May 21, 201016 yr Hi, I have just started using unraid (started with 4.5.3 and now 4.5.4) over the last couple of weeks and am seeing a problem that I am hoping to get help with. It seems that some of my files are getting corrupted! I have run the mem test for 24 hours with out it reporting any problems. I have done a parity check which has been ok and a reiserfsk --check which also reported ok. When copying over files to the unraid box I do a compare to make sure that the files get copied over completely. I now also do a md5sum after they verify ok so that I can compare the copied files with the original at a later time. I seems that after a few files are written to the unraid some of the files get corrupt (usually not the last files that were copied). Attached is a syslog which for some reason at startup only disk10 reports a read error. Disk 10 however does not currently show and file corruption (I have mainly been copying to disk11). Any suggestion would be most appreciated. syslog.txt.zip
May 21, 201016 yr I've copied about 500GB to two servers running 4.5.4, using Teracopy with Test mode on. No verify failures.
May 21, 201016 yr Author I don't get any verify errors when I copy the file over initially either (using muCommander in verify mode). In fact I also do an md5sum on the copied file and the original file and they check out identical. The issue comes sometime later after a few other copies / writes have been done to the disk, then all of a sudden a file that copied over ok and checksum'd ok no longer does. Files always verify OK when they are initially copied over.
May 21, 201016 yr I don't get any verify errors when I copy the file over initially either (using muCommander in verify mode). In fact I also do an md5sum on the copied file and the original file and they check out identical. The issue comes sometime later after a few other copies / writes have been done to the disk, then all of a sudden a file that copied over ok and checksum'd ok no longer does. Files always verify OK when they are initially copied over. Then you have a different HARDWARE error of some kind. It might be power supply related, it might be memory related, it might be network related, it could even be disk drive related. If writing one file corrupts a different file, you might want to run reiserfsck file system checks on your disks.
May 21, 201016 yr Author Hi Joe, As noted in my first post reiserfsck reports back no errors and I have run the memory test for 24 hours with no reported errors. I am in the process of checking each drive with the WD Data Lifeguard Diagnostic CD. Currently disks6-11 all check out fine using the full scan, and disk 5 is currently halfway through with no errors. If these all check out, my power supply will be the next on the list to check. It is a SeaSonic M12II 620 Bronze 620W, that had been working without any problems as my old fileserver.
May 22, 201016 yr Author All 11 drives test fine with WD's Extended test... Power supply test next (temp swapping mine with a friends). If that checks out mother board test is next. I currently have a gigabyte GA-MA790XT-UD4P, if anyone else using this with unraid could you please post to let me know your results with it.. Thanks
May 25, 201016 yr Author Thought that the power supply replacement had licked this one... I was wrong. Corruption again tonight. Motherboard stress test next using Phoronix Test Suite Live.
May 27, 201016 yr Author Motherboard came out of the stress test fine. I am now replacing my drives with 3 old ones and rebuilding a unraid on those for further testing.
June 12, 201016 yr Author Just a quick update. Put in the three old drives. After multiple copies, deletes and renames of files corruption occurred (on a file that was fairly static). Did a parity check after finding the corrupt file and the parity check completed with no errors. Moved to ubuntu on the same hardware and did not have any problems. Reverted back to unraid to see if i could pinpoint exact steps that trigger my issue but it has been working flawlessly so far (about 1 week now with heavy stress testing). I am really hesitant about reverting back the original 9 Tb array with all my data.
Archived
This topic is now archived and is closed to further replies.