November 20, 20187 yr Your parity rebuild completed successfully. Yay! Yes, a parity check would be a good thing to do next. Your cache pool is badly broken with a lot of BTRFS errors, but I think you were dealing with that in your other thread.
November 20, 20187 yr Author 22 minutes ago, John_M said: Your parity rebuild completed successfully. Yay! Yes, a parity check would be a good thing to do next. Your cache pool is badly broken with a lot of BTRFS errors, but I think you were dealing with that in your other thread. Do you mean running a SCRUB on the cache discs? If yes, I should probably wait till the parity check completes as it may be too intense for the system to do both. Edited November 20, 20187 yr by levster
November 20, 20187 yr It was just my observation from looking at your diagnostics. Yes, indeed. Wait until you're happy with the main parity protected array (i.e. for the parity check to complete) before trying to deal with the cache.
November 21, 20187 yr Author So, the parity rebuild and check went through without a hiccup. I did perform SCRUB with about 135 errors corrected. Here is the latest diagnostic report. Anything in it that is concerning? Thanks, Lev tower2-diagnostics-20181121-1611.zip
November 21, 20187 yr Parity looks good. The cache scrub certainly fixed a lot of errors. It might be worth running it again to see if it picks up any more. Keep an eye on disk ST32000542AS_5XW18GZS. I has had problems in the past, but not recently, it seems. The rest of the SMART reports look good. Your server is in a much better place than it was a few days ago. There's an update for the Unassigned Devices plugin that will get rid of some annoying error messages about another server being offline.
November 22, 20187 yr Author 1 hour ago, John_M said: Parity looks good. The cache scrub certainly fixed a lot of errors. It might be worth running it again to see if it picks up any more. Keep an eye on disk ST32000542AS_5XW18GZS. I has had problems in the past, but not recently, it seems. The rest of the SMART reports look good. Your server is in a much better place than it was a few days ago. There's an update for the Unassigned Devices plugin that will get rid of some annoying error messages about another server being offline. Sounds good. Just re-ran the SCRUB and no errors were identified. I am going to run some zeroing on spare discs from my other controller cards to see if they are the culprit. Thank you to all who offered all of the help. I truly appreciate the unRaid community! Edited November 22, 20187 yr by levster
November 22, 20187 yr Community Expert 10 hours ago, levster said: I did perform SCRUB with about 135 errors corrected. As long as there were no uncorrectable errors you're fine, but note that any NOCOW shares can't be checked or corrected.
November 27, 20187 yr Author I was hoping that all the problems were behind me. Yesterday, for an unknown reason a parity check was started and finished with no errors. All seemed OK and then about an hour later I received a message that two of the disks were in error state, Parity and disc 2. I am at my wits end. Is it the controller, cable or something else? I moved the discs around and am still having issues. Any ideas? Thank you, tower2-diagnostics-20181127-0159.zip
November 27, 20187 yr Community Expert Diagnostics are after rebooting so not much to see, but almost certainly not a disk problem.
November 27, 20187 yr Author 5 hours ago, johnnie.black said: Diagnostics are after rebooting so not much to see, but almost certainly not a disk problem. I think that I stopped the array and grabbed the diagnostics before I rebooted. But, I could be wrong. So, how should I proceed to restore the two missing discs? Edited November 27, 20187 yr by levster
November 27, 20187 yr Community Expert Unassign both disks, start the array, reassign both disks to begin parity sync and disk rebuild, might be worth checking cables/connections, controllers well seated, etc, before doing it.
November 27, 20187 yr Author 7 minutes ago, johnnie.black said: Unassign both disks, start the array, reassign both disks to begin parity sync and disk rebuild, might be worth checking cables/connections, controllers well seated, etc, before doing it. Ok. Will start. I checked and at least the cards and cables seem ok. But, I am strongly suspecting that there is a serious hardware issue. I am putting together another server, but it will take me some time to get all the dockers working. Also, I will need to get another Win 10 license for a VM that I have running on this server. Hope that this server lasts without me loosing data. Edited November 27, 20187 yr by levster
November 27, 20187 yr Community Expert 5 minutes ago, levster said: I am strongly suspecting that there is a serious hardware issue. Probably, also if you have a spare you could use it for rebuilding the data disk, in case something goes wrong, so you still have the old data disk intact.
