I have had some odd issues and I want to make sure i am not losing Data.  I have 6 disk plus parity.  1 6tb parity drive, 2 6tb drives and 4 3tb drives.  So 24TB total with 9tb free.

 

I had one 3TB drive that starting errors so I removed it and rebuilt the parity.  Everything seemed to go well.  Now I now have another 3TB drive that is randomly showing up as Unmountable: No File System.  I was able to get to finally get it to come up and mount doing a xfs_repair but after reboot its tends to go back to the Unmountable: No File System state.  

 

Here is my question.  If I am able to get it to report in and mount after doing another xfs_repair can i do a parity check and then remove the drive as well?  I have a replacement on the way but do not want to lose what's on it.  If the parity is good and the drive is mounted can I remove it and add a new and let it rebuild with out concern?

tower-diagnostics-20220406-1331.zip

Edited April 6, 20224 yr by stuss02
added Diagnostics

You're going to want to post your diagnostics 

3 minutes ago, Squid said:

You're going to want to post your diagnostics 

IS it ok to run that with parity check running?

 

On mobile now so can't look at diagnostics yet. Just some forum advice.

 

Since you attached them to a post already read by some they won't know there was anything new to see.  But since I made a new post this thread will show as unread again. 

With your continuing issues and this:

Apr  6 11:36:13 Tower kernel: BTRFS error (device sdc1): bdev /dev/sdc1 errs: wr 0, rd 0, flush 0, corrupt 2042, gen 0

I'd start by running memtest from the bootmenu for at least a couple of passes (By and large "corrupt" on a btrfs drive is because of memory issues)

 

If you're booting via UEFI mode, you will have to temporarily switch to Legacy mode in order to run the memtest

16 minutes ago, Squid said:

With your continuing issues and this:

Apr  6 11:36:13 Tower kernel: BTRFS error (device sdc1): bdev /dev/sdc1 errs: wr 0, rd 0, flush 0, corrupt 2042, gen 0

I'd start by running memtest from the bootmenu for at least a couple of passes (By and large "corrupt" on a btrfs drive is because of memory issues)

 

If you're booting via UEFI mode, you will have to temporarily switch to Legacy mode in order to run the memtest

Interesting.... Oddly enough I was doing the disk replacement in prep for hardware migration.  Just wanted to sort this out before throwing something new in the mix.

26 minutes ago, stuss02 said:

Just wanted to sort this out before throwing something new in the mix.

You don't want to do anything at all with any computer unless the RAM is trustworthy.

19 minutes ago, trurl said:

You don't want to do anything at all with any computer unless the RAM is trustworthy.

 

I will do a memtest after the Parity finishes.  It might be because the machine has 256g of ECC ram.  I know it cant use all that.

Builtin memtest won't work with ECC. Probably some other reason for btrfs corruption if using ECC. Didn't notice anything logged about memory which would usually be the case with ECC since it knows if it had to correct.