Jump to content

rgliebe

Members
  • Content count

    12
  • Joined

  • Last visited

Community Reputation

0 Neutral

About rgliebe

  • Rank
    Occasional Forum Poster
  1. I have confirmed a problem with Retrospect 5.1.175 with the SCSI update installed on a PowerMac G4 Gigabit Ethernet that has a 1.3 GHz G4 upgrade. I use a Compaq DDS-4 4mm DAT tape drive on a SCSI channel with a lot of hard drives. The problem did not show up immediately when I swapped out a troublesome Adaptec 39160 (their OS X driver support is weak) for an Atto UL3D. I am running OS X 10.2.8 with all patches and the latest Atto driver (2.1.0) and firmware (1.6.6). I did incremental backups (about 1 GB at a time) for weeks without any problems. However, when I did a full backup at the beginning of this year (about 45 GB), I stumbled into this problem on the second tape and was able to repeat it again after skipping to a third tape. The problem is that Retrospect gets frozen trying to write to the tape drive, which is also hung up in an endless loop of attempting to write. The only solution is to force quit Retrospect and reboot to reset the SCSI bus. Only the tape drive is unaddressable on the SCSI bus until it is power cycled on reboot. The hard drives are just fine. For some reason, each time after rebooting, a catalog update was necessary for the ENTIRE contents of the tape, not just the area where the write problem occurred. When the catalog update process reaches the bad point on the tape, it hangs up and freezes Retrospect, putting the tape drive into an endless read loop. This time the update process can be cancelled within Retrospect, which then reports back that there is a media read error due to bad media, dirty head, etc. (the -206 error). Without human intervention, Retrospect will stay hung up trying to address the tape drive forever. The biggest problem is that nothing else can be written to the tape again, and one must skip onto new blank media to continue the storageset. To see if this problem was unique to this particular configuration, or all Atto cards in general, I moved the tape drive to a Beige G3 Server running OS X Server 10.2.8 with an Atto PSC card in it. The same firmware release and driver were installed, which are the latest. The results so far have been good, with nearly two full tapes written to without any problems. So, I believe that the SCSI update does not fully address all the problems with at least one Atto card and needs to be looked at again for another update. This problem is hard to reproduce and could take many hours of writing to tape to observe. Retrospect also needs to be more robust when encountering read/write problems and not freeze up. It should stop and report the appropriate error after a reasonable timeout period occurs.
  2. A new problem showed up as soon as I upgraded the backup server from 5.0.205 to 5.0.536. I began to experience frequent problems when doing operations with the ATL Powerstor L200 tape library. Erasing tapes and moving tapes in and out of the drive from the Devices window suddenly became hazardous. There is now about a 20% chance that the operation will hang Retrospect and require a force quit to get out of it. No other changes were made, and this problem has never occurred before with an older version of Retrospect that supports this library. Maybe a change to the Retrospect Driver file has created this problem.
  3. rgliebe

    OS X Client Hangs Backup Server

    The hanging problem was resolved by upgrading the clients to 5.0.540. Other problems, like the Defer button not working on the client countdown, still remain. There also seems to be a memory issue on the backup server that comes up about once a week, usually while trying to back up an OS X client. Retrospect starts acting like the server only has 100 MB of RAM and refuses to back up the volumes with over 50,000 files until I reboot the backup server. So it looks like I traded one very annoying problem for a less annoying problem by doing the client upgrade.
  4. I have a similar problem with a Sun Ultra 5 running Solaris. I checked to make sure that everything that is required by Retrospect is properly installed. I cannot figure out why the client remains reserved and in use after the backup finishes. The only way to free it up is kill the process and restart it. Until Dantz addresses these types of problems and problems with full restores, I have to continue using Legato to back up our UNIX based servers.
  5. rgliebe

    Memory Issues?

    I am seeing the same problem backing up OS X clients and Windows clients from a server with 512 MB of real RAM and over 1 GB of virtual memory. It appears even the 6.0 release is not able to use virtual memory properly, which was a problem with previous Windows releases, but has never been a problem with the Mac releases. As soon as it gets even slightly close to using the real RAM minus what the OS (Windows 2000) is using, it starts giving out these sorts of error messages. It appears to be worse in the 6.0 release than it was in the 5.6 release, since I have seen about 200 MB of real RAM go to waste when it logs a "not enough application memory" error message. At least 5.6 waited to log this error until it really had run out of real RAM. This must be fixed by Dantz, since not every backup server is going to be stuffed with several GB of RAM, and UNIX and Windows systems typically have over 100,000 files on a single volume nowdays. I cannot upgrade the backup server that backs up our servers until this problem is fixed, since we have very large volumes with up to 1 million files on several of our servers.
  6. I have finally been able to confirm the cause of a problem that was killing my Windows based Retrospect 6.0 backup server. Not only does doing a threshold speed test cause problems, but so does having a client countdown, irregardless of how the user responds to it (if at all). Having either of these two options selected in the backup script causes the backup server to hang indefinitely on a OS X 10.2.x client, without logging an error message, until someone intervenes. The hanging occurs at the very beginning, even before the first volume is scanned for the files that are on it. Unfortunately, this problem is not consistent, but it does happen most of the time. I am using the latest Retrospect client for OS X, version 5.0.536, and the Windows backup server version is 6.0.206.
  7. I figured out what the problem was yesterday. Dantz does have a tech note out on this problem in the knowledgebase at http://www.dantz.com/index.php3?SCREEN=knowledgebase_article&id=835, but it is listed by the Sony mechanisms that have this problem (SDT-10000 and SDT-11000). The problem basically revolves around the driver Adaptec supplies for the 39160. I had to downgrade to the driver that Apple supplies for the 39160 with OS X, which is slightly older. The other person who started this thread is probably also having a similar problem with his SCSI card driver. For some reason these Sony mechanisms seem to have special problems with some OS X SCSI card drivers.
  8. I figured out what the problem was yesterday. Dantz does have a tech note out on this problem in the knowledgebase at http://www.dantz.com/index.php3?SCREEN=knowledgebase_article&id=835, but it is listed by the Sony mechanisms that have this problem (SDT-10000 and SDT-11000). The problem basically revolves around the driver Adaptec supplies for the 39160. I had to downgrade to the driver that Apple supplies for the 39160 with OS X, which is slightly older. The other person who started this thread is probably also having a similar problem with his SCSI card driver. For some reason these Sony mechanisms seem to have special problems with some OS X SCSI card drivers.
  9. I have had the same problem with this drive. It worked great with Retrospect 4.3, but starting with the OS X preview, the drive would fail with bad header errors during verification (which means the tape was not been written to properly). I have the same problem with the release version of Retrospect 5.0, including rev 203. I am running 10.1.3 with all the latest patches. I have used several SCSI cards, and the current one is an Adaptec 39160. Dantz needs to address this problem ASAP. It has nothing to do with the tape drive or the SCSI card. It is either the OS or Retrospect that needs to be fixed, and I vote for the latter, since it used to work great with Retrospect 4.3. Right now the 5.0 upgrade is a big waste of money for me, since I cannot use it yet to back up my video files (over 2 GB in size). By the way, Retrospect 5.0 does work fine under OS X with other SCSI tape loaders I have at work, which use DLT7000 drives.
×