November 27, 20187 yr Author 1 minute ago, johnnie.black said: Probably, also if you have a spare you could use it for rebuilding the data disk, in case something goes wrong, so you still have the old data disk intact. Opps. data rebuild started already. Too late... Edited November 27, 20187 yr by levster
November 29, 20187 yr Author So, the data rebuild went fine. All was well for a day or so and then this AM multiple disc errors showed up. At this point I am beyond frustrated. I've changed out the order of the discs between different controllers. I changed out the cables. I am really at a loss as to what to do next. I have a second server that I can start to transition to. So, should I start to move over the data discs there, one at a time, mount them and copy the data on to the array? Should I try to backup the what data I have still intact from this server on to an external disk? Is the data on he discs with errors still intact or are there true errors? Lev tower2-diagnostics-20181129-1149.zip
November 29, 20187 yr How many data need to transition, and why transition from an unstable system ? Does 2nd server confirm stable, why can't just attach all disks to it ? May be your X58 board was age (2011), would you consider replace it by a new one have 3 PCIe slot. I think you need a stable system to host all disks rather then move data in & out. Edited November 29, 20187 yr by Benson
November 29, 20187 yr Author The new system is up and running with a few data discs that I used to test it out. It is based on an MSI Z77A-G45 board. It has three PCI-e slots that I will move over the SIL PERC H310 cards into. I am just wondering what would be the safest way to migrate? Move one data disc at a time, copy it onto the array, then format the disc as a member of the new array, move to the next disc? Also, how can I make sure that the data on the current discs is OK?
November 29, 20187 yr Since you can't sure original disk/file have corrupt or not, copy or move won't make change. So I will put all data disk in this new platform, start array in maintenance mode ( so no writes or minmize ) and perform read check several time. If got stable then add parity disks, rebuild and test. Edited November 29, 20187 yr by Benson
November 29, 20187 yr Author 7 minutes ago, Benson said: Since you can't sure original disk/file have corrupt or not, copy or move won't make change. So I will put all data disk in this new platform, start array in maintenance mode ( so no writes or minmize ) and perform read check several time. If got stable then add parity disks, rebuild and test. What do you mean by "read check"? Do you mean to mount them as an unassigned device and read them?
November 29, 20187 yr No, just normal assign data disk as array member. If no parity disk in array, then array "check" will read all disk in same time instead of parity check / correct. Those are read check only, nothing about file system, the purpose was verify system stable or not. Edited November 29, 20187 yr by Benson
November 29, 20187 yr Author Just now, Benson said: No, just normal assign those disk as array member. If no parity disk in array, then array "check" will read all disk in same time instead of parity check / correct. Oh. OK. That system does have a parity. So, will that make a difference?
November 29, 20187 yr The aim of no parity disk was avoid make mistake assign data disk be parity and destory any data. Suggest perform new config and insert all data disk from problem server, make correct assignment and test. Stop array auto-start, only start array in maintenance mode, read check and verify stable or not. Edited November 29, 20187 yr by Benson
November 29, 20187 yr Author 4 minutes ago, Benson said: The aim of no parity disk was avoid make mistake assign data disk be parity and destory any data. Suggest perform new config and insert all data disk from problem server, make correct assignment and test. Stop array auto-start, only start array in maintenance mode, read check and verify stable or not. That sounds good. I will try it tonight. Thank you for the advise. If errors are founds, what would be the best way to mitigate them?
November 29, 20187 yr Depends on what error, anyway, if need, you can putback all disks to problem server like nothing happen. Keep posting the result here, you will got many advice. Edited November 29, 20187 yr by Benson
November 29, 20187 yr Author 2 minutes ago, Benson said: Depends on what error, anyway, if need, you can putback all disks to problem server like nothing happen. Keep posting the result here, you will got many advice. Will do and report back. Thank you.
Archived
This topic is now archived and is closed to further replies.