January 18, 200917 yr I've put 2 used 400GB IDE drives in my unRaid recently to mix with 1TB SATA drives. I also upgraded to 4GB or RAM from 2GB (had it lying around) Over the past 2 days I've had to run the parity check twice, first time found some errors that I cleaned up by hand: duplicate files in certain places-deleted them files on one disk when they are supposed to be on another-moved them at cmd prompt now the first time I had to run the parity check it shot back with 2000+ errors. Checked syslog, no mention of anything in there. Running it again as it crapped out last night, 300 errors at last count, nothing in the syslog. tail /var/log/syslog shows nothing. Dmesg has nothing either. Am I missing something? Is there a way to have all the data sync up without casuing these errors? Where is it putting these error messages?
January 18, 200917 yr I've put 2 used 400GB IDE drives in my unRaid recently to mix with 1TB SATA drives. I also upgraded to 4GB or RAM from 2GB (had it lying around) Over the past 2 days I've had to run the parity check twice, first time found some errors that I cleaned up by hand: duplicate files in certain places-deleted them files on one disk when they are supposed to be on another-moved them at cmd prompt now the first time I had to run the parity check it shot back with 2000+ errors. Checked syslog, no mention of anything in there. Running it again as it crapped out last night, 300 errors at last count, nothing in the syslog. tail /var/log/syslog shows nothing. Dmesg has nothing either. Am I missing something? Is there a way to have all the data sync up without casuing these errors? Where is it putting these error messages? I would run a memory test overnight. It sounds as if the memory is not working as expected. Did you set its timings and voltage correctly? It could easily be the cause. Errors such as yours have been either memory or motherboard related. I can only remember one case where a disk drive did anything similar, and it resulted in lots of errors in the syslog. Joe L.
January 18, 200917 yr Author i'll run the memory though again. I'm fairly certain that i have 4 matche DIMMs in there. could be mistaken but i'll double check. This won't harm the intergrity of the raid or should I shut it down now and check it ASAP?
January 18, 200917 yr Author Just as I posted that I thought to myself "Is there really anything I'm doing on here that's more important then the array itself?" Within a few minutes I shut down the system, ripped out the old "new" ram and now running memtest. I tried to put in another pair of DIMMs but it wasn't liking that. Odd that it was only showing 3.4GB of RAM when I had at least 4GB in there. hmm could be the mother board not liking 4GB of RAM. That's odd though.
January 18, 200917 yr tail /var/log/syslog shows nothing. Dmesg has nothing either. Am I missing something? Is there a way to have all the data sync up without casuing these errors? Where is it putting these error messages? unRAID used to log locations of sync errors in the syslog. Not sure why Tom took that out, except that occasionally someone would run a parity check and would have so many sync errors that it consumed all of available memory and crashed the server. I'd rather put up with that risk than have no indication of where the sync errors were. Remember that each time there is a sync error, unRAID will update parity to address the out-of-sync condition. Tom should consider adding the word "corrected" to the end of the message to make it clearer (e.g, "10 sync errors corrected").
January 18, 200917 yr Odd that it was only showing 3.4GB of RAM when I had at least 4GB in there. hmm could be the mother board not liking 4GB of RAM. That's odd though. Not odd at all and quite normal in 32 bit systems without memory re-mapping....
January 18, 200917 yr Author first pass complete with no errors and only 2GB or RAM. Rebooting and running parity check. let's see what happens
January 18, 200917 yr Author Odd that it was only showing 3.4GB of RAM when I had at least 4GB in there. hmm could be the mother board not liking 4GB of RAM. That's odd though. Not odd at all and quite normal in 32 bit systems without memory re-mapping.... I checked the manufacturers website for my motherboard. it does support 4GB of RAM but only ECC at that high. I'm spoiled as I'm used to work servers and new motherboards that can suck up tens of GBs of RAM without a problem. Forgot some of this old hardware can't support that much RAM. now i'm running a parity check and it is still finding errors. nothing in the message logs, is this normal for it to do this?
