cofin Posted November 3, 2012 Share Posted November 3, 2012 Hey all, Long time user and lurker here. I have had several unraid builds, and I finally decided to build a monster server with Norco 24 bay chassis. It is stacked out with 3TB drives and with the exception of parity, is functioning perfectly. I am on 5.0 rc8a, and I have tried multiple 3tb drives (I originally thought one of the 24 drives I purchased was bad) and I get the exact same number of errors (384) when attempting to sync to the parity disc. I've never had these types of issues in the past, so I'm a bit stumped on how to resolve. I am handy with the command line and use linux on a daily basis, so hopefully I'll be able to follow any suggestions. I've attached my syslog file, and any help would be greatly appreciated. EDIT: I reseated all SAS cables and swapped one of the SAS2LP cards into a free PCIX slot. Rebuilt parity in maintenance mode and it failed again at same spot. I've attached the output of the SMART report for the drive. EDIT: After disabling the PCI-E option ROMS and the drive spindowns, I was able to successfully sync parity. Hopefully this has resolved my isses. syslog.zip smartreport.txt Quote Link to comment
dikkiedirk Posted November 3, 2012 Share Posted November 3, 2012 Can you give some more input on system component? I can be a flaky drive, cable, backplane, cabling..........and many more. Quote Link to comment
cofin Posted November 3, 2012 Author Share Posted November 3, 2012 Sure thing, here are the current specs: I3 2100 Supermicro MBD-X9SCM-F-O 4 x KVR1333D3E9S/4G RAM 3 x AOC-SAS2LP-MV8 Corsair 750AX PSU 24 x Western Digital WD30EZRX Norco RPC-4224 Chassis All temps in the chassis are within normal range and all other drives appear to function normally. Quote Link to comment
dgaschk Posted November 3, 2012 Share Posted November 3, 2012 Attach a SMART report for the parity disk. Quote Link to comment
cofin Posted November 3, 2012 Author Share Posted November 3, 2012 Thanks, for helping with this. I've attached the SMART report to my original posting. Quote Link to comment
dgaschk Posted November 3, 2012 Share Posted November 3, 2012 Power down and check the cabling to the parity disk. When restarted parity will need to be rebuilt. Quote Link to comment
cofin Posted November 3, 2012 Author Share Posted November 3, 2012 I did that this morning. Reseated all add-on cards and SAS/power cables. Any other ideas? Quote Link to comment
cofin Posted November 4, 2012 Author Share Posted November 4, 2012 Just to give an update on this: I disabled all option ROMs on PCI-X cards in the BIOS. Disabled all disk spindowns in unraid. And attempted parity build again. I've gotten further than ever, and it's been running for about 15 hours. If it all goes well, it should complete successfully in ~80 minutes. Quote Link to comment
Glimmerman911 Posted January 2, 2013 Share Posted January 2, 2013 I believe I am having the same issue as you. I can read, write, and rebuild drives no problem, but on parity sync (maintenance mode or not), one of the (random?) disks on my AOC-SAS2LP-MV8 disconnects and red-balls. The server loses connection to the drive, I have to reboot for the server to see the disc again, then I can rebuild it. I assume the option to disable PCI-E optional ROMs in somewhere in my motherboard BIOS? Can you confirm which firmware version your AOC-SAS2LP-MV8 is running? Have you ever upgraded it? Thank you! Quote Link to comment
warhead Posted January 3, 2013 Share Posted January 3, 2013 Could it be a power supply issue? Parity sync requires all drives to be spun up Quote Link to comment
Glimmerman911 Posted January 3, 2013 Share Posted January 3, 2013 I thought it might have been as well, so I pulled out my 2 existing PSU units, and replacement them with a single Corsair 750w Professional series. More than enough for my server plus 15 green drives. Quote Link to comment
cofin Posted January 3, 2013 Author Share Posted January 3, 2013 I believe that the kernel support for these cards (or using multiples of these cards) is flaky. I swapped to 3 M1015 cards and all of my issues were instantly resolved. Faster writes and no parity rebuilding issues. Quote Link to comment
Glimmerman911 Posted January 3, 2013 Share Posted January 3, 2013 Any idea where I can get an M1015 in Canada, I cannot find any. *Edit* I did find one on tigrdirect.ca for $360, but at that point I would be better buying synology. Quote Link to comment
cofin Posted January 3, 2013 Author Share Posted January 3, 2013 I bought mine used on ebay. They were extremely expensive new. I got 3 for less than that price you mentioned. Quote Link to comment
Glimmerman911 Posted January 3, 2013 Share Posted January 3, 2013 Ebay it is! I just ordered 2 from an ebay seller out of South Korea, and he shipped them out already, they should be here in about 5 days. Quote Link to comment
cofin Posted January 3, 2013 Author Share Posted January 3, 2013 Be sure to follow the flashing guides in the other forum. I was able to flash all 3 with the other bios in about 30 minutes. Quote Link to comment
Glimmerman911 Posted January 3, 2013 Share Posted January 3, 2013 Could I trouble you to post the link for me, to ensure I follow the same instructions you did? Quote Link to comment
cofin Posted January 3, 2013 Author Share Posted January 3, 2013 Sure, no problem at all. Here it is: http://lime-technology.com/forum/index.php?topic=12767.msg121131#msg121131 Quote Link to comment
Glimmerman911 Posted January 3, 2013 Share Posted January 3, 2013 Thank you cofin. While I wait for my m1015 to arrive, I would like to continue testing the supermicro card. How did you disable the PCI-E option ROMS, I'm not sure what this means, and if this change could get my server to parity sync I would be ecstatic. Quote Link to comment
cofin Posted January 3, 2013 Author Share Posted January 3, 2013 The option ROM is a setting in the motherboard BIOS. Depending on what mobo you have, it could be in different places. Just poke around and see if you see something like that. I don't think that it will make a huge difference though. You might want to try setting this at the shell and see if it makes a difference: sysctl vm.highmem_is_dirtyable=1 This made an impact in my performance. Quote Link to comment
Glimmerman911 Posted January 3, 2013 Share Posted January 3, 2013 Ok, I looked through my BIOS and disabled anything resembling that already, was hoping I was missing something. I have read about that command from Tom, that it can greatly improve write performance, but there was some warnings about unknown effects? Quote Link to comment
Glimmerman911 Posted January 10, 2013 Share Posted January 10, 2013 Finally found the problem, my supermicro controller didn't like the PCI-E slot it was in. As soon as I switched it, parity check has completed successfully 3 times in a row now. My replacement supermicro arrived, so I am now testing that controller in the PCI-E slot to see if it was just a problem with that particular controller in that slot, or if the slot is bad and I need to RMA the motherboard. So far 60% complete a parity check completed without a dropped disk. Quote Link to comment
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.
Note: Your post will require moderator approval before it will be visible.