table, USER2.READTAB2, cannot be found.

14:56:36.12 Info: ORA: 5988: ora-5988-4724-refresh05559: Autorefresh rolled back.

14:56:36.12 Info: ORA: 5988: ora-5988-4724-refresh04458: Autorefresh number 9

4-14

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Optimize Performance for IMDB Cache finished for interval 15000ms with error.

14:56:36.12 Err : ORA: 5988: ora-5988-4724-fresher01606: Autorefresh number 9 failed for cache groups with interval 15000 ms after 10 retries.

The error message for thread ID 4724 shows that the change log table, TT_03_81799_L,

is missing. The introduction to "Autorefresh not refreshing cache at the specified interval" on page 4-16 has a table entry that describes what to do in this situation.

Diagnosing autorefresh performance problems

You can use the ttTraceMon

utility to diagnose autorefresh performance problems.

See

"AUTOREFRESH tracing" on page 1-16.

TimesTen tracing severely impacts application performance and consumes a great deal of disk space if trace output is directed to a file. When you are finished, reset tracing to the default values.

Using SNMP traps for alerts about autorefresh problems

Enable SNMP traps to alert you when autorefresh problems occur.The SNMP traps related to autorefresh include:

■ ttCacheAutoRefQueFullTrap

■ ttCacheIncAutoRefFailedTrap ttCacheValidationErrorTrap

■ ttCacheValidationWarnTrap

■ ttCacheValidationAbortedTrap

See "Diagnostics through SNMP Traps" in the

Oracle TimesTen In-Memory Database

Error Messages and SNMP Traps

.

Optimize Performance for IMDB Cache

The following recommendations optimize performance for the IMDB Cache:

Pin the IMDB Cache meta tables and cache group base tables in the SGA. by executing the "ALTER TABLE

<table_name>

CACHE" statement to indicate to the

Oracle database that these tables should be stored in the keep portion of the SGA buffer cache. Pinning IMDB Cache tables in the SGA increases the probability that any given datablock needed for a IMDB Cache refresh operation will be available in the SGA when the refresh is performed and will not force a disk read. This minimizes physical disk reads executed during TimesTen cache refresh operations. For more information about Oracle buffer cache management, see

"Configuring and Using Memory" in the

Oracle Database Performance Tuning Guide

.

Pin IMDB Cache triggers into the shared pool using the dbms_shared_ pool.keep

procedure. Pinning triggers into the shared pool for applications where updates to the cache group base tables are infrequent keeps the trigger from having to be reloaded and reparsed. This is not necessary for highly volatile tables where the trigger will be executed frequently and will remain in the shared pool under any circumstances.

Enable parallel query. For very large base tables with 10 million rows or more, consider using the Oracle database parallel query facility. The primary join query between the log table and the base table is the kind of query which the Oracle database parallel query is designed to handle. When parallel processing is

Troubleshooting Oracle In-Memory Database Cache

4-15

Autorefresh not refreshing cache at the specified interval enabled, the parallel query optimizer generates a query plan that allows the original query to be broken into sections to be worked concurrently by different parallel query slave processes. When using parallel query, users should assign a default degree of parallelism of (2*N) to the cache group base table, where "N" is the number of CPUs on the machine. Then, experiment to understand what level of parallelism works best for their environment. Experiment with different table structures for base tables, as follows:

Use a standard heap table with default degree of parallelism assigned during table creation or by use of the ALTER TABLE PARALLEL command. Build an

N-partition primary key index against the table.

Use an N-way partitioned table structure with partition range key based either on the table primary key or, in the case of a concatenated primary key, the high-order column of the primary key. The number of partitions should be set to the degree of parallelism. Use a local primary key index with the same number of partitions.

Use an N-way hashed partition structure using the primary key as the hash key, a local paritioned primary key index, and both index and table partitions equal to the degree of parallelism. The log table should not be partitioned, as the cardinalities of the log table should never be large enough that a paritioned log table would have any performance benefit. Further, given the continuously increasing value of the log table primary key column, range partitions cannot be used.

Autorefresh not refreshing cache at the specified interval

The following table shows possible causes for autorefresh problems.

Possible cause

Cache agent not started with a cache administration user

Object ID of the base table has changed.

Autorefresh trigger not enabled

What to do

Specify a cache administration user ID and password when starting the cache agent, as described in "Starting the cache agent" in the

Oracle In-Memory Database Cache User's Guide

.

See "Recover and reset autorefresh Oracle objects" on page 4-17.

See "Recover and reset autorefresh Oracle objects" on page 4-17.

See "Recover and reset autorefresh Oracle objects" on page 4-17.

Current log sequence number recorded in the TT_

version

_USER_COUNT table is less than to the maximum log sequence number in the autorefresh log table.

There is no row in the TT_

version

_

USER_COUNT table with

usercount

> 0 for every active incrementally autorefresh table

Change log table is empty.

User count is less than 0 or any TT_

version

anomalies

_USER_COUNT log sequence

See "Recover and reset autorefresh Oracle objects" on page 4-17.

See "Recover and reset autorefresh Oracle objects" on page 4-17.

See "Recover and reset autorefresh Oracle objects" on page 4-17.

4-16

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Autorefresh not refreshing cache at the specified interval

Possible cause

Autorefresh log table, trigger, or sequence associated with a cached table does not exist or is not valid.

TT_

TT_

version

missing.

version

_USER_COUNT table is

If the current log sequence number in the

_USER_COUNT table changes, is different from the bookmark and the associated cached table is not refreshed by the next committed autorefresh.

Resource problem

What to do

Check whether the cache agent was started with the correct cache administration user ID. If the cache administration user ID is correct, follow the

procedure described in "Recover and reset autorefresh Oracle objects" on page 4-17.

Check the user error log for messages about "fatal anomalies". This indicates corrupt or missing

Oracle objects.

Check whether the cache agent was started with the correct cache administration user ID. If the cache administration user ID is correct, follow the procedure in

"Recover and reset autorefresh

Oracle objects" on page 4-17.

Check the user error log for messages about "fatal anomalies". This indicates corrupt or missing

Oracle objects.

Restart the cache agent. If that does not work, follow the procedure in

"Recover and reset autorefresh Oracle objects" on page 4-17.

Restart the cache agent.

Reset autorefresh state

Incremental autorefresh does not work if the TRUNCATE statement is used on an

Oracle base table. If TRUNCATE is used on an Oracle base table, then you must reset autorefresh by using the ALTER CACHE GROUP statement to set the autorefresh state to OFF followed by another ALTER CACHE GROUP to reset the autorefresh state to ON.

Recover and reset autorefresh Oracle objects

If you know or suspect the Oracle objects used by autorefresh are the cause of the problem, use the following procedure to re-create the Oracle objects.

1.

Use ALTER CACHE GROUP to reset the autorefresh state to OFF on all cache groups on all data stores that have the affected cached table:

ALTER CACHE GROUP

cache_group_name

SET AUTOREFRESH STATE OFF;

2.

3.

Shut down all cache agents on all affected data stores.

Check if the user count is zero for each table in the cache group.

On the Oracle database, execute the following statement:

SELECT usercount FROM

autorefresh_id

.tt_

version

_user_count

WHERE tablename ='

owner

.

tablename

';

If the count is not zero, set the count to zero:

UPDATE

autorefresh_id

.tt_

version

_user_count SET usercount = 0

WHERE tablename ='

owner

.

tablename

';

Troubleshooting Oracle In-Memory Database Cache

4-17

Incremental autorefresh not progressing

4.

Start one of the cache agents. The cache agent performs a clean up operation. It displays the following message to the support log after it has completed the cleanup:

Cleanup of the Oracle objects completed

5.

After the cache agent has completed the clean up, use ALTER CACHE GROUP to reset the autorefresh state back to ON:

ALTER CACHE GROUP

cache_group_name

SET AUTOREFRESH STATE ON;

6.

7.

Start all other cache agents.

Use ALTER CACHE GROUP to reset the autorefresh state back to ON for all of the affected cache groups on all data stores.

Incremental autorefresh not progressing

If incremental autorefresh is not progressing, verify that:

Autorefresh state is ON

Cache agent is running

Inspect the support log for the conditions described in the following table:

Table summary is in the first heading cell.

Condition

Oracle server connection errors or warnings

Lock timeout errors or warnings on

TimesTen

What to do

See

"Troubleshooting Client/Server problems" on page 2-6 for information about resolving connection

problems.

This usually occurs because of an open DDL transaction on the cache group. Commit the DDL transaction so that autorefresh can get the necessary locks.

Increase PermSize.

Insufficient permanent data partition errors on TimesTen

Autorefresh Oracle object validations errors or warnings

Cache agent exits unexpectedly.

Core files in main daemon directory

Warnings about incremental autorefresh becoming full refresh

Warnings that autorefresh has not finished for a long time

See

"Recover and reset autorefresh Oracle objects" on page 4-17.

Contact Technical support.

Contact Technical support.

See

"Incremental autorefresh becomes full autorefresh" on page 4-19.

The autorefresh transaction can take a long time if many transactions have occurred since the last autorefresh.

Note

: Cache groups with the same autorefresh interval are autorefreshed in one transaction.

Validate autorefresh Oracle objects

The cache agent automatically verifies that Oracle objects exist and that they are valid so that autorefresh can progress. In normal operation, you should not see object validation errors or warnings in the user error log. If you see object validation errors, contact Technical support.

unless

one of the following conditions has occurred:

4-18

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Incremental autorefresh becomes full autorefresh

The TimesTen data store has been destroyed without using the DROP CACHE

GROUP statement.

A customer application inadvertently modifies the objects directly in the Oracle database.

A DDL operation occurs on the base table on the Oracle database. This disables the trigger that controls autorefresh operations.

The cache group needs to be re-created if one of the preceding conditions has occurred.

Incremental autorefresh becomes full autorefresh

Incremental autorefresh can become full autorefresh if the cache administration user tablespace becomes full.

This section includes the following topics:

Detecting when incremental autorefresh becomes full

Understanding the cache administration user tablespace

Diagnosing a full cache administration user tablespace

Monitoring the usage of the cache administration user's tablespace

Considerations when the cache administration user's tablespace is full

Detecting when incremental autorefresh becomes full

You can detect when incremental autorefresh becomes full refresh by several methods:

Check for messages in the support log that indicate full autorefresh operations are occurring. For example:

2007-08-08 08:06:51.35 Warn: ORA: 22119: ora-22119-0015-refresh05652: A full autorefresh will be performed for Incremental autorefresh table USER1.READTAB because change log table T_03_55555_L on Oracle has been truncated.

Use the ttCacheAutorefreshStatsGet

procedure.

If autorefresh is occurring.

InProgress

for longer than usual, full autorefresh may be

If a much larger number of rows (

autoRefNumRows

) was autorefreshed than usual, full autorefresh may have occurred.

Check the support log for messages about full autorefresh.

If SNMP traps are enabled, the ttCacheRecoveryAutorefreshTrap

SNMP trap indicates a full autorefresh.

Understanding the cache administration user tablespace

TimesTen strongly recommends creating a separate tablespace for the cache administration user. This tablespace is used as the cache administration user's default tablespace. The tablespace contains autorefresh triggers for each Oracle table, change log tables for each Oracle table, and other objects that TimesTen needs for each cache administration user. If you do not specify a separate tablespace, then these objects are placed in the Oracle system tablespace.

Troubleshooting Oracle In-Memory Database Cache

4-19

Incremental autorefresh becomes full autorefresh

Specify the tablespace when you create the cache administration user on Oracle. You can also specify the tablespace after user creation with the DEFAULT TABLESPACE clause of the Oracle ALTER USER statement.

Change log tables for each of the cached Oracle tables reside in the cache administration user tablespace. For each update on an Oracle table, one row (a change log record) is inserted into the change log table for that Oracle table. The size of a change log record in bytes is as follows: size of change log record = size of primary key on Oracle table + 250

The number of records in a change log table depends on the update rate on the Oracle table and on the autorefresh interval on TimesTen. Every 20 seconds, TimesTen removes change log records that have been applied to all data stores that cache the associated Oracle table.

When change logs are removed, a message similar to the following is displayed in the support log:

16:32:26.73 Info: ORA: 5652: ora-5652-4756-ogTblGC01036: Garbage collector deleted 1 rows from TT_03_383270_L where logseq < 1

There are options on how to manage what happens when the cache administration

user tablespace is filled. See "Considerations when the cache administration user's tablespace is full" on page 4-21 for more information.

