January 5, 201313 yr I've been trying to get things going for the past few weeks with my unraid build. I'm using 5.0 rc8a. The build is: ASUS M2N32-SLI w/5400 x2 4GB DDR2 800 Corsair RAM (memtested and passed, but memtest reports it running at 900 instead of 800) Hard drives: 6 1TB caviar blacks, and 5 2TB greens (EARS unjumpered) HBA is an IBM M1015 crossflashed to latest IT firmware I happened to have the M2N32-SLI laying around doing nothing (I actually have two of these motherboards) My goal was to get a pair of M1015's running out of the dual pci-x 16 slots, but only one will work in the lower 16x slot and neither will work in the primary 16x slot. This has forced me to use the motherboard SATA connectors, which as far as I know should be ok. I connected 3 1TB blacks to the M1015, and the 2TB parity drive (new out of package, but didn't run preclear) to the motherboard SATA connectors. I tried copying about 250-300GB of multimedia files a few times, in the unraid web manager the parity drive racked up about 200-340 errors each time. I thought this was pretty bad, so I moved the parity drive to the M1015. The errors went away on the next test copy. I didn't run a parity check at the time (looking back I should have). Next, I connected a 1TB black to the motherboard SATA port where the parity drive was connected, I copied about 500GB of data to that drive to see what would happen, no errors. I powered off and forgot to shut the array down, and next powerup it ran a parity check: Total size: 2 TB Current position: 21.31 GB (1%) Estimated speed: 35.36 MB/sec Estimated finish: 933 minutes Sync errors corrected: 5170742 Is that a high number of sync errors? I'm guessing I should rebuild the array and keep testing, as well as preclear the parity drive. Should I be looking for a new motherboard anyway at this point? I'd really like to be able to use the boards I have laying around but not if there's a risk of corruption or errors.
February 2, 201313 yr Author 1 is too many errors. Run a memetest overnight. I'm scratching my head on this one. I ran memtest for 24 hours nothing, I re-ran it just with test #5 for 24hours and nothing. I rebuilt the array to test it using the same 1TB disks I originally did for data but put in a 1TB black for the parity drive (instead of the 2TB green). I can't seem to reproduce the errors after writing over 1TB of data to the array. I'm gunshy now of putting all my eggs onto this thing for fear of having it turn around, run a parity check and everything gets corrupted.
February 2, 201313 yr 1 is too many errors. Run a memetest overnight. I'm scratching my head on this one. I ran memtest for 24 hours nothing, I re-ran it just with test #5 for 24hours and nothing. I rebuilt the array to test it using the same 1TB disks I originally did for data but put in a 1TB black for the parity drive (instead of the 2TB green). I can't seem to reproduce the errors after writing over 1TB of data to the array. I'm gunshy now of putting all my eggs onto this thing for fear of having it turn around, run a parity check and everything gets corrupted. You need to attached a syslog for anyone to be of much more help to you. If you don't how, read the first post in this thread: http://lime-technology.com/forum/index.php?topic=9880.0 I would also suggest that you run a non-correcting parity check as if there are errors that they will be detected. Also, if there be should errors found in a non-correcting parity check, a second test can be run to see if the errors replicate themselves.
February 2, 201313 yr Author 1 is too many errors. Run a memetest overnight. I'm scratching my head on this one. I ran memtest for 24 hours nothing, I re-ran it just with test #5 for 24hours and nothing. I rebuilt the array to test it using the same 1TB disks I originally did for data but put in a 1TB black for the parity drive (instead of the 2TB green). I can't seem to reproduce the errors after writing over 1TB of data to the array. I'm gunshy now of putting all my eggs onto this thing for fear of having it turn around, run a parity check and everything gets corrupted. You need to attached a syslog for anyone to be of much more help to you. If you don't how, read the first post in this thread: http://lime-technology.com/forum/index.php?topic=9880.0 I would also suggest that you run a non-correcting parity check as if there are errors that they will be detected. Also, if there be should errors found in a non-correcting parity check, a second test can be run to see if the errors replicate themselves. Ok thanks, Would a syslog from now matter or do I need the one from when the millions of sync errors occurred? I was kind of wondering in more general terms if its better to just run everything off of HBA cards and avoid plugging into the motherboards (especially older ones). I thought that slew of errors might be an nforce driver issue or something like that.
February 2, 201313 yr Have you run a non-correcting parity check and not gotten any errors??? Have you done it a second time??? Of course, one with errors is what is needed. The experts who answer the most difficult problems on this board are not clairvoyant. I realize you are a new user and have some doubts after your initial experience. But there was something wrong in the beginning. You said you swapped out a HD and PERHAPS that was the problem. You need to confirm that all is well. To give yourself some peace of mind and confidence in both unRAID and your hardware. If you don't know how to run a non-correcting parity check, here are the simple instructions: Go to the main page of the Web GUI. Under the 'Array Status' section of the page, look for the check box that is labeled 'Write corrections to parity disk'. Uncheck the box. Click on the 'Check' button. It will take about from four to eight hours to run the test. I usually do it over night and it is finished in the morning. (Typically takes seven and quarter hours for my 3TB wide array.) If you don't get any errors after two checks, everything is working properly. There are ways to test that HD to see if it has issues...
February 6, 201313 yr Author Have you run a non-correcting parity check and not gotten any errors??? Have you done it a second time??? Of course, one with errors is what is needed. The experts who answer the most difficult problems on this board are not clairvoyant. I realize you are a new user and have some doubts after your initial experience. But there was something wrong in the beginning. You said you swapped out a HD and PERHAPS that was the problem. You need to confirm that all is well. To give yourself some peace of mind and confidence in both unRAID and your hardware. If you don't know how to run a non-correcting parity check, here are the simple instructions: Go to the main page of the Web GUI. Under the 'Array Status' section of the page, look for the check box that is labeled 'Write corrections to parity disk'. Uncheck the box. Click on the 'Check' button. It will take about from four to eight hours to run the test. I usually do it over night and it is finished in the morning. (Typically takes seven and quarter hours for my 3TB wide array.) If you don't get any errors after two checks, everything is working properly. There are ways to test that HD to see if it has issues... I took the original array apart, never obtained the syslog just took note of the errors first pass went something like this (correction was on): Total size: 2 TB Current position: 21.31 GB (1%) Estimated speed: 35.36 MB/sec Estimated finish: 933 minutes Sync errors corrected: 5170742 stopped the parity check, the array. powered off waited a bit, powered back on. restarted array and ran non-correcting parity check up unitl it hit 5% and it went something like this: Total size: 2 TB Current position: 95.58 GB (5%) Estimated speed: 53.16 MB/sec Estimated finish: 597 minutes Sync errors detected: 17169748 I don't have the syslogs, but I know to get them for next time. I rebuilt the array with the 3x1TB caviar blacks, copied over about 1TB of data, ran parity check twice and couldn't reproduce any errors. Right now I'm just building the array with my 5 2TB greens, copied over 2TB so far and no errors. If all goes well, I'll add the 6x1TB blacks and see what happens. I'm changing the motherboard/cpu/ram either way. I just found the amount of errors pretty unsettling for a first-run through.
February 7, 201313 yr I would add in those 1TB Black drives one at a time. After you are sure that each new drive is working properly, add the next one. That way if you encounter a problem, you can be reasonably assured that the problem with the recent addition. (If it were me, I would pre-clear each one of those six drives with a triple cycle before committing it to my array. But then I am super cautious... )
Archived
This topic is now archived and is closed to further replies.