How to resolve SCSI bus timeouts

Problem

A SCSI bus timeout occurs when either the SCSI controller or the SCSI device detects that a command has not been responded to within a specific timeout period. Normally, commands are issued by the SCSI controller, which then waits for a response from the device being addressed. If no response is detected, the controller may reset the bus to reinitiate communication. While disk drives normally recover well from SCSI bus timeouts and resets, tape drives have much greater difficulty in doing so.
 

Cause

To try to detect problems such as these, the architects of the SCSI bus, and Backup Exec (tm) programmers have defined time limits for possible operations being performed by tape drives attached to the SCSI bus. Normally, these timeout values are large enough to accommodate any delay, including retries, in executing the specific command being attempted. In the event that the device does not respond in the amount of time indicated by the timeout value, a timeout occurs.

Usually, the first indication of a timeout is that the tape drive ceases activity. However, because some operations can take several hours on some tape drives, some timeout values are very long, which can make it seem as if the program itself has stopped responding, or that there is some kind of application error.
Given below are some suggested changes and checks to help resolve the SCSI timeout issue.
 

Solution

• Set the SCSI ID of the tape drive as close as possible to the SCSI ID of the HBA controller/interface. Example: If the HBA controller ID is 7 then the tape drive should have 6 or 8, or any available ID which is closest to 7, if 6 and 8 are unavailable.
• Avoid 0 or 1 for SCSI ID's, these are usually reserved for bootable devices
• Set INITIATE WIDE NEGOTIATION to off for controllers with a wide bus
• Set INITIATE SYNC NEGOTIATION for the ID of the tape drive to off
• Set MAXIMUM SYNC TRANSFER RATE to the slowest possible setting
• DISCONNECT must be enabled for all tape drive SCSI IDs
• Set SCSI PARITY CHECKING to off

• Check with the controller and tape drive vendors for firmware and driver updates.
• Use good quality cabling that follows the SCSI specification.
• For external cables, use heavy duty shielded cables.
• Under no circumstances should an external SCSI cable be detached from any connection while power is still applied to any device on the bus.
• For single-ended internal cables, make sure there is at least one foot of cable between devices.
• Also, for single-ended buses, make sure the overall cable length from termination to termination does not exceed the maximum bus length of three meters (roughly 9 to 10 feet).
• Use active termination on both ends of the SCSI bus.
• The last external device (if any) on the SCSI chain is terminated.
• The last internal device (if any)) on the SCSI chain is terminated.
• The internal termination of the SCSI host adapter is not set to Automatic, but to the appropriate setting for the hardware configuration (Refer to the Host Adapter documentation for the correct settings).
• All external devices should be set to supply termination power.
• Put the tape devices on a separate controller by itself or change the existing controller.
• Replace the tape drive.

• "Device I/O" errors.
• "Bad Block" errors.