STORAGE

Applying a Modular, Layered Approach to

Troubleshooting Tape Drive Hardware

In storage environments, tape drives and libraries can be cost-effective backup and restore devices, but they may incur support costs for troubleshooting and maintenance.

This article introduces a modular, layered approach that can ease system administration of tape devices by methodically troubleshooting problems to determine root causes and resolve issues.

BY GAJANAN MUDALIAR

Related Categories:

Dell PowerVault storage

Linux

Microsoft Windows

Novell NetWare

Storage

Tape library

Troubleshooting

Visit www.dell.com/powersolutions for the complete category index to all articles published in this issue.

I n the typical enterprise data center, tape drives, tape libraries, and other tape automation devices can be cost-effective workhorses for offline data storage. However, the total cost of ownership for the many tape devices attached to a data center’s servers can escalate when support and service costs are factored into the equation.

Given these considerations, administrators must take a proactive approach to troubleshooting, determining root causes, and correcting problems.

A basic system for troubleshooting tape drive devices can be categorized in a four-layer stack, similar to a TCP/IP stack. As shown in Figure 1, the predominant layers include the hardware layer, the BIOS layer, the OS layer, and the application layer.

Although the hardware layer and the BIOS layer can be combined into a single layer, various troubleshooting techniques can be carried out independently on each of these layers. This article provides details and troubleshooting techniques for the four layers.

Troubleshooting at the hardware layer

The scope of the hardware layer extends from the tape drive to the SCSI cable connectivity of a SCSI controller.

Loose cable connections can cause problems at this layer.

Administrators can perform the following troubleshooting procedures to diagnose tape device problems at the hardware layer:

• Check for improper SCSI termination, which can lead to SCSI errors during backups or to the tape

Application layer

OS layer

BIOS layer

Hardware layer

Figure 1. Layered model for modular approach to troubleshooting

42 DELL POWER SOLUTIONS Reprinted from Dell Power Solutions, May 2005. Copyright © 2005 Dell Inc. All rights reserved .

May 2005

STORAGE

•

• device not being detected on the BIOS level.

Check that the SCSI installation is properly configured— terminators must exist at both ends of the bus. Most host bus adapters (HBAs) are terminated by default, so be sure that the last device in the chain is also terminated.

1

Determine how the SCSI device is terminated. Some SCSI devices are internally terminated, which can be disabled through a jumper or a dual in-line package (DIP) switch, while others use an external terminator plug.

Tape drives and libraries usually perform a power-on self-test

(POST) operation, in which the system initializes itself and initiates an internal self-test to check for hardware issues. Some tape drives and tape libraries have a display panel, which provides a hex code that corresponds to the hardware error that occurred within the system. The hex code may indicate a self-diagnostic mode or a particular display of blinking or colored lights. Consult the device documentation for more information.

Troubleshooting at the BIOS layer

If a tape drive is not detected on the BIOS level of a controller, the problem usually resides in the controller or the tape drive. Administrators can perform the following troubleshooting procedures to determine why a tape drive is not being detected: mine whether the SCSI ID is causing the problem. For example, the tape drive’s SCSI ID could conflict with a device on the same SCSI bus, which usually occurs when the SCSI ID of the tape drive is 7—the default SCSI ID of the HBA (SCSI controller). When attaching a tape drive with a SCSI ID above 7 on a narrow SCSI channel, change the SCSI ID to a value below 7.

devices on the same SCSI bus is not responding, it should be separated from the SCSI bus. whether the controller is supported. Commonly used HBAs in

Dell ™ servers include the Adaptec SCSI Card 39160 and SCSI

Card 29160 and the LSI Logic LSI53C1030 embedded SCSI controller.

a tape drive is connected to a narrow SCSI channel controller, the narrow SCSI channel can support only seven SCSI IDs.

that the SCSI Disconnection and Parity settings are enabled.

Synchronous Negotiation should be disabled. The SCSI transfer rate should be set to 5 Mbps.

2 If hard drives are being duplexed, both SCSI cards should be configured identically.

and not Level Triggering. Level Triggering involves the use of shared interrupts, which can cause problems in some systems.

