Re-adding parity drive with known correct data

[JorgeB]

Try running xfs_repair on the emulated disk, start the array in maintenance mode and from the console type:

 

xfs_repair -v /dev/md5

 

[JimBob]

My array is running in btrfs, is there an equivalent command for that?

[JorgeB]

Also post a screenshot of current main page just to check everything else looks normal.

[JorgeB]

Just now, JimBob said:

My array is running in btrfs, is there an equivalent command for that?

Yes, but not as easy, post your diagnostics instead then.

[JimBob]

Screenshot of my current main page:

 

 

Thanks

 

And my diagnostics file is attached

alexandria-diagnostics-20180206-1758.zip

[JorgeB]

Doesn't look good, most likely parity2 wasn't valid, or at least not 100% valid, run this after starting in maintenance mode, this is read only just to see if a valid filesystem is detected:

 

btrfs check /dev/md5

[JimBob]

Damn, will do.

 

thanks

[JimBob]

Looks bad, this is the output.

 

Is this going to be a reformat 5 and just restore data from my other backups job?

 

 

[JorgeB]

You can try a new config and this time leave the old disk5 assigned as is, instead of trying to emulate it, so just these steps:

 

-Tools -> New Config -> Retain current configuration: All -> Apply
-assign missing disk(s), you can also assign parity1 this time

-check "parity is already valid", though it isn't, but just so that it doens't start a sync
-start the array

 

If the old disk 5 still unmountable post new diags.

[JimBob]

Well, this is a turn of events I didn't see happening.

 

 

Seems everything is back to normal. Just going to check some files I know where missing before.

 

Fingers crossed!

[JimBob]

Okay all files seem to be back. Time to do a parity check to lock all this in. Anything more you suggest I do to make sure all is running well?

[JorgeB]

Parity check should be enough, need to be a correcting check since sync  errors are expected, if there are lot of errors a sync can be faster.

[JorgeB]

Almost forgot, since it's btrfs and after the parity check finishes you can run a scrub on disk5, if zero errors ale files are OK, if there are errors you can see any corrupt file names on the syslog.

[JimBob]

Thanks so much johnnie! You've saved my array. Just doing a parity check now and indeed there are lots of sync errors corrected.

 

I'll update this thread when all is corrected to list my final steps for anyone else that has this issue :0

 

 

[JimBob]

Hi @johnnie.black update to the situation.

 

Parity check completed and corrected errors, tested some media from drive 5 and it seemed OK, one or two not working but I don't know if they ever worked, or if this rebuilding messed it up.

 

Attempted to do the btrfs scrum like you suggested but it didn't seem to run, so i tried it on other drives and it ran fine. Thinking it may be a temp error I rebooted my server and upon boot was greeted by the unmountable: no file system error for drive 5 again

 

I find it odd that it is reporting this after successfully playing media from it so presumably the FS was OK (ish) before the reboot. DO you have any possible advice, or is it seeming like a reformat and redo the FS for disk5?

 

Thanks,

J

[JorgeB]

Please post current diagnostics

[JimBob]

Will do, attached are my current diags

 

Thanks

 

alexandria-diagnostics-20180208-2351.zip

[JorgeB]

The filesystem is corrupt, and it's kind of expected since you had mentioned that the drive had been partially rebuilt before, you can try the options on the FAQ post below to recover your data, but for first two and if they work you'll need a new disk to copy the data to, or you could for example temporarily unassign and use parity2, but test first and if either works then decide on the best option, btrfs restore is probably your best hope and the 3rd option should only be used as a last resort, and it's not very likely to work with the type of corruption the disk has.

 

https://lime-technology.com/forums/topic/46802-faq-for-unraid-v6/?do=findComment&comment=543490

 

[JimBob]

Hi @johnnie.black

 

The 2nd option has beenr unning for a while now and all seemed good, data being recovered at 5TB so far out of the 8TB that existed on there. Just had an unexpected power outage and my UPS didn't kick in for long enough causing the server to reboot. Is there any way to resume the restore or do I have to wipe the recovered 5TB and start again?

 

Thanks

[JorgeB]

If you run btrfs restore again to the same folder it will skip the existing files, you might want to check the last skipped file as it might be incomplete.

[JimBob]

Thant's a huge relief, I'll do that now. Side question, during the restore sometimes it says it is looping on a file and gives me an y/N/a option, I have been doing Y and testing the file and all seems fine, is there a way to get it to just to yes for everything?

 

[JorgeB]

4 minutes ago, JimBob said:

y/N/a option

a is for all

[JimBob]

Thanks again, I'll update when it's all done.

 

J

[JimBob]

So everything seems to be back data wise, to get the once disk5 (previously corrupted one I've moved data off) as my 2n'd parity, would I do it via new config? Also would I set mark parity as valid when doing this or let it rebuild parity?

 

Thanks,

J

[JorgeB]

Depends, is old disk5 still part of the array? Was btrfs restore done to the array or to a new disk outside the array? Maybe post a screenshot of current array config also.