hoa Posted December 16, 2009 Report Share Posted December 16, 2009 We have used Retrospect for very long time with good results. The last two years we have used Retro 6.1 on an Intel xserve with two drives mirrored for files and system and the third for Retro-filebackup. Two backup sets swapped on a weekly basis, and the extra stored offsite. We have tried to maintain at least a year of history. Lately the other backup-disk got corrupted. I reformatted the drive, took a full backup, and all went well for a week or so, and I had to redo the formatting. It was always backup-set A that got corrupted, set-B performs fine. I took the problem disk to the local apple-store and they tested it and told me it was fine. Retrospect is probably doing something wrong when writing to the disk. I was told version 6 is an old program not written for intel, and they recommended Crashplan pro. Does anybody have an advice for me. I'm considering upgrading to Retro-8, and also evaluating Crashplan, but best would be to try to fix the problem with Retro.6.1 Kind regards hoa Quote Link to comment Share on other sites More sharing options...
Lennart_T Posted December 16, 2009 Report Share Posted December 16, 2009 Did the store (or you) do a surface scan (checking all the sectors) on the failing drive? When you reformatted the drive, did you check the option to zero out all data? (It would probably take many hours to format the drive that way, but it's much safer than just write a new (empty) disk directory in less than a minute.) Quote Link to comment Share on other sites More sharing options...
hoa Posted December 16, 2009 Author Report Share Posted December 16, 2009 Hej Lennart! Thank you for your cooperation. The store told me they had scanned the disk and checked all blocks, i had the impression it had taken some hours. ( and it was quite expensive ) The disk was formatted and empty when I used it, but i don't know how it was formatted. Do you think I should reformat and give it a try, for a third time? Hoa Quote Link to comment Share on other sites More sharing options...
twickland Posted December 16, 2009 Report Share Posted December 16, 2009 How did you determine that the volume had gotten corrupted? A Retrospect backup can give hard drives a real workout, and so can sometimes reveal a failing drive mechanism, weak power supply, etc., before other tests do so. You don't mention the brand of your drive. I do know that LaCie drives a few years ago were known for having flaky power supplies. If your unit has an outboard power supply brick, you might try a replacement. The fact that you're experiencing trouble with just one drive is pretty suggestive of a hardware problem. Big hard drives are pretty cheap these days, and at some point, it'd probably be worth your while to purchase a replacement rather than to spend a lot of time and effort in diagnosing the problem. Quote Link to comment Share on other sites More sharing options...
hoa Posted December 16, 2009 Author Report Share Posted December 16, 2009 (edited) Thanks twickland. Im logging in to xserver with ARD. When the problem first occurred, retrospect was operating and the remote desktop was slow and almost unresponsive. The backup disk was invisible on desktop, probably unmounted. When i tried to mount it with disk utility, I got some error message and an option to reformat the drive. The disk is original apple server disks (750GB SATA) in xserve drive bays. Edited December 16, 2009 by Guest Quote Link to comment Share on other sites More sharing options...
rhwalker Posted December 16, 2009 Report Share Posted December 16, 2009 Retrospect is probably doing something wrong when writing to the disk. Retrospect doesn't write to the disk; the Mac OS and Mac OS drivers do. There is nothing that Retrospect can do to destroy the formatting of your disk. File corruption is a different matter. What do the xServe's SMART indicators say (in Server Monitor)? I bet that the disk is failing. As for the machine being slow and unresponsive when Retrospect is running, that is normal. Takes a lot of effort by Retrospect to keep the I/O pipe filled. Russ Quote Link to comment Share on other sites More sharing options...
hoa Posted December 16, 2009 Author Report Share Posted December 16, 2009 What do the xServe's SMART indicators say (in Server Monitor)? Russ I cant check at the moment, but I´m quite certain Disk utility said: SMART status: Verified -and all green "lamps" in server monitor. Pre-failure Warning: No Warning Quote Link to comment Share on other sites More sharing options...
rhwalker Posted December 16, 2009 Report Share Posted December 16, 2009 There is missing information in your post. (1) what version of Mac OS X Server (10.x.x) is running on your xServe? (2) What version of Retrospect 6.1 (6.1.x) do you have? This should print in the Retrospect log each time Retrospect launches. (3) What version (6.x.x.x) of Retrospect Driver Update ("RDU") do you have? This should print in the Retrospect log each time Retrospect launches. Russ Quote Link to comment Share on other sites More sharing options...
hoa Posted December 16, 2009 Author Report Share Posted December 16, 2009 (1) MacOsX server 10.5.8 (9L34) (2) Retrospect version 6.1.230 (3) Retrospect Driver Update, version 6.1.16.100 Quote Link to comment Share on other sites More sharing options...