Troubleshooting at the OS layer

If the tape drive is detected on the BIOS level, it should also be detected on the OS level. To facilitate device detection, administrators should proactively check with hardware vendors for

A basic system for

updated device drivers.

troubleshooting tape drive

Microsoft Windows OS

The Microsoft ® Management

devices can be categorized

Console (MMC) provides a user interface through which

in a four-layer stack. The

administrators can coordinate systems management

predominant layers include

for Microsoft Windows ® environments. In the Computer

the hardware layer, the

Management>Device Manager menu, administrators can

BIOS layer, the OS layer, and

view all hardware components within the Windows-

the application layer.

based environment. From this menu, they can check whether a tape device is listed. The “?” icon beside a tape device indicates that the device driver is not loaded for that particular device. Device drivers usually can be downloaded from the vendor’s Web site or from the CD provided by the vendor.

Novell NetWare OS

System administrators managing systems within Novell ® NetWare ® environments can troubleshoot tape hardware from the commandline interface. To check whether a tape drive is listed in NetWare, administrators should issue the following command: list storage adapters

Sample output of the preceding command is shown in Figure 2.

1 If both internal and external devices are attached to one host adapter, it may be necessary to disable the termination on the adapter card if it is physically located in the middle of the bus.

2 Different host adapters will set these parameters in different ways. Some utilize jumpers or DIP switches, while others use a software-based configuration utility. Not all host adapters will be capable of adjusting all of these parameters.

www.dell.com/powersolutions Reprinted from Dell Power Solutions, May 2005. Copyright © 2005 Dell Inc. All rights reserved .

DELL 43

STORAGE

The following command will list only the devices attached to the system: list devices

Sample output of the preceding command is shown in Figure 3.

If the device cannot be viewed on the OS level, administrators should issue the following command to scan all the SCSI buses:

NW2600:list devices

0x0007: [V025-A0-D0:0] SAMSUNG CD-ROM SN-124 N104 [CD]

0x0008: [V024-A5-D1:0] Legacy Floppy [FLOPPY]

0x0009: [V321-A4-D1:0] IBM ULTRIUM-TD2 37RH [TAPE]

0x000B: [V358-A1-D0:0] FUJITSU MAP3367NC rev:5608 [HDD]

0x000C: [V358-A1-D1:0] FUJITSU MAP3367NC rev:5608 [HDD]

0x000D: [V358-A1-D2:0] FUJITSU MAP3367NC rev:5608 [HDD]

0x000E: [V358-A1-D6:0] Unbound Device Object

Figure 3. Sample output for Novell NetWare command to list devices attached to the system scan for new devices

In the device list output, the NetWare OS may not list the tape drive with its actual name or it may list the tape drive as an “unbound device object.” In such cases, a proper compatible driver should be loaded so that the output will reflect the inquiry name of the tape drive. The common device driver for all tape devices is nwtape.cdm. For Digital Linear Tape (DLT) tape devices, the common driver is dlttape.cdm.

If the SCSI card is enabled for logical storage units (LUNs), the changer and tape drive may be using the same SCSI ID but different LUNs. If so, administrators must enable LUN support. For example, if using an Adaptec SCSI card, the administrator would add LUN_ENABLE=FF to the line that is loading the driver for this card in startup.ncf:

Linux OS

In Linux, /proc/scsi/scsi is the file that shows the inquiry string for the tape drive. The /var/log/dmesg file will have an entry for the tape drive after the kudzu utility is executed and the device driver of the controller is loaded. If a tape library has a changer module and the tape drive and changer have the same SCSI ID but reside in different LUNs—which can occur on Dell PowerVault ™

122T autoloaders with DLT VS 80 technology—the Linux kernel will not update the /proc/scsi/scsi file with the inquiry string. In this case, system administrators should update the file with the following command: echo “scsi add-single-device 2 0 0 13” >

/proc/scsi/scsi

In the preceding command, the four numbers refer to the

HBA ID (2), the SCSI bus (0), the SCSI ID (0), and the LUN (13).

Devices in Linux are referenced by a

NW2600:list storage adapters

0x00 [V100-A100] USB UHCI Controller [slot 0]

0x01 [V025-A0] Novell ATA/IDE Host Adapter Module [slot 10008]

0x07 [V025-A0-D0:0] SAMSUNG CD-ROM SN-124 N104

0x02 [V358-A1] LSI_53C1030:10023 [slot 10023]

0x0E [V358-A1-D6:0] Unbound Device Object

0x0B [V358-A1-D0:0] FUJITSU MAP3367NC rev:5608

0x0C [V358-A1-D1:0] FUJITSU MAP3367NC rev:5608

0x0D [V358-A1-D2:0] FUJITSU MAP3367NC rev:5608

0x03 [V358-A2] LSI_53C1030:10024 [slot 10024]

0x04 [V321-A3] Adaptec SCSI Card 39160/3960D - Ultra160 SCSI [slot 201]

0x05 [V321-A4] Adaptec SCSI Card 39160/3960D - Ultra160 SCSI [slot 202]

0x0A [V321-A4-D4:0] DELL PV-132T 227D

0x09 [V321-A4-D1:0] IBM ULTRIUM-TD2 37RH

0x06 [V024-A5] Legacy FLOPPY Controller [slot 0]

0x08 [V024-A5-D1:0] Legacy Floppy device file. A device file can be a raw device file or a logical device file.