Diagnosing a full cache administration user tablespace

Check for the following conditions if the cache administration user tablespace is full:

Is the autorefresh state set to PAUSED? Change log records accumulate when the state is PAUSED.

Has the cache group been created but not loaded? The default autorefresh state for cache group creation is PAUSED.

Is a cache group being created or is a data store being duplicated? Both of these operations temporarily stop clean-up operations on the change log table.

Are the cache agents on all TimesTen data stores running? If a cache agent is not running, change log records accumulate.

Has a data store been abandoned without dropping autorefresh cache groups in the data store? Abandoned data stores result from scenarios such as the following:

The data store is destroyed by ttDestroy -force

.

The application connected to the data store with the attribute set to 1, but the cache groups that were in the old data store are not re-created.

Overwrite

connection

If the data store still exists, connect to the abandoned data store and drop the cache group.

Use the cacheInfo.sql

script to find out how large the change log tables are for each cached Oracle table. Use the output to verify that the data stores are still in use.

See

"Displaying information from the change log tables" on page 4-13.

If the data stores are still in use, verify that the cache agents are running.

Compare the autorefresh progress on TimesTen to the maximum log sequence number on the change log table. If TimesTen is behind, then call the ttCacheAutorefreshStatsGet

procedure to see whether the autorefresh

4-20

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Incremental autorefresh becomes full autorefresh operations are successful. See

"Using the ttCacheAutorefreshStatsGet procedure" on page 4-11.

If the status is

InProgress

longer than seems reasonable, see "Poor autorefresh performance" on page 4-22.

You may need to decrease the autorefresh interval or increase the size of the cache administration user tablespace.

There are options on how to manage what happens when the cache administration

user tablespace is filled. See "Considerations when the cache administration user's tablespace is full" on page 4-21 for more information.

Monitoring the usage of the cache administration user's tablespace

To monitor the cache administration user tablespace, you can use either Oracle

Enterprise Manager alerts or set the TimesTen tablespace threshold parameter.

The cache agent can be configured to periodically monitor the tablespace usage and issue a warning when it exceeds a specified threshold. Set the tablespace threshold percentage with the

TblspaceThreshold

parameter of the ttCacheConfig

built-in procedure. For example, if you set the

TblspaceThreshold

parameter to 80, then a warning is issued when more than 80% of the tablespace is used.

If the threshold is set to zero, then no warning is issued. This is the default.

If the threshold is set between 1 and 99, a warning is issued when the tablespace threshold exceeds that number.

If the threshold is set to 100, then a warning is issued when the tablespace is full.

For example, to configure for a warning to be issued if the tablespace exceeds 80%, execute the following: call ttCacheConfig(’TblspaceThreshold’,,,’80’);

For full details of the ttCacheConfig

built-in procedure, see the "ttCacheConfig" section in the

Oracle TimesTen In-Memory Database Reference

.

Considerations when the cache administration user's tablespace is full

With Oracle tables that are cached in a TimesTen database, you can configure them to use incremental automatic refresh. For these tables, you can specify which one of the following is to occur when the cache administration user's tablespace is full:

The application performing the DML is to fail. This is the default.

The tablespace full recovery is set to none. The application receives an "Out of

Tablespace" error from Oracle when the tablespace is full. At that point, the application will need to rollback the transaction.

Setting the tablespace full recovery to none is configured when you set the

Param parameter to

TblSpaceFullRecovery

and the

Value

parameter to

None

with the ttCacheConfig

built-in procedure. For example, the following configures

Param

to

TblSpaceFullRecovery

and

Value

to

None

for the employees

table that is owned by terry

: call ttCacheConfig('TblSpaceFullRecovery',’terry’, ’employees’,'None');

Truncate the change log table to free up space and cause a full autorefresh.

When the cache administration user's tablespace is full, any application that is executing DML statements on the autorefresh cached Oracle tables continues to

Troubleshooting Oracle In-Memory Database Cache

4-21

Poor autorefresh performance execute. A trigger executes to free up space for new change log records by deleting existing change log records. This can result in a full automatic refresh on cache groups that have the incremental automatic refresh mode configured. However, if the Oracle table is not configured for incremental automatic refresh, then no trigger executes.

To set the operation to allow the application to continue and cause an autorefresh, set the

Param

parameter to

TblSpaceFullRecovery

and the

Value

parameter to

Reload

with the ttCacheConfig

procedure. The user will see stale data until the full autorefresh is complete.

However, even if the user sets the cache configuration parameter

TblSpaceFullRecovery

with the value of

Reload

, the tablespace may not be able to be emptied enough to handle the case of a growing index. Deleting rows from the change log table may not free up enough space for the index that is on the change log table. If the index is growing so fast that it uses all the tablespace to the point where purging the change log tables does not help, then the user's application may receive the following error:

ORA-01654: unable to extend index <index> by 128 in tablespace <tblspace>

For full details of the ttCacheConfig

built-in procedure, see the "ttCacheConfig" section in the

Oracle TimesTen In-Memory Database Reference

.

Poor autorefresh performance

Poor autorefresh performance is usually the result of large autorefresh operations. Use the ttCacheAutorefreshStatsGet

procedure to check the autorefresh duration and observe whether the status remains

InProgress

for a long time.

Factors that can cause large autorefresh operations include:

Incremental autorefresh becomes full autorefresh

Unresponsive or dead TimesTen database degrades autorefresh performance

Excessive deadlocks, buffer busy and row lock waits during autorefresh cache group refresh

Abnormally large log and base tables degrade autorefresh performance

Performance degrades when autorefresh interval is small

Large autorefresh interval

Large number of cache groups with the same interval

High rate of changes to the Oracle tables

The number of generations of child tables in a cache group

The number of rows in the cached Oracle tables

The size of the rows in the cached Oracle tables

Enable an AUTOREFRESH trace to diagnose autorefresh performance problems. See

"AUTOREFRESH tracing" on page 1-16.

4-22

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Poor autorefresh performance

Unresponsive or dead TimesTen database degrades autorefresh performance

Note:

Automatic recovery for TimesTen cache groups only applies to read-only and user managed cache groups that use the

AUTOREFRESH cache group attribute. In this section, all references to autorefresh cache groups are read-only and user managed cache groups that use the AUTOREFRESH cache group attribute.

If any TimesTen databases containing autorefresh cache groups are destroyed or no longer in use, TimesTen continues to track autorefresh changes to the Oracle tables for the TimesTen database for which the cache agent is not running. This causes automatic refresh to cache groups in active TimesTen databases to slow down.

The cache agent is responsible for detecting if a database is unresponsive or no longer in use. You can specify if and how a dead TimesTen database is to be recovered.

However, you cannot recover a TimesTen database if all of the Oracle objects have been removed.

The following sections describe how you can avoid a degraded autorefresh performance for inactive TimesTen databases:

Setting cached TimesTen database timeout

Configuring recovery method for certain cache groups

Setting cached TimesTen database timeout

You can instruct TimesTen to mark the database as dead and no longer accepting updates if the cache agent has not communicated with the Oracle server within a specific timeout period.

Set the timeout for the TimesTen database and the recovery method for each autorefresh cache group with the

AgentTimeOut

parameter in the ttCacheConfig built-in procedure. The timeout value applies to the all TimesTen databases that use the same cache administration user. You should set the timeout value greater than the time necessary to load the TimesTen database into memory on first connect and start the cache agent. Otherwise, the TimesTen database could be incorrectly marked as dead. For any planned maintenance for the TimesTen instance, you could temporarily set the

AgentTimeOut

value to zero to disable the timeout. For full details of the ttCacheConfig

built-in procedure, see the "ttCacheConfig" section in the

Oracle

TimesTen In-Memory Database Reference

.

For example, the following sets the timeout value for the TimesTen database to 6000 seconds or 100 minutes. If the cache agent does not contact the Oracle server within a

100-minute period, then the TimesTen database is marked as dead.

ttIsql> call ttCacheConfig('AgentTimeOut',,,'6000');

Configuring recovery method for certain cache groups

You can recover a TimesTen database and autorefresh cache groups if they are not synchronizing with the Oracle database. If there is no synchronization, then updates on the Oracle tables are not automatically refreshed to the corresponding TimesTen cache tables.

You can configure the

DeadDbRecovery

parameter of the ttCacheConfig

built-in procedure to specify how to recover the synchronization for the TimesTen database and all autorefresh cache groups. The setting for

DeadDbRecovery

applies to all

TimesTen databases that use the same cache administrator user. Set the

Troubleshooting Oracle In-Memory Database Cache

4-23

Poor autorefresh performance

DeadDbRecovery

parameter to

Normal

,

Manual

or

None

to describe how TimesTen is to recover the database and all autorefresh cache groups. The

DeadDbRecovery setting applies to all TimesTen databases that use the same cache administration user.

While TimesTen is recovering the database and its autorefresh cache groups, there is an autorefresh status for the TimesTen database and the autorefresh cache groups that describes the recovery status for each of these entities. The TimesTen database can have an automatic refresh status of Alive, Dead or Recovering. The autorefresh cache groups can have an automatic refresh status of OK, Dead or Recovering. The

TimesTen database status changes are linked to changes in the status for the autorefresh cache groups, as follows:

If the recovery method is set to Normal, then when TimesTen starts a full automatic refresh on an autorefresh cache group, the cache group’s status is set to

Recovering and the database’s status is also set to Recovering.

The TimesTen database’s status is only set to Alive when all of the autorefresh cache groups have either been recovered to OK or have been dropped.

When the database status is set to Dead, then all of its autorefresh cache groups are also set to Dead.

Note:

You can determine the autorefresh status of the TimesTen database and autorefresh cache groups with the ttCacheDbCgStatus

built-in procedure, which is described in the

"ttCacheDbCgStatus" section in the

Oracle TimesTen In-Memory

Database Reference

.

When communication between the cache agent and the Oracle server is re-established,

TimesTen determines how to recover the autorefresh cache groups. TimesTen follows the recovery method you configured in the

DeadDbRecovery

parameter in the ttCacheConfig

built-in procedure. This parameter can be set to one of the following:

Normal

: This is the default. The autorefresh cache groups will each be recovered with a full automatic refresh. After the first full refresh, the cache group is recovered and will incrementally perform autorefresh.

The autorefresh cache groups within the same automatic refresh interval will be transactionally consistent. Because it is a full refresh, it is not as performant as an incremental refresh.

The autorefresh sets the status to Recovering. When the full autorefresh is completed successfully, the autorefresh cache group status is set to OK.

Manual

: You must manually refresh an autorefresh cache group to recover it, or unload it if the cache group is dynamic.

None

: The autorefresh cache group will never be recovered by a TimesTen autorefresh. Drop and recreate the cache group to recover it.

The database status changes as the first autorefresh cache group status changes. If there is at least one cache group that is in the process of recovery, then the database status is set to Recovering. Once all cache groups have been recovered, the status of the TimesTen database is marked as Alive.

The following example sets the

DeadDbRecovery

parameter to

Normal

for all autorefresh cache groups. The dead TimesTen database will be recovered when all of its autorefresh cache groups have each been recovered with a full automatic refresh. ttIsql> call ttCacheConfig('DeadDbRecovery',,,'Normal');

4-24

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Poor autorefresh performance

When TimesTen databases participating in an active standby pair replication scheme contains cache groups, if the autorefresh status of the active master database is Dead and the autorefresh status of the standby master database is Alive, the standby master does not automatically assume the role of the active master. The recovery requires that you manually ensure that the cache and replication agents are executing. The specifics for each situation is as follows:

Table 4–2 Recovery for cache groups involved in active standby replication pair

DeadDbRecovery

Setting

Normal

Normal

Normal

Manual

Manual

Manual

Active

Master

Alive

Dead

Dead

Alive

Dead

Dead

Standby

Master

Dead

Alive

Dead

Dead

Alive

Dead

Resulting

Behavior

Make sure that the cache and replication agents are executing on the standby master. Once the cache agent can connect to the

Oracle Database, then the status of all autorefresh cache groups is set to Recovering. This sets the database to Recovering. Once a single cache group has received enough data to resume autorefresh, the status is set to OK. After all cache group are set to OK, the database is set to Alive.

Alternatively, you can fail the standby master and rollout a new standby master.

Make sure that the cache and replication agents are executing on the active master. Once the cache agent can connect to the

Oracle Database, then the status of all autorefresh cache groups is set to Recovering. This sets the database to Recovering. Once a single cache group has received enough data to resume autorefresh, the status is set to OK. After all cache group are set to OK, the database is set to Alive.

Alternatively, you can fail the active master, switch the standby master as the new active and then rollout a new standby master.

