April 18, 201115 yr Hey guys, i have had a lot of trouble so far getting this unRAID server going. To make it short, im currently running a parity check and am 0.5% in with 2564056 sync errors. Im not sure how i should proceed from here. Do i let it finish, correcting all the errors and then run another to make sure all errors have been corrected? Or should i stop this and start over (for the 2nd time)?
April 18, 201115 yr A little more information and perhaps a syslog might help people troubleshoot. Look at the pinned posts for pointers on what information is useful and on getting a syslog posted. Specifically this post >click<
April 19, 201115 yr Author Ok. Some history on what happened. I originally setup the array without setting MBR: 4k-aligned. When i noticed transfer speed performance problems and did some searching, i realized my mistake and wanted to correct it before putting anymore data on the array. I searched the forums and found a post saying to use dd if=/dev/zero count=8 of=/dev/sdX on each disk and then initconfig to start fresh and reformat the drives. This worked fine and when i started the array, all the drives said unformatted so i checked the box and formatted. When that was done, it ran the initial parity check. I thought there may be an issue during this because disk1 was getting read at over 100MB/sec while disk2 was at less than 40. I let this process finish and when it was done, i noticed that the read I/O on disk1 was still over 100MB/sec even after all activity on disk2 and the parity drive had stopped. So i stopped the array, shutdown the server and restarted. When it booted back up, it once again said that disk1 was unformatted. This was weird, but i checked the box and let it format again since there was no data on it yet, so i wasnt worried about losing anything. When it was complete. The array showed that parity was in sync and everything looked good. So i copied some data over last night and when i got home from work today i decided to do a parity check with data on the array so that i could make sure there were no issues anymore. This is where i am now. Sitting at 15.5% done with the parity check and currently at 46,777,623 sync errors. Im hoping that everything was seriously off from my formatting issues and that once this sync completes and i do another, there will be no errors. I just wanted to see what others thought. I am attaching the syslog which shows parity errors at the end. It looks like it stopped logging them 2 hours ago, not far into the check. Thanks for any input anyone may have. syslog.txt
April 19, 201115 yr In theory, you should just let it finish and then do another check and see what happens. But, disk1 coming back unformatted just doesn't sound good and I wouldn't trust it without more testing. I personally would stop the parity check, unassign that drive, do an initconfig and let the parity build without it. With disk1 out of the array, I would run a preclear (or 2) on it to see if it acts up any further. If the preclear is OK and I can't see any problems at all, then it will only take minutes to assign it back to the array and begin to use it. Peter
April 19, 201115 yr Author Ok, so I let that parity-check finish, there were over 200mil corrections. As soon as it finished i started it again and went to work. When i got home, it had finished with 0 errors. I moved some data to both disks 1 and 2 as well as to the main Media share i setup. Then i moved the data i copied to each individual disk between the two drives, to try and create different scenarios of data moving around between the two drives. Then i stopped the array and restarted the server. The array was started and looked good upon restart and i started another parity check. Assuming this check completes with no errors, i think everything is ok.
April 19, 201115 yr Did you ever pull SMART reports from the drive? I'd at least check that just in case something appears there. I didn't see anything bad in your log but from your description of events the log from before the reboot probably would have shown something bad going on. Peter
Archived
This topic is now archived and is closed to further replies.