rhwalker Posted December 16, 2009 Report Share Posted December 16, 2009 Ok, then you have all of the right bits. You seem to have a failing drive (or perhaps even a memory issue, but that would show up as ECC errors in your console log). Or perhaps at some point you didn't have a clean shutdown of the server, and that messed up the bits on the drive. Russ Quote Link to comment Share on other sites More sharing options...
hoa Posted December 16, 2009 Author Report Share Posted December 16, 2009 We had a electrical failure some weeks ago, and it might have happened during backup. Can that be the reason for the problems with the drive, and could a new "deep" formatting clean the disc? (zero out all data) If so, it would be worth a try. Quote Link to comment Share on other sites More sharing options...
rhwalker Posted December 17, 2009 Report Share Posted December 17, 2009 Yes, yes. If the electrical failure happened during the backup, it could have corrupted the Retrospect files and might also have damaged the filesystem, causing it to wind about on itself. There are rare failure modes whereby the disk head can erase sector information too. You might consider getting an Uninterruptable Power Supply for the server. I wouldn't run a server without one. But you aren't talking "zero out all data"; you are talking about laying down new tracks and a new filesystem. I may not fully understand your configuration, because it seems that you are removing the sled (and drive) for the disk that you are calling your Retrospect backup, correct? If so, be sure that you unmount the volume before you take out the sled, because, if you pop the drive before you unmount the volume, that could cause the same bad results (corrupted files, corrupted filesystem). Sounds like a weekend project. Russ Quote Link to comment Share on other sites More sharing options...
hoa Posted December 17, 2009 Author Report Share Posted December 17, 2009 Thank you Russ, its great to have access to this forum and it's service minded people! Yes, I do remove the sled and disk after unmounting it. (always) I have 4 identical disks. Two containing the system and files, mirrored with SoftRaid. The other two are formatted with diskutility as regular apple drives used for backup, rotating on a weekly basis. I haven't experienced problems with the other three disks. What do you mean by " laying down new tracks and a new filesystem." Shouldn't it be enough to reformat the one problem drive? I live in Iceland and we have very stable electricity. But your probably right with the UPS. I will look into it. Kind regards hoa Quote Link to comment Share on other sites More sharing options...
rhwalker Posted December 17, 2009 Report Share Posted December 17, 2009 Two containing the system and files, mirrored with SoftRaid. Ok. Those two are fine, and SoftRaid always "does the right thing", and you know when it lets you split the mirror. We use SoftRaid on our xServe, too, for our system drive (one mirror is a high-speed SCSI attached to an ATTO card, one mirror is a RAID 5 array on an Apple Hardware RAID). It's one of the few pieces of software that I have never seen make a mistake. Excellent program, excellent software QC/QA, excellent support, very bright group of people. The other two are formatted with diskutility as regular apple drives used for backup, rotating on a weekly basis. I haven't experienced problems with the other three disks. Shouldn't be a problem there as long as you remember to unmount them before you remove the sleds. What do you mean by "laying down new tracks and a new filesystem." Shouldn't it be enough to reformat the one problem drive? Yes. The reason I made a comment was because you parenthetically remarked "zero all files". Disk Utility has a zero option that erases free space, that is not the same thing as reformatting. It can also just lay down a new filesystem, which only reconstructs the filesystem tree, and that is not the same thing as a reformat. Good luck. Stay warm. Russ Quote Link to comment Share on other sites More sharing options...
hoa Posted December 17, 2009 Author Report Share Posted December 17, 2009 I checked the disk today and its SMART status was "Verified" -and all green "lamps" in server monitor. Pre-failure Warning: No Warning I ran Disk utility and erased the disk again, Zero out data, it took some two hours. But after reading your latest post I think I should have chose partition and created a new partition. I will try that, and see if it will run smoothly for a new backup set. I have only good experience with SoftRaid too. Quick and helpful support and a simple and easy to understand interface. I always split my mirrored disks before applying upgrades to the server, checking that i have two bootable drives. If something goes wrong I have a copy of the servers state before the upgrade. If it goes well I can erase the second disk and add it to the mirror, and it rebuilds in a couple of hours. Quote Link to comment Share on other sites More sharing options...
rhwalker Posted December 17, 2009 Report Share Posted December 17, 2009 I have only good experience with SoftRaid too. Quick and helpful support and a simple and easy to understand interface. I always split my mirrored disks before applying upgrades to the server, checking that i have two bootable drives. If something goes wrong I have a copy of the servers state before the upgrade. If it goes well I can erase the second disk and add it to the mirror, and it rebuilds in a couple of hours. You and I have the same procedure, then. When the next version of SoftRaid is released, it will be scriptable, so that you will be able to replicate the server each night as a part of the mail, etc., checkpointing procedure. That, world peace, and a reliable Retrospect 8 (that can read backups made by earlier versions) will make my life complete. Russ Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.