Make sure that the cache and replication agents are executing on both masters. Once the cache agent can connect to the Oracle

Database, then the status of all autorefresh cache groups is set to Recovering. This sets the database to Recovering. Once a single cache group has received enough data to resume autorefresh, the status is set to OK. After all cache group are set to OK, the database is set to Alive.

Alternatively, you can rollout new masters.

Make sure that the cache and replication agents are executing on the standby master. Once the cache agent can connect to the

Oracle Database, then the status of all autorefresh cache groups is set to Recovering. This sets the database to Recovering. Once a single cache group has received enough data to resume autorefresh, the status is set to OK. After all cache group are set to OK, the database is set to Alive.

Alternatively, you can fail the standby master and rollout a new standby master.

Make sure that the cache and replication agents are executing on the active master. Use a manual refresh to recover the autorefresh cache groups on the active master. After all cache group are set to OK or have been dropped, the database is set to Alive.

Make sure that the cache and replication agents are executing on the active master. Use a manual refresh to recover the autorefresh cache groups on the active master. After all cache group are set to OK or have been dropped, the database is set to Alive. Changes are then replicated to the standby master.

Troubleshooting Oracle In-Memory Database Cache

4-25

Poor autorefresh performance

Table 4–2 (Cont.) Recovery for cache groups involved in active standby replication pair

DeadDbRecovery

Setting

None

None

None

Active

Master

Alive

Dead

Dead

Standby

Master

Dead

Alive

Dead

Resulting

Behavior

Mark the standby master as failed. Execute ttDestroy

utility for the standby master database. Duplicate the active master by executing ttRepAdmin -duplicate

utility from the active master.

Destroy the dead active master with the ttDesctroy

utility.

Recover the dead active master by duplicating the standby master with the ttRepAdmin -duplicate

utility.

Rollout new masters.

Excessive deadlocks, buffer busy and row lock waits during autorefresh cache group refresh

During an autorefresh cache group refresh, there can be excessive buffer busy waits, row lock waits, and deadlocks on updates in the Oracle database, which can negatively affect the throughput performance. When there are multiple deadlocks on updates in the Oracle database involving the autorefresh log tables, the following may appear in the support log:

Oracle native error code = 60, msg = ORA-00060: deadlock detected while waiting for resource

An error occurred while preparing or executing the following Oracle sql statement: <some statement involving <cache admin user>.TT_##_#######_L where the # is some number>

You can improve your performance by modifying the INITRANS and FREELISTS settings, which can affect the concurrent inserts into the autorefresh log table as well as internal maintenance of these tables. The application updating the base table that is being autorefreshed encounters a throughput performance hit when these settings are not appropriately configured.

You can automatically or manually manage these settings as follows:

Use ASSM tablespace, which automatically manages FREELISTS.

1.

Manually adjust FREELISTS and INITRANS for the autorefresh log table on the

Oracle database.

The following details how to manually modify INITRANS and FREELISTS for the autorefresh log table on the Oracle database:

Retrieve the name of the autorefresh log table that is on the Oracle database.

Under the cache administration user login, execute the SQL*Plus script cacheInfo.sql

that lists the autorefresh change log table name, along with other items. The following example executes the cacheInfo.sql

script that lists the autorefresh change log table name as tt_05_1216726_L, as shown in bold:

SQL> @cacheInfo.sql

*************Autorefresh Objects Information ***************

Host name: gordon-tt

Timesten datastore name: /scratch/ds/myDB

Cache table name: SCOTT.ALOBN

Change log table name: tt_05_1216726_L

Number of rows in change log table: 1

Maximum logseq on the change log table: 2

Timesten has autorefreshed updates upto logseq: 1

Number of updates waiting to be autorefreshed: 1

4-26

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Poor autorefresh performance

2.

Number of updates that has not been marked with a valid logseq: 0

****************************

Host name: conobar-tt

Timesten datastore name: /scratch/ds/myDB

Cache table name: SCOTT.A

Change log table name: tt_05_1279699_L

Number of rows in change log table: 7

Maximum logseq on the change log table: 0

Timesten has autorefreshed updates upto logseq: 0

Number of updates waiting to be autorefreshed: 5

Number of updates that has not been marked with a valid logseq: 5

****************************

Manually alter the table on the Oracle database. The following example uses the table from the previous example. This example alters the INITRANS and

FREELISTS settings for the bar.tt_05_1279699_L table.

Note:

See "INITRANS integer" and "FREELISTS" in the

Oracle

Database SQL Language Reference

for details on what are the correct values for configuring these settings.

ALTER TABLE BAR.TT_05_1279699_L INITRANS 10;

ALTER TABLE BAR.TT_05_1279699_L STORAGE(FREELISTS 5); or

ALTER TABLE BAR.TT_05_1279699_L MOVE STORAGE(FREELISTS 5);

3.

Alter the INITRANS and FREELISTS settings for the index for this table, which have the same name as the autorefresh change log table with an additional "L" at the end of it. For example, the index for table bar.tt_05_1279699_L

is bar.tt_05_1279699_LL

These settings should be the same as what you set for the autorefresh change log table.

ALTER INDEX BAR.TT_05_1279699_LL INITRANS 10;

ALTER INDEX BAR.TT_05_1279699_LL STORAGE(FREELISTS 5);

Abnormally large log and base tables degrade autorefresh performance

The cache thread SQL refresh joins the log table and the base table, which identifies rows needed to be refreshed into TimesTen. The larger the cardinalities of the base table and the log table, the longer the time necessary to perform this join. Performance degradation may occur if either the log table or the base table is abnormally large.

The following describe scenarios where the log table can become abnormally large:

If the log table is never purged in configurations where cache groups from multiple DSNs all reference the same base table, it increases in size indefinitely. If one or more of the cache agents for these groups are turned off, those DSNs will not properly refresh their cache groups and the log tables will not be purged. If the autorefresh state is turned to paused on one of multiple nodes, the other nodes may slow down.

The log table can grow abnormally large if some of the cache agents have been shutdown. Resolve this issue by restarting the cache, which will purge all of the backlogged log rows to be purged and all of the cache groups to be synchronized after the completion of the refresh cycle for all cache groups.

Troubleshooting Oracle In-Memory Database Cache

4-27

Poor autorefresh performance

The log table can be abnormally large if rows inserted into the log table are never purged and can never be purged by normal processing. This occurs when one or more DSNs are destroyed or rebuilt without first removing the cache groups. The cache group tables on the Oracle database have no information that the cache groups have been destroyed, which corrupts the entire cache group. Rebuild and reinitialize all of the cache groups associated with this base table. Alternatively, never destroy a DSN with cache groups. Instead, always drop the cache groups before destroying a DSN.

Performance degrades when autorefresh interval is small

When a relatively short refresh interval, such as a few hundred milliseconds, is combined with a large number of entries in the log table or in the base table, a cache refresh operation does not complete before the next refresh operation is scheduled to begin. In this case, the entries in the log table can be un-marked when the current autorefresh cycle finishes.

Thus, the same rows can be refreshed from the base table to the cache group in the next autorefresh cycle, by which time the rows will be marked. Make sure that the time it for the refresh is greater than the refresh interval. Set the refresh interval to a value where redundant refreshes will not occur.

4-28

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

5

5

Troubleshooting AWT Cache Groups

Creating an asynchronous writethrough (AWT) cache group automatically creates a replication scheme that allows the data store to communicate with the Oracle database. You must start the replication agent after you create an AWT cache group and start the cache agent. See "Creating an AWT cache group" in the

Oracle In-Memory

Database Cache User's Guide

.

Material in

Chapter 6, "Troubleshooting Replication"

is useful for troubleshooting

AWT cache group problems. Useful replication topics are summarized in the current chapter in these sections:

Unable to start or stop replication agent

Replication does not work

Using SNMP traps for notification of replication events

This chapter also contains the following sections:

Monitoring AWT performance

Possible causes of poor AWT performance

Permanent Oracle errors reported by TimesTen

Transient Oracle errors reported by TimesTen

Unable to start or stop replication agent

This section describes what to check if you are unable to start or stop a replication agent.

Possible cause What to do

You do not have ADMIN privileges. You must have ADMIN privileges to use the ttAdmin utility or the ttRepStart

or ttRepStop

procedures to start or stop a replication agent.

TimesTen daemon not started Check the state of the TimesTen daemon, as described in

"Check the TimesTen user error log" on page 2-2. If

necessary, start the TimesTen daemon as described in

"Working with the Oracle TimesTen Data Manager

Daemon" in the

Oracle TimesTen In-Memory Database

Operations Guide

.

Troubleshooting AWT Cache Groups

5-1

Replication does not work

Replication does not work

If you are unable to get replication working, the problem may be one or more of the following:

Possible Cause

TimesTen daemon or replication agents not running

Replication agents not communicating

Replication not in Start state

See...

"Check status of TimesTen daemon and replication agents" on page 6-3

"Check that replication agents are communicating" on page 6-5

"Check replication state" on page 6-5

Using SNMP traps for notification of replication events

TimesTen can send SNMP traps for certain replication events to enable network management software to take immediate action. TimesTen can send the following

SNMP traps:

■ ttRepAgentExitingTrap

■ ttRepAgentDiedTrap

■ ttRepAgentStartingTrap

These traps are described in "Diagnostics through SNMP Traps" in the

Oracle TimesTen

In-Memory Database Error Messages and SNMP Traps

.

Monitoring AWT performance

You can monitor the performance of asynchronous writethrough (AWT) cache groups to determine how much time is spent performing tasks in the AWT workflow. Use the ttCacheAwtMonitorConfig

built-in procedure to enable monitoring.

For example, enable monitoring and set the sampling frequency to 16. A sampling factor of 16 is recommended for accuracy and performance.

Command> CALL ttCacheAwtMonitorConfig('ON',16);

< ON, 16 >

1 row found.

Use the ttRepAdmin

utility with the

-awtmoninfo

and

-showstatus

commands to display the monitoring results. The AWT monitoring statistics include:

TimesTen processing time: The total number of milliseconds spent in processing

AWT transaction data since monitoring was enabled.

Oracle bookmark management time: The total number of milliseconds spent in managing AWT metadata on Oracle since monitoring was enabled.

Oracle execute time: The total number of milliseconds spent in OCI preparation, binding and execution for AWT SQL operations since monitoring was enabled.

This statistic includes network latency between TimesTen and Oracle.

Oracle commit time: The total number of milliseconds spent in committing AWT updates on Oracle since monitoring was enabled. This statistic includes network latency between TimesTen and Oracle.

Time since monitoring was started

5-2

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Permanent Oracle errors reported by TimesTen

Total number of TimesTen row operations: The total number of rows updated in

AWT cache groups since monitoring was enabled.

Total number of TimesTen transactions: The total number of transactions in AWT cache groups since monitoring was enabled.

Total number of flushes to Oracle: The total number of times that TimesTen data has been sent to Oracle.

The output also includes the percentage of time spent on TimesTen processing, Oracle bookmark management, Oracle execution and Oracle commits.

For example: ttRepAdmin -showstatus -awtmoninfo myDSN

[other -showstatus output]

...

AWT Monitoring statistics

-------------------------

TimesTen processing time : 0.689000 millisecs (0.164307 %)

Oracle bookmark management time : 3.229000 millisecs (0.770027%)

Oracle execute time : 342.908000 millisecs (81.774043 %)

Oracle commit time : 72.450000 millisecs (17.277315 %)

Time since monitoring was started: 8528.641000 millisecs

Cache-connect Operational Stats :

Total Number of TimesTen row operations : 2

Total Number of TimesTen transactions : 2

Total Number of flushes to Oracle : 2

Possible causes of poor AWT performance

This section addresses issues that may degrade AWT performance.

Possible cause

Slow network

Log buffer too small

Frequent or inefficient disk writes

Reading from transaction log files on disk instead of the log buffer

See...

"Check network bandwidth" on page 6-11

"Check size of log buffer" on page 6-12

"Check durability settings" on page 6-12

"Check for reads from transaction log files" on page 6-12

Permanent Oracle errors reported by TimesTen

Insert, update, or delete errors that occur while applying changes to Oracle are saved in an error file located in the data store directory with the following name:

DatastoreName.awterr

Errors reported to this file are

permanent

errors. TimesTen does not retry the transaction. The errors may be reported in the AWT error file long after the commit to

TimesTen occurs.

The format of the messages in the AWT error file is similar to those generated for conflict and transaction errors in replication, as shown in

Example 5–1 . Oracle error

messages are also reported in the support log and the user log.