January 18, 200917 yr Odd that it was only showing 3.4GB of RAM when I had at least 4GB in there. hmm could be the mother board not liking 4GB of RAM. That's odd though. Not odd at all and quite normal in 32 bit systems without memory re-mapping.... I checked the manufacturers website for my motherboard. it does support 4GB of RAM but only ECC at that high. I'm spoiled as I'm used to work servers and new motherboards that can suck up tens of GBs of RAM without a problem. Forgot some of this old hardware can't support that much RAM. now i'm running a parity check and it is still finding errors. nothing in the message logs, is this normal for it to do this? Normal for nothing to be in the logs. You might be "correcting' the errors written to parity earlier. (In other words, it was changing parity to incorrect values it thought were correct) Let this parity check finish, it is expected that the initial run find errors, then run a second parity check. It should find NO errors. If parity errors continue, you still have problems. Joe L.
January 18, 200917 yr Author should i be copying files to/from the raid at this point or just stop doing anything on it all together? ETA is going to be 10hrs on it as it is.
January 18, 200917 yr Author I stopped all traffic to it. It's doing nothing but checking at the moment.
January 18, 200917 yr Author so far Total size: 976,762,552 KB Current position: 243,943,676 (24.9%) Estimated speed: 32,004 KB/sec Estimated finish: 381.5 minutes Sync errors: 79 This is allot less than the 400 i was getting for the same time. I still want to know what files it's "fixing" is there anyway that's possible?
January 18, 200917 yr so far Total size: 976,762,552 KB Current position: 243,943,676 (24.9%) Estimated speed: 32,004 KB/sec Estimated finish: 381.5 minutes Sync errors: 79 This is allot less than the 400 i was getting for the same time. I still want to know what files it's "fixing" is there anyway that's possible? Short answer - no. I am actually working on creating PAR2 files for my drives so that I can answer this type of question after I get some parity errors. Without having done something like that, there is no way that I know of. Even if you had the cluster numbers (which we don't anymore), trying to convert them into filenames would be more than a little difficult. But without them, it is impossible.
January 18, 200917 yr Author are you creating a specific volume set aside for just these logs? I'd almost be willing to hookup a 2nd USB device into my system or something to dump error logs to. I could care less about the device if it fails, I just want to see what is fixed and what was broken. Now I'm using this thing heavily for bittorrent, could this affect my raid in anyway? I don't think so but I'm trying to rule out any and all possibilities as this only started happening when i put in those 2 IDE drives and added the RAM and started massively downloading torrents.
January 19, 200917 yr Author After a 6hrrun I've got the following. Total size: 976,762,552 KB Current position: 923,732,080 (94.5%) Estimated speed: 40,409 KB/sec Estimated finish: 21.8 minutes Sync errors: 757 I'll guestimate a few more errors in the next 20 or so minutes. I'm running it again to make sure. Could my parity disk be out of sync? I'm noticing that some of my files are not being shown as changed on my xbox via streaming but are still attempting to play.
January 19, 200917 yr You probably know this, but I just wanted it to be clear. A parity check (or build) does not change the data in any way. It only changes the parity bits on the parity drive. So no matter what happens with the errors, your data is safe. The risk is, if parity info was incorrect and a drive died, you may have problems accurately rebuilding parts of that data drive.
January 19, 200917 yr Author My overall concern was that the parity mismatch may result in data loss. I've seen RAID arrays get out of sync, lose their build information, and then go teats up. Since this is my own data and unRAID is new to me (wild ride and all) I didn't know how this parity sync would handle the rebuild. Would it sync the data with my own disks or would it say "my stuff is newer use me" and overwrite it. I'm used to our servers losing a Drive(s) and then put back in replacements with the RAID rebuilding itself making sure all data across all disks are running smoothly. Of course I don't get to SEE anything besides the pretty tickers go by or the messages saying "RAID complete in X minutes" Why should this be any different? I guess since it's home built I thought there would be more to it, but it's KISS at it's finest. Thanks for the hints. My next parity check will finish in: 515.6 minutes . On your advice I decided to run it again to make sure there isn't anything else wrong with my array. By this time I would've had allot more errors, so far so good. I'll cross my fingers. All my trials and tribulations, and the members on this forum, have built a nice learning curve/experience for me. Even though some of my questions sound silly or trivial, i'm glad I took the plunge.
Archived
This topic is now archived and is closed to further replies.