Because a tape drive is a sequential device, it can be referenced using a raw device path.

The Dell support Web site

(support.dell.com) provides numerous tools to help troubleshoot hardware-related tape device issues.

Administrators also can initiate a selfdiagnostic test to obtain an error code and messages on the front panel of the tape device. The user guide of the tape device can then provide additional information about the error code and troubleshooting steps to help resolve the issue.

Figure 2. Sample output for Novell NetWare command to list storage devices

44 DELL POWER SOLUTIONS Reprinted from Dell Power Solutions, May 2005. Copyright © 2005 Dell Inc. All rights reserved .

May 2005

STORAGE

Troubleshooting at the application layer

From the application layer, it is often difficult to drill down to the root cause of problems because most applications do not interact with the hardware directly. Application software generally communicates with the device drivers through the OS kernel, and the device drivers communicate with the firmware of the storage unit. Some applications may have native device drivers loaded to communicate with storage units. Administrators can perform the following troubleshooting procedures to diagnose tape device problems at the application layer: drive, which can prevent tape media from being populated to maximum capacity.

amount of debris has accumulated on the head. Media debris can prevent data from being written to a tape cartridge.

System administrators managing systems within

Novell NetWare environments can troubleshoot tape

with all the respective SCSI commands so that they can check the conditions and the trace communications over the SCSI bus.

Whenever SCSI errors or hardware errors occur in the backup device, the

hardware from the command-line interface.

target—the device that receives service requests for processing—sets a unit attention (UA) flag, which eventually follows a SCSI reset condition. Administrators can determine the cause of a SCSI bus reset by taking a SCSI trace of the communication path between the target and the initiator—a SCSI device containing application clients and SCSI ports that originate device service and task management requests to be processed by a

SCSI target device. Obtaining the sense data information from the

SCSI trace session can show which flags have been set and help administrators identify a clear path to problem resolution.

tion and whether the appropriate device drivers have been installed if the tape hardware is not detected on the application level but is detected on the OS level.

updated version of the tape device firmware is available if

SCSI errors occur while backing up large amounts of data.

3

Software vendors typically provide diagnostic utilities to aid in hardware troubleshooting. For example, VERITAS NetBackup software includes a utility through which administrators can initiate a self-diagnostic confidence test on the tape device.

Efficiently resolving tape hardware problems

Administrators can enhance the process of troubleshooting tape devices by following the modular, layered approach described in this article. To determine root causes rapidly, administrators should first categorize the condition as a hardware issue or a software issue, then determine the layer at which to begin troubleshooting. This method can help administrators find an efficient path to problem resolution.

Troubleshooting other general operational issues

Tape drives can break down during a backup or restore procedure.

In Windows, administrators would detect this error in a system event log; in Linux and

The Microsoft Management

NetWare, the error would appear on the system con-

Console provides a user

sole or logger screen.

In some cases, to deter-

interface through which

mine the root cause of a failure, administrators must take

administrators can coordinate

a SCSI trace and then analyze it on a SCSI analyzer. SCSI

systems management

analyzers show time diagrams, command listings,

for Microsoft Windows

and state listings. When using SCSI analyzers, admin-

environments.

istrators should be familiar

Gajanan Mudaliar is an engineering analyst on the tape storage engineering team in the Dell Product Group. His primary area of expertise is testing and troubleshooting tape hardware and automation devices. Prior to joining Dell, Gajanan had seven years of IT experience in technical support and system administration for enterprise storage installations. He has a B.E. in Electronics and Telecommunication from the University of

Kolhapur in India.

F O R M O R E I N F O R M AT I O N

Dell support: support.dell.com

VERITAS support: support.veritas.com

3 As a general safety rule, do not disconnect or power down the tape device during the firmware upgrade.

46 DELL POWER SOLUTIONS Reprinted from Dell Power Solutions, May 2005. Copyright © 2005 Dell Inc. All rights reserved .

May 2005