Troubleshooting AWT Cache Groups

5-3

Transient Oracle errors reported by TimesTen

Example 5–1 Cache violation occurs when update is propagated to Oracle

If a constraint violation occurs when a cache group update is propagated to Oracle, the message in the AWT error file is similar to the following:

Error occurred 14:48:55 on 03-22-2007

Datastore: c:\temp\cgDSN

Oracle Id: system1

Transmitting name: cgDSN

Error message:

TT5210: Oracle unique constraint violation error in OCIStmtExecute():

ORA-00001: unique constraint (GUSER.SYS_C00357240) violated rc = -1 -- file

"bdbTblH.c", lineno 1205, procedure "ttBDbStmtForce()"

TT5025: Commit failure in Oracle. Transaction must be rolled back in TimesTen.

-- file "bdbConnect.c", lineno 885, procedure "ttBDbXact()"

Operation that caused the error:

Insert into table TESTUSER.T1 <9,1000>

Failed transaction:

Insert into table TESTUSER.T1 <9, 1000>

End of failed transaction

Example 5–2 An object that TimesTen has placed on Oracle is dropped

If an object that TimesTen has placed on Oracle is dropped, the message in the AWT error file is similar to the following:

May 04 18:12:36 HOST1 TimesTen Replication 7.0[2136]:

[Err ] DEFAULT:meta.c(639):

TT16062: Failed to compile command: select p.commit_timestamp, p.commit_seqnum, p.protocol from owner1.TT_03_REPPEERS p where p.replication_name = :rname and p.replication_owner = :rowner and p.tt_store_id = :oid and p.subscriber_id = :sid

May 04 18:12:36 HOST1 TimesTen Replication 7.0[2136]:

[Err ] DEFAULT:meta.c(639):

TT5221: TT5221: Oracle syntax error in OCIStmtExecute():

ORA-00942: table or view does not exist rc = -1 -- file "bdbStmt.c", lineno 1041, procedure "getOraOutTypesNLengths()"

In this example, the

TT_03_REPPEERS

table does not exist. To recover from this error, perform the following tasks:

1.

2.

3.

Stop the replication agent.

Drop and re-create the cache group.

Restart the replication agent.

Transient Oracle errors reported by TimesTen

The support log for data stores with AWT cache groups may contain Oracle errors if the replication agent encounters a problem on the Oracle database. If the replication agent encounters one of these errors, AWT rolls back the transaction and retries it. If the support log becomes full, the oldest messages are deleted and replaced by new messages.

The Oracle errors in the support log are considered

transient

because AWT retries the transaction.

5-4

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Transient Oracle errors reported by TimesTen

Some transient errors indicate an underlying problem on the Oracle database must be solved before AWT operations can continue. For example:

ORA-01536: space quota exceeded for tablespace

ORA-01034: ORACLE not available

After the underlying problem has been fixed, AWT retries the operation.

For more information about the Oracle errors, see

Oracle Database Error Messages

for the Oracle release you are using.

The following Oracle errors are transient:

ORA-00018: maximum number of sessions exceeded

ORA-00019: maximum number of session licenses exceeded

ORA-00020: maximum number of processes (%s) exceeded

ORA-00025: failed to allocate %s

ORA-00028: your session has been killed

ORA-00038: Cannot create session: server group belongs to another user

ORA-00051: timeout occurred while waiting for a resource

ORA-00052: maximum number of enqueue resources (%s) exceeded

ORA-00053: maximum number of enqueues exceeded

ORA-00054: resource busy and acquire with NOWAIT specified

ORA-00055: maximum number of DML locks exceeded

ORA-00057: maximum number of temporary table locks exceeded

ORA-00058: DB_BLOCK_SIZE must be %s to mount this database (not %s)

ORA-00059: maximum number of DB_FILES exceeded

ORA-00060: deadlock detected while waiting for resource

ORA-00063: maximum number of LOG_FILES exceeded

ORA-00064: object is too large to allocate on this O/S (%s,%s)

ORA-00099: timed out while waiting for resource, potential PDML deadlock

ORA-00104: deadlock detected; all public servers blocked waiting for resources

ORA-00107: failed to connect to ORACLE listener process

ORA-00115: connection refused; dispatcher connection table is full

ORA-00125: connection refused; invalid presentation

ORA-00126: connection refused; invalid duplicity

ORA-00284: recovery session still in progress

ORA-00370: potential deadlock during kcbchange operation

ORA-00371: not enough shared pool memory

ORA-00376: file %s cannot be read at this time

ORA-00379: no free buffers available in buffer pool %s for block size %sK

ORA-00384: Insufficient memory to grow cache

ORA-00568: Maximum number of interrupt handlers exceeded

ORA-00579: osndnt: server received malformed connection request

ORA-00600: internal error code, arguments: [%s], [%s], [%s], [%s], [%s], [%s],

[%s], [%s]

ORA-00603: ORACLE server session terminated by fatal error

ORA-01000: maximum open cursors exceeded

ORA-01012: not logged on

ORA-01014: ORACLE shutdown in progress

ORA-01019: unable to allocate memory in the user side

ORA-01031: insufficient privileges

ORA-01033: ORACLE initialization or shutdown in progress

ORA-01034: ORACLE not available

ORA-01035: ORACLE only available to users with RESTRICTED SESSION privilege

ORA-01037: maximum cursor memory exceeded

ORA-01046: cannot acquire space to extend context area

ORA-01073: fatal connection error: unrecognized call type

ORA-01089: immediate shutdown in progress - no operations are permitted

ORA-01090: shutdown in progress - connection is not permitted

ORA-01092: ORACLE instance terminated. Disconnection forced

ORA-01094: ALTER DATABASE CLOSE in progress. Connections not permitted

Troubleshooting AWT Cache Groups

5-5

Transient Oracle errors reported by TimesTen

ORA-01109: database not open

ORA-01147: SYSTEM tablespace file %s is offline

ORA-01154: database busy. Open, close, mount, and dismount not allowed now

ORA-01155: the database is being opened, closed, mounted or dismounted

ORA-01219: database not open: queries allowed on fixed tables/views only

ORA-01237: cannot extend datafile %s

ORA-01456: may not perform insert/delete/update operation inside a READ ONLY transaction

ORA-01536: space quota exceeded for tablespace '%s'

ORA-01539: tablespace '%s' is not online

ORA-01542: tablespace '%s' is offline, cannot allocate space in it

ORA-01562: failed to extend rollback segment number %s

ORA-01573: shutting down instance, no further change allowed

ORA-01628: max # extents (%s) reached for rollback segment %s

ORA-01629: max # extents (%s) reached saving undo for tablespace %s

ORA-01630: max # extents (%s) reached in temp segment in tablespace %s

ORA-01631: max # extents (%s) reached in table %s.%s

ORA-01632: max # extents (%s) reached in index %s.%s

ORA-01650: unable to extend rollback segment %s by %s in tablespace %s

ORA-01651: unable to extend save undo segment by %s for tablespace %s

ORA-01652: unable to extend temp segment by %s in tablespace %s

ORA-01653: unable to extend table %s.%s by %s in tablespace %s

ORA-01654: unable to extend index %s.%s by %s in tablespace %s

ORA-01655: unable to extend cluster %s.%s by %s in tablespace %s

ORA-01656: max # extents (%s) reached in cluster %s.%s

ORA-01658: unable to create INITIAL extent for segment in tablespace %s

ORA-01659: unable to allocate MINEXTENTS beyond %s in tablespace %s

ORA-01680: unable to extend LOB segment by %s in tablespace %s

ORA-01681: max # extents (%s) reached in LOB segment in tablespace %s

ORA-01683: unable to extend index %s.%s partition %s by %s in tablespace %s

ORA-01684: max # extents (%s) reached in table %s.%s partition %s

ORA-01685: max # extents (%s) reached in index %s.%s partition %s

ORA-01686: max # files (%s) reached for the tablespace %s

ORA-01688: unable to extend table %s.%s partition %s by %s in tablespace %s

ORA-01691: unable to extend lob segment %s.%s by %s in tablespace %s

ORA-01692: unable to extend lob segment %s.%s partition %s by %s in tablespace %s

ORA-01693: max # extents (%s) reached in lob segment %s.%s

ORA-01694: max # extents (%s) reached in lob segment %s.%s partition %s

ORA-03113: end-of-file on communication channel

ORA-03114: not connected to ORACLE

ORA-03134: Connections to this server version are no longer supported.

ORA-03135: connection lost contact

ORA-03136: inbound connection timed out

ORA-03232: unable to allocate an extent of %s blocks from tablespace %s

ORA-03233: unable to extend table %s.%s subpartition %s by %s in tablespace %s

ORA-03234: unable to extend index %s.%s subpartition %s by %s in tablespace %s

ORA-03235: max # extents (%s) reached in table %s.%s subpartition %s

ORA-03236: max # extents (%s) reached in index %s.%s subpartition %s

ORA-03237: Initial Extent of specified size cannot be allocated

ORA-03238: unable to extend LOB segment %s.%s subpartition %s by %s in tablespace

%s

ORA-03239: maxextents (%s) reached in LOB segment %s.%s subpartition %s

ORA-04020: deadlock detected while trying to lock object %s%s%s%s%s

ORA-06019: NETASY: invalid login (connect) string

ORA-06021: NETASY: connect failed

ORA-06030: NETDNT: connect failed, unrecognized node name

ORA-06031: NETDNT: connect failed, unrecognized object name

ORA-06032: NETDNT: connect failed, access control data rejected

ORA-06033: NETDNT: connect failed, partner rejected connection

ORA-06034: NETDNT: connect failed, partner exited unexpectedly

5-6

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Transient Oracle errors reported by TimesTen

ORA-06035: NETDNT: connect failed, insufficient resources

ORA-06036: NETDNT: connect failed, no response from object

ORA-06037: NETDNT: connect failed, node unreachable

ORA-06039: NETDNT: connect failed

ORA-06040: NETDNT: invalid login (connect) string

ORA-06108: NETTCP: connect to host failed

ORA-06113: NETTCP: Too many connections

ORA-06114: NETTCP: SID lookup failure

ORA-06143: NETTCP: maximum connections exceeded

ORA-06315: IPA: Invalid connect string

ORA-06316: IPA: Invalid database SID

ORA-06317: IPA: Local maximum number of users exceeded

ORA-06318: IPA: Local maximum number of connections exceeded

ORA-06319: IPA: Remote maximum number of users exceeded

ORA-06320: IPA: Remote maximum number of connections exceeded

ORA-06404: NETCMN: invalid login (connect) string

ORA-06413: Connection not open.

ORA-10435: enable tracing of global enqueue service deadlock detetction

ORA-10626: specify timeout for online index rebuild to wait for DML

ORA-10906: Unable to extend segment after insert direct load

ORA-12150: TNS:unable to send data

ORA-12151: TNS:received bad packet type from network layer

ORA-12152: TNS:unable to send break message

ORA-12153: TNS:not connected

ORA-12154: TNS:could not resolve service name

ORA-12155: TNS:received bad datatype in NSWMARKER packet

ORA-12156: TNS:tried to reset line from incorrect state

ORA-12157: TNS:internal network communication error

ORA-12158: TNS:could not initialize parameter subsystem

ORA-12159: TNS:trace file not writeable

ORA-12160: TNS:internal error: Bad error number

ORA-12161: TNS:internal error: partial data received

ORA-12162: TNS:service name is incorrectly specified

ORA-12163: TNS:connect descriptor is too long

ORA-12166: TNS:Client can not connect to HO agent.

ORA-12168: TNS:Unable to contact Directory Server.

