Blog: Backup

When working with HP Ultrium tape drives, it is sometimes a challenge to get them to run at speeds even close to the advertised maximum. Sometimes there are questions as to how to troubleshoot and how know you are getting the most out of the drive. Here are some tips: [more]

• Most HP modern Ultrium drives have a hardware compression feature. In order to get even close to the advertised speed of the drive, you must ensure that the hardware compression is enabled on the drive. You can do this with the HP Library and Tape Tools utility. Without it, you will get approximately half of the advertised speed.
• If you are using Backup Exec with an Ultrium drive, use the Symantec drivers that are provided on their website and make sure the drive you are using is on the HCL for the version of BE you are using. Only switch to the HP/OEM driver if you experience issues with media robotics when moving/inventorying media.
• If you are using a tape library or autoloader, use the HP/OEM driver for the media changer/robotics and the Symantec driver for the drive. Symantec does not make drivers for the robotics.
• Make sure that you are not using a SCSI channel that is connected to a RAID card. This can seriously impact performance of the drive. Best possible scenario is an Ultra320 SCSI card on a dedicated 64-bit PCI bus. Make sure the SCSI card drivers are up to date.
• There are two types of autoloaders/libraries: LUN based and SCSI ID based. The difference is the way the device presents itself to the OS. LUN based devices share a SCSI ID and present different LUNs to the OS for the media changer and the drive. SCSI ID devices present two different SCSI IDs . Backup Exec works best on SCSI ID devices, but either will work. If the device you are working on allows you to choose LUN based or SCSI based, choose SCSI based.
• Make sure the SCSI bus of your device chain is properly terminated. If using LVD SCSI, make sure you are using an LVD/SE terminator. Do not put more than two SCSI devices on the same SCSI chain as the tape device.
• When testing performance, do a backup to local disk to test performance. If speeds are significantly faster, you should further troubleshoot the tape device because data CAN be backed up faster; the bottleneck is the tape device. If speeds are about the same, the bottleneck is the BE remote agent. Data cannot be presented to the device fast enough to keep it busy. In this scenario, troubleshoot the hosts for performance increases.
• On the Backup Exec media server, disable the "Removable Storage" service. It has been known to cause issues with Backup Exec since version 8 and up.

We used to run bemcmd91.exe with a long list of parameters to export and import Backup Exec jobs and settings.  With version 10, there is now a gui, BEUtility.exe. Read http://seer.support.veritas.com/docs/274778.htm.  This document says "BEUtility will not copy media sets that are not currently assigned to a job". I assumed this means that it will copy media sets that were assigned to a job, but it does not. It also says it will copy the alert configuration, but it does not. You need to go through all the settings and check them.

When using Backup Exec 10.x, you may experience problems backing up SQL Server 2000 or MSDE databases if the file path contains double backslashes in the file path.

The error within Backup Exec is:  [more]

Final error: 0xe0008492 - Database Query Failure. See the job log for details.
Final error category: Resource Errors

To fix the issue, you must detach and reattach each database that contains the double backslash in the file path. The issue is caused by including a trailing slash in the “New Database Default Location” when declaring a file location.

If you do include a trailing slash (which will be included by default if you use the browse button) and try to create new databases using T-SQL (i.e. CREATE DATABASE …), your file paths will end up with the double backslash notation. Creating databases with the Enterprise Manager GUI does not produce the double backslash no matter if the “New Database Default Location” includes a trailing backslash or not. So, make sure the trailing backslashes are removed from the file path.

The recovery partition on ThinkPads installs in two different sizes depending on how you install it.  If you revert to factory settings, the recovery partition should be 4-5 GB.  If you restore the system from a Rescue and Recovery backup, the recovery partition will only be about 500MB.  The down side is that Rescue and Recovery installs/upgrades write files to the recovery partition.  The 500MB partition fills up very quickly and prevents you from upgrading/installing Rescue and Recovery.  Your two options at that point are 1) research Rescue and Recovery to see if there is an install parameters that prevents files from being written to the recovery partition (in the interest of time, I have not done this), or 2) remove the recovery partition and then install Rescue and Recovery.

Following daylight savings time change, some backup applications will modify the scheduled backup job start time to preserve a 24-hour time gap between the current and previous day’s job. In particular, CoNetrix has determined Veritas’ Backup Exec version 9.x and greater will push the scheduled backup job back one hour from the original start time. For example, a backup job typically scheduled to start at 10:00pm on Monday will start at 11:00pm on Monday. This may cause conflicts with overlapping scheduled jobs. Veritas states that the problem should automatically correct itself following the first run after the time change. Users should check their scheduled backup jobs to ensure the start time for each job is correct.