ORA-12169: TNS:Net service name given as connect identifier is too long

ORA-12170: TNS:Connect timeout occurred

ORA-12171: TNS:could not resolve connect identifier: %s

ORA-12196: TNS:received an error from TNS

ORA-12197: TNS:keyword-value resolution error

ORA-12198: TNS:could not find path to destination

ORA-12200: TNS:could not allocate memory

ORA-12201: TNS:encountered too small a connection buffer

ORA-12202: TNS:internal navigation error

ORA-12203: TNS:unable to connect to destination

ORA-12204: TNS:received data refused from an application

ORA-12205: TNS:could not get failed addresses

ORA-12206: TNS:received a TNS error during navigation

ORA-12207: TNS:unable to perform navigation

ORA-12208: TNS:could not find the TNSNAV.ORA file

ORA-12209: TNS:encountered uninitialized global

ORA-12210: TNS:error in finding Navigator data

ORA-12211: TNS:needs PREFERRED_CMANAGERS entry in TNSNAV.ORA

ORA-12212: TNS:incomplete PREFERRED_CMANAGERS binding in TNSNAV.ORA

ORA-12213: TNS:incomplete PREFERRED_CMANAGERS binding in TNSNAV.ORA

ORA-12214: TNS:missing local communities entry in TNSNAV.ORA

ORA-12215: TNS:poorly formed PREFERRED_NAVIGATORS Addresses in TNSNAV.ORA

ORA-12216: TNS:poorly formed PREFERRED_CMANAGERS addresses in TNSNAV.ORA

ORA-12217: TNS:could not contact PREFERRED_CMANAGERS in TNSNAV.ORA

Troubleshooting AWT Cache Groups

5-7

Transient Oracle errors reported by TimesTen

ORA-12218: TNS:unacceptable network configuration data

ORA-12219: TNS:missing community name from address in ADDRESS_LIST

ORA-12221: TNS:illegal ADDRESS parameters

ORA-12222: TNS:no such protocol adapter

ORA-12223: TNS:internal limit restriction exceeded

ORA-12224: TNS:no listener

ORA-12225: TNS:destination host unreachable

ORA-12226: TNS:operating system resource quota exceeded

ORA-12227: TNS:syntax error

ORA-12228: TNS:protocol adapter not loadable

ORA-12229: TNS:Interchange has no more free connections

ORA-12230: TNS:Severe Network error ocurred in making this connection

ORA-12231: TNS:No connection possible to destination

ORA-12232: TNS:No path available to destination

ORA-12233: TNS:Failure to accept a connection

ORA-12235: TNS:Failure to redirect to destination

ORA-12236: TNS:protocol adapter not loaded

ORA-12316: syntax error in database link's connect string

ORA-12326: database %s is closing immediately; no operations are permitted

ORA-12329: database %s is closed; no operations are permitted

ORA-12500: TNS:listener failed to start a dedicated server process

ORA-12501: TNS:listener failed to spawn process

ORA-12502: TNS:listener received no CONNECT_DATA from client

ORA-12504: TNS:listener was not given the SID in CONNECT_DATA

ORA-12505: TNS:listener could not resolve SID given in connect descriptor

ORA-12506: TNS:listener was not given the ALIAS in CONNECT_DATA

ORA-12507: TNS:listener could not resolve ALIAS given

ORA-12508: TNS:listener could not resolve the COMMAND given

ORA-12509: TNS:listener failed to redirect client to service handler

ORA-12510: TNS:database temporarily lacks resources to handle the request

ORA-12511: TNS:service handler found but it is not accepting connections

ORA-12512: TNS:service handler found but it has not registered a redirect address

ORA-12513: TNS:service handler found but it has registered for a different protocol

ORA-12514: TNS:listener could not resolve SERVICE_NAME given in connect descriptor

ORA-12515: TNS:listener could not find a handler for this presentation

ORA-12516: TNS:listener could not find available handler with matching protocol stack

ORA-12517: TNS:listener could not find service handler supporting direct handoff

ORA-12518: TNS:listener could not hand off client connection

ORA-12519: TNS:no appropriate service handler found

ORA-12520: TNS:listener could not find available handler for requested type of server

ORA-12521: TNS:listener could not resolve INSTANCE_NAME given in connect descriptor

ORA-12522: TNS:listener could not find available instance with given INSTANCE_ROLE

ORA-12523: TNS:listener could not find instance appropriate for the client connection

ORA-12524: TNS:listener could not resolve HANDLER_NAME given in connect descriptor

ORA-12525: TNS:listener has not received client's request in time allowed

ORA-12526: TNS:listener: all appropriate instances are in restricted mode

ORA-12527: TNS:listener: all instances are in restricted mode or blocking new connections

ORA-12528: TNS:listener: all appropriate instances are blocking new connections

ORA-12529: TNS:connect request rejected based on current filtering rules

ORA-12531: TNS:cannot allocate memory

ORA-12532: TNS:invalid argument

ORA-12533: TNS:illegal ADDRESS parameters

ORA-12534: TNS:operation not supported

ORA-12535: TNS:operation timed out

5-8

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Transient Oracle errors reported by TimesTen

ORA-12536: TNS:operation would block

ORA-12537: TNS:connection closed

ORA-12538: TNS:no such protocol adapter

ORA-12539: TNS:buffer over- or under-flow

ORA-12540: TNS:internal limit restriction exceeded

ORA-12541: TNS:no listener

ORA-12542: TNS:address already in use

ORA-12543: TNS:destination host unreachable

ORA-12544: TNS:contexts have different wait/test functions

ORA-12545: Connect failed because target host or object does not exist

ORA-12546: TNS:permission denied

ORA-12547: TNS:lost contact

ORA-12549: TNS:operating system resource quota exceeded

ORA-12550: TNS:syntax error

ORA-12551: TNS:missing keyword

ORA-12552: TNS:operation was interrupted

ORA-12554: TNS:current operation is still in progress

ORA-12555: TNS:permission denied

ORA-12556: TNS:no caller

ORA-12557: TNS:protocol adapter not loadable

ORA-12558: TNS:protocol adapter not loaded

ORA-12560: TNS:protocol adapter error

ORA-12561: TNS:unknown error

ORA-12562: TNS:bad global handle

ORA-12564: TNS:connection refused

ORA-12566: TNS:protocol error

ORA-12569: TNS:packet checksum failure

ORA-12570: TNS:packet reader failure

ORA-12571: TNS:packet writer failure

ORA-12574: TNS:redirection denied

ORA-12582: TNS:invalid operation

ORA-12583: TNS:no reader

ORA-12585: TNS:data truncation

ORA-12589: TNS:connection not bequeathable

ORA-12590: TNS:no I/O buffer

ORA-12591: TNS:event signal failure

ORA-12592: TNS:bad packet

ORA-12593: TNS:no registered connection

ORA-12595: TNS:no confirmation

ORA-12596: TNS:internal inconsistency

ORA-12600: TNS: string open failed

ORA-12602: TNS: Connection Pooling limit reached

ORA-12606: TNS: Application timeout occurred

ORA-12607: TNS: Connect timeout occurred

ORA-12608: TNS: Send timeout occurred

ORA-12609: TNS: Receive timeout occurred

ORA-12612: TNS:connection is busy

ORA-12615: TNS:preempt error

ORA-12623: TNS:operation is illegal in this state

ORA-12624: TNS:connection is already registered

ORA-12636: Packet send failed

ORA-12637: Packet receive failed

ORA-12829: Deadlock - itls occupied by siblings at block %s of file %s

ORA-12993: tablespace '%s' is offline, cannot drop column

ORA-14117: partition resides in offlined tablespace

ORA-14268: subpartition '%s' of the partition resides in offlined tablespace

ORA-16000: database open for read-only access

ORA-16003: standby database is restricted to read-only access

ORA-16403: shutdown in progress - remote connection is not permitted

ORA-16724: the intended state for resource has been set to OFFLINE

Troubleshooting AWT Cache Groups

5-9

Transient Oracle errors reported by TimesTen

ORA-16903: Unable to connect to database

ORA-16914: Missing connect string. Try \"help\"

ORA-18014: deadlock detected while waiting for resource %s

ORA-21521: exceeded maximum number of connections in OCI (object mode only)

ORA-21522: attempted to use an invalid connection in OCI (object mode only)

ORA-23317: a communication failure has occurred

ORA-24401: cannot open further connections

ORA-24418: Cannot open further sessions.

ORA-24778: cannot open connections

ORA-25400: must replay fetch

ORA-25401: can not continue fetches

ORA-25402: transaction must roll back

ORA-25403: could not reconnect

ORA-25405: transaction status unknown

ORA-25407: connection terminated

ORA-25408: can not safely replay call

ORA-25409: failover happened during the network operation,cannot continue

ORA-25425: connection lost during rollback

ORA-29306: datafile %s is not online

ORA-30006: resource busy; acquire with WAIT timeout expired

ORA-30036: unable to extend segment by %s in undo tablespace '%s'

ORA-30040: Undo tablespace is offline

ORA-31443: deadlock detected while acquiring lock on %s

ORA-37013: (XSACQUIRE_DEADLOCK) Cannot wait to acquire object %j since doing so would cause a deadlock.

ORA-44317: database open read-only

5-10

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

6

6

Troubleshooting Replication

This chapter describes how to troubleshoot some of the problems you may encounter when replicating data stores.

This chapter includes the following topics:

Unable to create a replication scheme

Unable to alter a replication scheme

Unable to start or stop replication agent

Using SNMP traps for notification of replication events

Replication does not work

Replication unresponsive, appears hung

Poor replication or XLA performance

Problems using ttRepAdmin

Problems with conflict checking

Unable to create a replication scheme

This section describes what to check if you are unable to use CREATE REPLICATION to create a replication scheme.

Possible cause

You do not have ADMIN privilege

Incorrect data store name, host name, or element name.

The local host is not part of the replication scheme.

What to do

You must have ADMIN privilege to use the CREATE

REPLICATION or DROP REPLICATION statements.

Check the CREATE REPLICATION statement for typographical errors.

See "Check host names" on page 6-7.

Use official host names instead of aliases.

The host name should match the value returned by the hostname

command on your system and should be used consistently throughout the replication scheme.

Create the replication scheme on a host that will be part of the replication scheme.

Troubleshooting Replication

6-1

Unable to alter a replication scheme

Possible cause What to do

Replication tables defined in the CREATE

REPLICATION statement do not exist.

Other problems

The name, owner, and column definitions of the tables participating in the replication scheme must be identical on both the master and subscriber data stores. Use CREATE TABLE to create tables on the data store, or use the ttRepAdmin

-duplicate

utility or the ttRepDuplicateEx

C function to duplicate the entire data store to be replicated.

Review the procedures and requirements described in "Defining

Replication Schemes" in the

Oracle TimesTen In-Memory Database

TimesTen to TimesTen Replication Guide

.

Unable to alter a replication scheme

This section describes what to check if you are unable to use ALTER REPLICATION to alter a replication scheme.

Possible cause What to do

You do not have ADMIN privilege

Replication agent in Start state

You must have ADMIN privilege to use the ALTER

REPLICATION statement.

Most ALTER REPLICATION operations are supported only when the replication agent is stopped ( ttAdmin -repStop

). Stop the replication agents on both master and subscriber data stores, alter the replication scheme on both master and subscriber data stores, then restart both replication agents.

Incorrect data store name, host name, or element name

Replication table defined in the ALTER

REPLICATION statement does not exist

Other problems

Check ALTER REPLICATION statement for typographical errors.

See "Check host names" on page 6-7.

Use CREATE TABLE to create a table on the data store.

Review the procedures and requirements described in "Altering Replication" in the

Oracle TimesTen

In-Memory Database TimesTen to TimesTen Replication

Guide

.

Unable to start or stop replication agent

This section describes what to check if you are unable to start or stop a replication agent.

Possible cause

You do not have ADMIN privileges

TimesTen daemon not started

What to do

You must have ADMIN privileges to use the ttAdmin

utility or the ttRepStart

or ttRepStop procedures to start or stop a replication agent.

Check the state of the TimesTen daemon, as described

in "Check the TimesTen user error log" on page 2-2. If

necessary, start the TimesTen daemon as described in

"Working with the Oracle TimesTen Data Manager

Daemon" in the

Oracle TimesTen In-Memory Database

Operations Guide

.

6-2

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Replication does not work

Possible cause

Data store does not participate in a replication scheme.

What to do

If a data store does not participate in a replication scheme, attempts to start a replication agent for that data store will fail. Use CREATE REPLICATION to create a replication scheme for the data store.

Using SNMP traps for notification of replication events

TimesTen can send SNMP traps for certain replication events to enable network management software to take immediate action. TimesTen can send the following traps for replication events:

■ ttRepAgentExitingTrap

■ ttRepAgentDiedTrap ttRepAgentStartingTrap ttRepCatchupStartTrap ttRepCatchupStopTrap

■ ttRepReturnTransitionTrap ttRepSubscriberFailedTrap

■ ttRepUpdateFailedTrap

These traps are described in "Diagnostics through SNMP Traps" in the

Oracle TimesTen

In-Memory Database Error Messages and SNMP Traps

.

Replication does not work

If you are unable to get replication working between a master and subscriber data store, the problem may be one or more of the following:

Possible cause See...

TimesTen daemon and/or replication agents not running

"Check status of TimesTen daemon and replication agents" on page 6-3

Master and subscriber agents not communicating

Replication not in Start state

Error in replication scheme

Inconsistent owner names for replication scheme and tables

Inconsistent replication tables

"Check that replication agents are communicating" on page 6-5

"Check replication state" on page 6-5

"Check replication scheme configuration" on page 6-6

"Check owner names" on page 6-8

"Check consistency between replicated tables" page 6-10

on

Check status of TimesTen daemon and replication agents

Use the ttStatus

utility to confirm the main TimesTen daemon is running and the replication agents are started for all of your master and subscriber data stores. The output from a simple replication scheme using a single master and subscriber data

store should look like that shown in Example 6–1 .

If the TimesTen daemon is running, but the replication agents are not, the output looks

like that shown in Example 6–2 . In this case, start the replication agents as described in

Troubleshooting Replication

6-3

Replication does not work

"Starting and stopping the replication agents" in the

Oracle TimesTen In-Memory

Database TimesTen to TimesTen Replication Guide

.

If neither the TimesTen daemon or replication agents are running, the output looks like that shown in

Example 6–3

. In this case, confirm you have correctly installed

TimesTen and the Data Manager service is started, as described in "TimesTen

Installation" in the

Oracle TimesTen In-Memory Database Installation Guide

.

Example 6–1 ttStatus output for one master and one subscriber

C:\>ttStatus

TimesTen status report as of Thu Jan 25 16:23:33 2007

Daemon pid 5088 port 17000 instance MYINSTANCE

TimesTen server pid 4344 started on port 17002

TimesTen webserver pid 4216 started on port 17004

------------------------------------------------------------------------

Data store c:\temp\subscriber1ds

There are 12 connections to the data store

Data store is in shared mode

Shared Memory KEY Global\DBI45b9471c.2.SHM.2 HANDLE 0x280

Type PID Context Connection Name ConnID

Process 1244 0x00d08fb0 subscriber1ds 1

Replication 4548 0x00aed2f8 LOGFORCE 4

Replication 4548 0x00b03470 TRANSMITTER 5

Replication 4548 0x00b725a8 RECEIVER 6

Replication 4548 0x00b82808 REPHOLD 2

Replication 4548 0x00b98980 REPLISTENER 3

Subdaemon 2752 0x00526768 Worker 2042

Subdaemon 2752 0x0072a758 Flusher 2043

Subdaemon 2752 0x007308c0 Checkpoint 2044

Subdaemon 2752 0x00736a28 HistGC 2046

Subdaemon 2752 0x067f02f8 Aging 2045

Subdaemon 2752 0x068364a0 Monitor 2047

Replication policy : Manual

Replication agent is running.

Cache agent policy : Manual

------------------------------------------------------------------------

Data store c:\temp\masterds

There are 12 connections to the data store

Data store is in shared mode

Shared Memory KEY Global\DBI45b945d0.0.SHM.6 HANDLE 0x2bc

Type PID Context Connection Name ConnID

Process 5880 0x00d09008 masterds 1

Replication 3728 0x00aed570 LOGFORCE 4

Replication 3728 0x00b036e8 TRANSMITTER 5

Replication 3728 0x00b168b8 REPHOLD 3

Replication 3728 0x00b1ca20 REPLISTENER 2

Replication 3728 0x00b22b88 RECEIVER 6

Subdaemon 3220 0x00526768 Worker 2042

Subdaemon 3220 0x0072e768 Flusher 2043

Subdaemon 3220 0x007348d0 Checkpoint 2044

Subdaemon 3220 0x067b0068 Aging 2045

Subdaemon 3220 0x067c0040 Monitor 2047

Subdaemon 3220 0x068404c8 HistGC 2046

Replication policy : Manual

Replication agent is running.

Cache agent policy : Manual

------------------------------------------------------------------------

Data store c:\temp\demo

There are no connections to the data store

6-4

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Replication does not work

Replication policy : Manual

Cache agent policy : Manual

------------------------------------------------------------------------

End of report

Example 6–2 Replication agent is not running

> ttStatus

TimesTen status report as of Tue Oct 28 10:31:30 2006

Daemon pid 3396 port 15000 instance MYINSTANCE

TimesTen server pid 3436 started on port 15002

-----------------------------------------------------------------

Data store c:\temp\subscriberds

There are no connections to the data store cache agent restart policy: manual

-----------------------------------------------------------------

Data store c:\temp\masterds

There are no connections to the data store cache agent restart policy: manual

-----------------------------------------------------------------

End of report

Example 6–3 TimesTen daemon and replication agent are not running

> ttStatus ttStatus: Could not connect to TimesTen daemon: Connection refused

Check that replication agents are communicating

Use ttRepAdmin -receiver -list

to see that the replication agents are communicating with each other. If the masterds

data store is replicating to subscriberds

, the output should look similar to the following:

Example 6–4 Check that the replication agents are communicating

> ttRepAdmin -receiver -list masterDSN

Peer name Host name

----------------

SUBSCRIBERDS

Port

------------------------ ------

MYHOST Auto

State Proto

------- -----

Start 10

Last Msg Sent Last Msg Recv Latency TPS RecordsPS Logs

------------- ------------- ------- ------- --------- ----

0:01:12 19.41 5 52 2

Check replication state

Use the ttReplicationStatus

procedure to check state of the subscriber data store with respect to its master. If the subscriber is in the

Stop

,

Pause

, or

Failed

state, use the ttReplicationStatus

procedure to reset the subscriber state to

Start

, as described in "Setting the replication state of subscribers" in the

Oracle TimesTen

In-Memory Database TimesTen to TimesTen Replication Guide

.

Example 6–5 Obtain status of the subscriber data store from the master data store

Use ttReplicationStatus

to obtain the status of the subscriberds

data store from its master data store, masterDSN

, enter:

> ttIsql masterDSN

Command> CALL ttReplicationStatus ('subscriberds');

Troubleshooting Replication

6-5

Replication does not work

< SUBSCRIBERDS, MYHOST, 0, pause, 1, 10, REPSCHEME, REPL >

1 row found.

To reset state to

Start

call the ttRepSubscriberStateSet

procedure:

Command> CALL ttRepSubscriberStateSet('REPSCHEME', 'REPL', 'SUBSCRIBERDS',

'MYHOST', 0)

Command> CALL ttReplicationStatus ('subscriberds');

< SUBSCRIBERDS, MYHOST, 0, start, 1, 152959, REPSCHEME, REPL >

1 row found.

Check replication scheme configuration

This section describes some procedures you can use to confirm the correct configuration of the various components in your replicated system. The basic procedure categories are:

Check ttRepAdmin -showconfig

Check the TTREP.TTSTORES table

Check host names

Check ttRepAdmin -showconfig

Use ttRepAdmin -showconfig

to confirm the configuration of your replication scheme.

What to look for:

Are all of the subscriber agents started and reported to be in the

Start

state? If not, reset the agents to the

Start

state. See "Setting the replication state of subscribers" in the

Oracle TimesTen In-Memory Database TimesTen to TimesTen

Replication Guide

.

Do the reported Peer names match the names given in the DataStore attributes in the DSN definitions for the replicated data stores? Replication does not work if you specified the names given for the Data Source Name attributes.

Is there anything under List of subscribers? If not, confirm the data store names you specified in the DSN definition are consistent with those you specified in your replication scheme configuration file.

Are the Host names correct? If in doubt, see "Check host names" on page 6-7.

Are the correct table names displayed under Table details? If not, correct the table names in your replication scheme configuration file.

Example 6–6 Confirm the configuration of the replication scheme

> ttRepAdmin -showconfig masterDSN

Self host "MYHOST", port auto, name "MASTERDS", LSN 4/2970276, timeout 120, threshold 0

List of subscribers

-----------------

Peer name

----------------

SUBSCRIBERDS

Host name

MYHOST

Last Msg Sent Last Msg Recv Latency TPS

Port

------------------------ ------

Auto

State Proto

------- -----

Start

RecordsPS Logs

10

------------- ------------- ------- ------- --------- ----

0:01:12 19.41 5 52 2

List of tables and subscriptions

--------------------------------

6-6

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Replication does not work

Table details

-------------

Table : REPL.TAB

Master Name

-----------

MASTERDS

Subscriber Name

-------------

SUBSCRIBERDS

Check the TTREP.TTSTORES table

Check the

TTREP.TTSTORES

table to confirm that replication associates the replication scheme with the local data store.

Example 6–7 Confirm that the replication scheme is associated with the local data store

Connect to the data store and enter:

SELECT * FROM ttrep.ttstores WHERE is_local_store <> 0x0;

Command> select * from ttrep.ttstores where is_local_store <> 0x0;

< -5193371075573733683, MYHOST, MASTERDS, 01, 0, 0, 4, 0 >

1 row found.

There should be exactly one row returned. If more than one row is returned, contact

Technical support . If no rows are returned, then none of the hosts returned by the

following statement is perceived to be a local system by TimesTen replication:

SELECT DISTINCT

host_name

FROM ttrep.ttstores;

It may also be that none of the data store names specified in your replication scheme match those specified in your DSN descriptions.

Check host names

Some hosts or IP addresses specified in a replication scheme cannot be resolved by the replication agent because:

Host names or IP addresses specified in the replication scheme are wrong or misspelled.

Host names or IP addresses cannot be resolved or found by DNS or in the

/etc/hosts

file

Entries in the

/etc/hosts file are incorrectly ordered in appearance. This error is most common when multiple NICs are used. You must have root privilege to make changes to the

/etc/hosts files.

See "Configuring host IP addresses" in the

Oracle TimesTen In-Memory Database

TimesTen to TimesTen Replication Guide

for details on how to configure DNS and

/etc/hosts files for host machines used for replication.

To check if a host name in the replication scheme matches the host name of the local machine, write an application to perform these tasks:

1.

Use a gethostname

OS function call to determine the host name of the running host.

2.

3.

4.

Call gethostbyname

with the output from Step 1.

Call gethostbyname

with the host name specified in the replication scheme.

Compare output of Step 2 and Step 3. If there is a match, then the running host is involved in replication. Otherwise, it is not involved in replication.

Troubleshooting Replication

6-7

Replication does not work

Check owner names

As described in "Table requirements and restrictions" and "Owner of the replication scheme and tables" in the

Oracle TimesTen In-Memory Database TimesTen to TimesTen

Replication Guide

, the owner names of your replication scheme and your replicated tables must be consistent across all participating data stores.

Checking replication owner

Check the owner name assigned to your replication scheme by calling the ttIsql repschemes

command or by listing the contents of the

TTREP.REPLICATIONS

table.

Example 6–8

shows that the replication scheme name,

REPSCHEME

, has a consistent owner name (

REPL

) in the data stores on both

SYSTEM1

and

SYSTEM2

.

Example 6–9

shows the scheme name with inconsistent owner names. This can occur if you omit the owner name from the replication scheme definition and the system uses the Id of the replication scheme creator.

Example 6–8 Consistent owner names for replication scheme

On

SYSTEM1

:

> ttIsql masterDSN

Command> select * from ttrep.replications;

< REPSCHEME

1 row found.

, REPL , C, 0, 0, -1 >

On

SYSTEM2

:

> ttIsql -connStr "dsn=subscriberDSN"

Command> select * from ttrep.replications;

< REPSCHEME

1 row found.

, REPL , C, 0, 0, -1 >

Example 6–9 Inconsistent owner names for replication scheme

On

SYSTEM1

:

> ttIsql masterDSN

Command> select * from ttrep.replications;

< REPSCHEME

1 row found.

, SYSTEM1 , C, 0, 0, -1 >

On

SYSTEM2

:

> ttIsql -connStr "dsn=subscriberDSN"

Command> select * from ttrep.replications;

< REPSCHEME , SYSTEM2

1 row found.

, C, 0, 0, -1 >

Checking table owner

Check the owner names assigned to the tables in each data store by using the ttIsql tables

command.

Example 6–10 Consistent table owner names

This example shows that the

TAB

table has a consistent owner name (

REPL

) in the data stores on both

SYSTEM1

and

SYSTEM2

.

6-8

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Replication does not work

Output for SYSTEM1

SYS.CACHE_GROUP

SYS.COLUMNS

SYS.COL_STATS

SYS.INDEXES

SYS.MONITOR

SYS.PLAN

SYS.TABLES

SYS.TBL_STATS

SYS.TRANSACTION_LOG_API

REPL.TAB

TTREP.REPELEMENTS

TTREP.REPLICATIONS

TTREP.REPPEERS

TTREP.REPSTORES

TTREP.REPSUBSCRIPTIONS

TTREP.REPTABLES

TTREP.TTSTORES

Output for SYSTEM2

SYS.CACHE_GROUP

SYS.COLUMNS

SYS.COL_STATS

SYS.INDEXES

SYS.MONITOR

SYS.PLAN

SYS.TABLES

SYS.TBL_STATS

SYS.TRANSACTION_LOG_API

REPL.TAB

TTREP.REPELEMENTS

TTREP.REPLICATIONS

TTREP.REPPEERS

TTREP.REPSTORES

TTREP.REPSUBSCRIPTIONS

TTREP.REPTABLES

TTREP.TTSTORES

Example 6–11 Inconsistent table owner names

This example shows the

TAB

table with inconsistent owner names, which were automatically assigned for each host.

Output for SYSTEM1

SYS.CACHE_GROUP

SYS.COLUMNS

SYS.COL_STATS

SYS.INDEXES

SYS.MONITOR

SYS.PLAN

SYS.TABLES

SYS.TBL_STATS

SYS.TRANSACTION_LOG_API

SYSTEM1.TAB

TTREP.REPELEMENTS

TTREP.REPLICATIONS

TTREP.REPPEERS

TTREP.REPSTORES

TTREP.REPSUBSCRIPTIONS

Output for SYSTEM2

SYS.CACHE_GROUP

SYS.COLUMNS

SYS.COL_STATS

SYS.INDEXES

SYS.MONITOR

SYS.PLAN

SYS.TABLES

SYS.TBL_STATS

SYS.TRANSACTION_LOG_API

SYSTEM2.TAB

TTREP.REPELEMENTS

TTREP.REPLICATIONS

TTREP.REPPEERS

TTREP.REPSTORES

TTREP.REPSUBSCRIPTIONS

Troubleshooting Replication

6-9

Replication unresponsive, appears hung

Output for SYSTEM1

TTREP.REPTABLES

TTREP.TTSTORES

Output for SYSTEM2

TTREP.REPTABLES

TTREP.TTSTORES

Check consistency between replicated tables

Replicated tables on both master and subscriber data stores must be exactly the same.

Example 6–12 Check consistency between replicated tables

This output from the user error log shows a mismatch on the number of columns for the subscriber table

TTUSER.MYDSN

.

11:37:58.00 Info: REP: 9430: REP1:transmitter.c(4936): TT16136: Sending definition for table TTUSER.MYDSN (1 columns)

11:37:58.00 Info: REP: 9412: REP2:receiver.c(5928): TT16193: Adding definition for table: TTUSER.MYDSN

11:37:58.00 Info: REP: 9412: REP2:meta.c(5580):TTUSER.MYDSN ptn 0: srcoff 0, destoff 0, length 8

11:37:58.00 Info: REP: 9412: REP2:meta.c(5580):TTUSER.MYDSN ptn 1: srcoff 8, destoff 12, length 12

11:37:58.00 Err : REP: 9412: REP2:receiver.c(6203): TT16198: Table definition mismatch on number of columns for table TTUSER.MYDSN. Local definition: 2; transmitting peer: 1

11:37:58.00 Err : REP: 9412: REP2:receiver.c(6380): TT16204: Table TTUSER.MYDSN marked invalid. Will not apply transactions received for it until a valid definition is received

11:37:58.00 Err : REP: 9412: REP2:receiver.c(7200): TT16078: Table definition for ID 637068 is invalid (Original failure 11:37:58

REP2:receiver.c(6203): TT16198: Table definition mismatch on number of columns for table TTUSER.MYDSN. Local definition: 2; transmitting peer: 1)

11:37:58.00 Err : REP: 9412: REP2:receiver.c(5002): TT16187: Transaction

1173958671/2; Error: transient 0, permanent 1

Replication unresponsive, appears hung

Table summary is in the first heading cell.

Possible cause

Failed subscriber

Return-receipt timeout period too long

See...

"Check replication state" on page 6-10

"Check return receipt timeout setting" on page 6-11

Check replication state

Use the ttReplicationStatus

procedure to check state of the subscriber data store with respect to its master. If the subscriber is in the

Failed

state, see "Managing data store failover and recovery" in the

Oracle TimesTen In-Memory Database TimesTen to

TimesTen Replication Guide

for information on how to deal with failed data stores.

Example 6–13 Check replication state

Use ttReplicationStatus

to obtain the status of the subscriberds

data store from its master data store, masterDSN

, enter:

> ttIsql masterDSN

6-10

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Poor replication or XLA performance

Command> CALL ttReplicationStatus ('subscriberds');

< SUBSCRIBERDS, MYHOST, 0, failed, 1, 10, REPSCHEME, REPL >

1 row found.

Check return receipt timeout setting

Use the ttRepSyncGet

procedure to check the return receipt timeout setting. A value of -1 indicates the application is to wait until it receives an acknowledgement from the subscriber. Network latency or other issues might delay receipt of the subscriber acknowledgment. You either address these issues or use the ttRepSyncGet procedure to reset the return receipt timeout period.

See "Checking the status of return service transactions" in the

Oracle TimesTen

In-Memory Database TimesTen to TimesTen Replication Guide

for more information.

Poor replication or XLA performance

Most of this section addresses issues that may impact replication performance. Some issues, such as log buffer too small and reading from the transaction log files on disk, can impact the performance of both replication and XLA applications.

Possible cause

Slow network

Using RETURN RECEIPT

Inefficient replication scheme

Log buffer too small

Frequent or inefficient disk writes

Reading from transaction log files on disk rather than the log buffer

High rate of conflicts

See...

"Check network bandwidth" on page 6-11

"Check use of return receipt blocking" on page 11

"Check replication configuration" on page 12

"Check size of log buffer" on page 6-12

"Check durability settings" on page 6-12

"Check for reads from transaction log files" on page 6-12

"Conflict reporting slows down replication" on page 6-16

Check network bandwidth

Replication agents typically communicate over some type of network connection. If replicating over a network slower than 10 MB per second (such as common with a 100

Base-T Ethernet typical in a LAN), you must be careful to match the transaction load to the available bandwidth of the network. see "Network bandwidth requirements" in the

Oracle TimesTen In-Memory Database TimesTen to TimesTen Replication Guide

for details.

Check use of return receipt blocking

Unless you need receipt confirmation for all your transactions, disable RETURN

RECEIPT BLOCKING. If you require receipt confirmation for some transactions, then set RETURN RECEIPT BY REQUEST and call the ttRepSyncSet

procedure to enable the return receipt service for specific transactions. See "RETURN RECEIPT BY

REQUEST" in the

Oracle TimesTen In-Memory Database TimesTen to TimesTen Replication

Guide

for details.

Troubleshooting Replication

6-11

Poor replication or XLA performance

Note:

The performance degradation caused by return-receipt becomes less of an issue when multiple applications (or threads) are updating the data store. If you must use return-receipt in a transaction, you can improve the performance of your application by using multiple threads to update the data store. Though each thread must block for receipt confirmation, the other threads are free to make updates.

Check replication configuration

In addition to return-receipt setting described above, other factors related to the configuration of your replication scheme could impact replication performance. As described in "Making decisions about performance and recovery trade-offs" in the

Oracle TimesTen In-Memory Database TimesTen to TimesTen Replication Guide

, you often have to weigh the ability to efficiently failover and recover a data store against replication performance.

For more information about direct replication, see "Direct replication or propagation" in the

Oracle TimesTen In-Memory Database TimesTen to TimesTen Replication Guide

.

Check size of log buffer

Setting your log buffer too small may impact replication performance. Instead, Set the

LogBufMB

DSN attribute to a larger size. For more information on this DSN attribute, see "Setting attributes for logging" in the

Oracle TimesTen In-Memory Database TimesTen to TimesTen Replication Guide

Check durability settings

You can improve replication performance by setting TRANSMIT NONDURABLE on the replication ELEMENT to eliminate the flush-log-to-disk operation from the replication cycle. See "Setting transmit durability on data store elements" in the

Oracle

TimesTen In-Memory Database TimesTen to TimesTen Replication Guide

for details.

Enabling the DURABLE COMMIT option in your replication scheme also impacts performance. See "DURABLE COMMIT" in the

Oracle TimesTen In-Memory Database

TimesTen to TimesTen Replication Guide

for more information.

Check for reads from transaction log files

In some situations a "log reader," such as a master replication agent 'transmitter' thread or a ttXlaNextUpdate

call in an XLA application, may not be able to keep up with the update rate of the applications writing to the data store. Normally, replication and XLA readers get update records from the log buffer in memory. When the readers fall behind the application update rate, transaction log files can accumulate on the disk until the backlog can be cleared. This forces the readers to read transactions from the transaction log files on disk, which is much slower. Should you detect reads from the transaction log files, you may want to respond by decreasing the rate of application updates to that sustainable by the log readers.

Applications can monitor whether log readers are obtaining update records from transaction log files on disk rather than from the log buffer in memory by tracking the

SYS.MONITOR table entry LOG_FS_READS. For example, you can check the value of

LOG_FS_READS for the data store,

MASTERDSN

, with the following ttIsql command:

6-12

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Poor replication or XLA performance

% ttIsql -v1 -e "select log_fs_reads from monitor; quit;" -connStr dsn=MASTERDSN

If the LOG_FS_READS counter is increasing, the log readers are falling behind or clearing out a backlog in the transaction log files.

For more complete monitoring of replication progress, create a simple shell script like the following:

!/bin/sh trap exit 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

DSN=$1 while [ 1 ] ; do

date

ttRepAdmin -receiver -list -connStr dsn=$DSN

echo -n "Log reads from disk: "

ttIsql -v1 -e "select log_fs_reads from monitor; quit;" -connStr dsn=$DSN

echo

ttRepAdmin -bookmark -connStr dsn=$DSN sleep 15 done

Example 6–14 Check the status of the transaction log

For example, you name the above script monitorLog

and your replication scheme replicates from the

MASTERDSN

data store to the

SUBSCRIBER1DSN

data store. You can then check the status of the transaction log by entering:

$ monitorLog masterdsn

This generates output similar to the following:

Mon Aug

Peer name

2 10:44:40 2004

Host name Port

--------------------------------------- ------

SUBSCRIBER1DSN MYHOST Auto

State Proto

------- -----

Last Msg Sent Last Msg Recv Latency TPS RecordsPS Logs

------------- ------------- ------- ------- --------- ----

00:00:05 -1.00

-1 -1 1

Log reads from disk: < 0 >

Replication hold LSN ...... 10/2656136

Last written LSN .......... 10/4015824

Last LSN forced to disk ... 10/3970152

The output from the script displays an updated status every 15 seconds until you enter

Ctrl-C to exit.

Following the date in the output in Example 6–14

is the name of the subscriber, its host, and so on. Next is latency and rate information, as well as the number of transaction log files being retained on behalf of this subscriber. The specific meaning of each value is described in "Using ttRepAdmin to display subscriber list" in the

Oracle

TimesTen In-Memory Database TimesTen to TimesTen Replication Guide

. The main interest here is the 'Last Msg Sent' and 'Logs' values. The 'Last Msg Sent' value indicates the elapsed time since the last message was sent by the master to the subscriber and 'Logs' indicates how many transaction log files behind the replication log reader is from the current log insertion point used by the writers (Last written LSN).

Troubleshooting Replication

6-13

Problems using ttRepAdmin

Normally the 'Logs' value should be '1', as shown in

Example 6–14 . A steadily

increasing 'Logs' value indicates latency is increasing and eventually log reads are satisfied from disk.

Note:

If the LogBufMB is larger than the LogFileSize, an increase in the 'Logs' value does not necessarily mean the log readers are reading from the transaction log files. This is because the log manager does not allow more than one log file's worth of data to be outstanding before writing it to the file system. After the log manager writes the data, the data remains in the log buffer to be read directly by the log readers.

So, when the LogBufMB is larger than the LogFileSize, the 'Logs" value alone may not be the best measure of whether log readers are reading from memory or from disk.

The output from: ttRepAdmin -bookmark -connStr dsn=$DSN displays the number of the transaction log files and the location of the bookmarks set by the log manager, as described in "From the command line: ttRepAdmin -bookmark" in the

Oracle TimesTen In-Memory Database TimesTen to TimesTen Replication Guide

. The difference between the Replication hold LSN and the last written LSN indicates the number of records in the transaction log that have not yet been transmitted to the subscribers. A steady increase in the difference between these values is another indication that replication latency is increasing and log file reads are likely to occur.

Example 6–15 Log reader must read from transaction log files

In this example, assume the

LogBufMB

is 16MB and the

LogFileSize

is 8MB. The following output indicates the log reader is approximately 1.8 MB behind the capacity of the log buffer and must read from the transaction log files, 14 and 15.

Peer name Host name Port State Proto

---------------- ------------------------ ------ ------- -----

SUBSCRIBER1DSN MYHOST

Last Msg Sent Last Msg Recv Latency TPS RecordsPS Logs

------------- ------------- ------- ------- --------- ----

00:00:03 - -1.00 -1

Log reads from disk: <20>

Replication hold LSN ...... 14/7007464

Last written LSN .......... 17/465336

Last LSN forced to disk ... 17/456152

Problems using ttRepAdmin

This section includes the following topics:

Problems using ttRepAdmin -duplicate

Returns 'Must specify -scheme' error

6-14

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

Problems with conflict checking

Problems using ttRepAdmin -duplicate

If you connected to your new subscriber DSN before running ttRepAdmin

-duplicate

, the data store has already been created. In this situation,

-duplicate returns:

Error : Restore not done : The datastore already exists.

Unable to restore datastore locally

Confirm the existence of the data store by running ttStatus

and checking to see if the data store is in the returned list. If the new subscriber data store exists, destroy it and try ttRepAdmin -duplicate

again:

> ttDestroy /tmp/newstore

> ttRepAdmin -dsn newstoreDSN -duplicate -name newstore

-from masterds -host "server1"

If you have made an error entering the subscriber data store name or host name in the replication scheme, you may see something like the following:

Unable to swap datastore locally

No receiver NEWSTORE on SERVER2 found to swap with

Returns 'Must specify -scheme' error

If you have more than one scheme specified in your

TTREP.REPLICATIONS

table, some ttRepAdmin

commands may return the error:

Must specify -scheme to identify which replication scheme to use

To check the names of the replication schemes used by your data store, use the ttIsql

utility to connect, and enter:

Command> SELECT * from TTREP.REPLICATIONS;

Example 6–16 Two replication schemes assigned to the data store

This example shows that two replication schemes,

REPSCHEME1

and

REPSCHEME2

, are assigned to the data store associated with subDSN

. In this case, it is necessary to use the ttRepAdmin -scheme

option.

> ttIsql -connStr "dsn=subDSN"

Command> SELECT * from TTREP.REPLICATIONS;

< REPSCHEME1 , REPL , C, 0, 0, -1 >

, REPL , C, 0, 0, -1 > < REPSCHEME2

2 rows found.

Command> exit

> ttRepAdmin -dsn subDSN -receiver -list -scheme REPSCHEME1

Peer name

----------------

SUBSCRIBER1

Host name

------------------------ ------

MYHOST

Port

Auto

State

Start

Proto

------- -----

10

Last Msg Sent Last Msg Recv Latency TPS RecordsPS Logs

------------- ------------- ------- ------- --------- ----

0:01:12 19.41 5 52 2

Problems with conflict checking

This section includes the following topics:

Column cannot be used for replication timestamp

Troubleshooting Replication

6-15

Problems with conflict checking

Timestamp does not exist

Conflict reporting slows down replication

Column cannot be used for replication timestamp

When attempting to set CHECK CONFLICTS for an element in a CREATE

REPLICATION statement, you may encounter an error similar to the following:

8004: Column REPL.TABS.TS cannot be used for replication timestamp checking if in an index or added by ALTER TABLE; and must be binary(8) with NULL values allowed.

In this situation, check:

That the timestamp column in the specified table is a nullable column of type

BINARY(8). In the above example, the TS column in the

REPL.TAB

table should have a type of BINARY(8).

The timestamp column is defined in the original CREATE TABLE statement, rather than added later using ALTER TABLE.

Timestamp does not exist

You may receive an error similar to the following:

2208: Column TS does not exist in table.

In this situation, confirm that you have specified the correct name for the timestamp

COLUMN in the CHECK CONFLICTS clause and that it exists in the specified table.

Also, make sure the timestamp column is not part of a primary key or index.

Conflict reporting slows down replication

If you have configured replication to check conflicts, TimesTen sends reports to the local host. You can also configure a report file. See "Diagnostics through SNMP Traps" in the

Oracle TimesTen In-Memory Database Error Messages and SNMP Traps

.

If there is a large number of conflicts in a short period of time, subscriber performance can slow down because of the reporting requirements. You can use store attributes in the CREATE REPLICATION or ALTER REPLICATION statements to suspend and resume conflict reporting at specified rates of conflict:

CONFLICT REPORTING SUSPEND AT

rate

CONFLICT REPORTING RESUME AT

rate

Information about conflicts that occur while reporting is suspended cannot be retrieved.

See "Reporting conflicts" in the

Oracle TimesTen In-Memory Database TimesTen to

TimesTen Replication Guide

.

6-16

Oracle TimesTen In-Memory Database Troubleshooting Procedures Guide

A

AgentTimeout parameter, 4-23

aging

monitoring, 1-14

ALTER REPLICATION

troubleshooting, 6-2

asynchronous writethrough cache group, see AWT cache group

autorefresh

diagnose, 4-22

Failed status, 4-14

fails, 4-9

full, 4-19 incremental, 4-19

interval too small, 4-28

monitoring

change log tables, 4-13

SQL script, 4-13 support log, 4-13 monitoring the change log tables, 4-13

performance, 4-22, 4-23, 4-26, 4-28

setting FREELISTS, 4-26 setting INITRANS, 4-26

trace, 1-16

AUTOREFRESH trace, 4-22

AWT cache group

dropped object, 5-4

error file, 5-3

operation failure, 4-9

Oracle errors, 5-3, 5-4

replication problems, 5-1

awterr file, 5-3

B

base table

performance, 4-27

buffer busy waits, 4-26

C

cache trigger

performance, 4-15

cache agent

Index

problems starting, 4-2

cache grid

attach, 4-3 detach, 4-3 recover, 4-3 unexpected failure, 4-3

cache group

autorefresh fails, 4-9

AWT operation failure, 4-9

DDL tracking, 4-9 load failure, 4-9

loading fails, 4-10

performance, 4-27

recover, 4-23

refreshing fails, 4-10

cache groups

performance, 4-15

cacheInfo.sql script, 4-13

change log record

size, 4-20 change log table, 4-20

CHECK CONFLICTS

timestamp problems, 6-16

client/server

out of space, 2-9

problems, 2-6

thread stack overflow, 2-8

concurrent connections

maximum for server, 2-7

conflict reporting

suspending, 6-16

connect identifier, 4-4

connection

maximum concurrent server, 2-7

connection status, 1-2

connection timeout

Oracle client/server versions, 4-5

connections

maximum number on Windows XP, 2-8

create cache group

null constraint, 4-9 unsupported data type mapping, 4-9

CREATE SESSION privilege, 2-4

Index-1

D

daemon

support log, 1-6 user error log, 1-6

database

dead, 4-23 recover, 4-23

datablock

availability, 4-15 dbms_shared_pool.keep procedure, 4-15

DDL

tracking, 4-9

DeadDbRecovery parameter, 4-23

DEADLOCK tracing, 1-10

deadlocks, 4-26

downgrade

version 7 to version 6, 3-1

E

error tracing, 1-13

F

failures

server, 2-7

FREELISTS

setting, 4-26

H

hanging application, 2-13

I

IMDB cache database

recover, 4-23 dead database, 4-23

performance, 4-15

INITRANS

setting, 4-26

ipcrm command, 2-5 ipcs command, 2-5

isolation

levels, 2-12

L

LOAD CACHE GROUP

failure, 4-10

lock

levels, 2-12

timeout, 1-19

tracing, 1-12

LockLevel attribute, 2-12

log buffer

replication, 6-12

log table

performance, 4-27

Index-2

LogBufMB first connection attribute, 6-12

M

memory shared

consumption, 2-16

used by query, 2-18

multiple connections

thread stack overflow, 2-8

multithreaded applications

conflicts, 2-14

O

OCI initialization failure, 4-8

ODBC tracing, 1-20

optimizer

update table statistics, 2-11

ORA-12154, 4-4

Oracle client/server interoperability, 4-5

Oracle errors

AWT cache group, 5-4

AWT cache groups, 5-3

Oracle tablespace, 4-19

ORACLE_HOME

changing, 4-3

out of space

client/server multiple connections, 2-9

P

parallel query

enable, 4-15

performance

autorefresh cache group, 4-26 buffy busy waits, 4-26 excessive deadlocks, 4-26

IMDB cache, 4-15

isolation levels, 2-12

large tables, 4-27

lock levels, 2-12

refresh, 4-26 row lock waits, 4-26

PermSize attribute, 2-5

problems

checking connection status, 1-2

client/server, 2-6

finding tables, 2-15

hanging application, 2-13

transaction log files accumulating, 2-19

ps command, 2-5

Q

query

checking memory used, 2-18

R

read committed isolation level

SELECT gives duplicate results, 2-20

REFRESH CACHE GROUP

failure, 4-10

replication agent

unable to stop or start, 6-2

log buffer, 6-12

performance

conflict reporting, 6-16

poor performance, 6-11

troubleshooting, 5-2, 6-1, 6-3

rollbacks

tracing, 1-10

row lock waits, 4-26

S

sar reporting tool, 2-16 sar tool, 2-16

SELECT statement

duplicate results, 2-20

semaphore

removal, 2-5

server

failures, 2-7 maximum concurrent connections, 2-7

SGA datablock

availability, 4-15 keep portion, 4-15

shared memory

estimate size, 2-5 maximum size reached, 2-5 query limits, 2-5 remove segment, 2-5

shared segment

too many in use, 2-3 unable to create, 2-3

SNMP traps, 1-20

replication, 6-3

space

monitoring, 1-10

SQL tracing, 1-9

statement preparation

importance of, 2-12

support log

monitoring autorefresh, 4-13

Oracle errors, 5-4

SYS tables, 1-20

system tables

monitoring, 1-20

troubleshooting contention, 2-14

T

table

update statistics, 2-11

tables

abnormally large, 4-27

large, 4-15

tablespace

cache administration user, 4-19 create, 4-19

diagnose, 4-20

full, 0-xiv, 4-20, 4-21

managing, 0-xiv, 4-21 monitor, 0-xiv, 4-21

Oracle, 4-19

recovery, 0-xiv, 4-21

set threshold warning, 0-xiv, 4-21

TblSpaceFullRecovery parameter, 4-21

TblspaceThreshold parameter, 4-21

TempSize attribute, 2-5

thread stack overflow

multiple client connections, 2-8

TimesTen daemon

support log, 1-6 user error log, 1-6

tnsnames.ora identifier, 4-4

top reporting tool

vmstat tool, 2-16

trace

output format, 1-8

tracing

AGING trace, 1-14

API trace, 1-10

AUTOREFRESH trace, 1-16

deadlocks, 1-10

error tracing, 1-13

how to turn on, 1-20

LOCK trace, 1-12

ODBC trace, 1-20

SQL trace, 1-9

transaction

log buffer, 6-12

log files

accumulation, 2-19

triggers

performance, 4-15

troubleshooting, 4-1

AWT dropped object, 5-4

client/server, 2-6

connection status, 1-2

finding tables, 2-15

hanging application, 2-13

low on space, 1-10

ODBC trace, 1-20

replication, 6-1

server failures, 2-7

transaction log files accumulating, 2-19

ttCacheAutorefreshStatsGet procedure, 4-22

Failed status, 4-14

ttCacheConfig built-in procedure, 4-21, 4-22, 4-23

ttCapture utility, 1-5

ttGridAttach utility, 4-3 ttGridDetachList utility, 4-3

ttIsql utility

using, 1-1

ttLockLevel procedure, 2-12

ttmodinstall utility

Index-3

using, 4-3

ttOptSetFlag procedure, 2-12

ttOptUpdateStats procedure, 2-11

ttRepAdmin utility

troubleshooting, 6-14

ttRepStart utility, 4-3

ttSize utility

estimate memory, 2-5

ttStatus utility

using, 1-2

ttTraceMon utility

AGING tracing, 1-14

API tracing, 1-10

AUTOREFRESH tracing, 1-16

DEADLOCK tracing, 1-10

error tracing, 1-13

LOCK tracing, 1-12

output format, 1-8

SQL tracing, 1-9

using, 1-6

ttXactAdmin utility

using for troubleshooting, 1-19

U

ulimit command, 2-5

UNIC

ps command, 2-5

UNIX

ipcrm command, 2-5 ipcs command, 2-5 ulimit command, 2-5

V

vmstat reporting tool, 2-16

X

XLA

poor performance, 6-11

Index-4

Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement

Table of contents