Cache pool - BTRFS crash, switching back to XFS

[limawaken]

Hello guys,

 

for years prior to this i used only a single drive xfs cache and never faced any problems.

Last year i decided to give multiple disk cache pools a go, to have a bit of peace of mind in case of a cache disk failure.

I know jonathanM isn't a big fan of BTRFS, and despite coming across several hints throughout the forums to avoid btrfs like the plague... I thought i'd give it a try anyway.

 

 

so i got 2 new 1tb SSDs and set it up.

I only used it for 8 months and I have had to rebuild my cache pool 3 times.

Same problem each time. VMs and dockers suddenly crash. I use pfsense VM so I know quite soon when the problem hits because everyone will be asking if the internet is down. its usually most inconvenient when this happens.

Usually what happens is the logs would be filled with BTRFS errors and the filesystem becomes read-only.

there are not many topics on repairing btrfs filesystems, so rebuilding seems to be the easiest solution.

 

i don't know if its really btrfs reliability... it could be my hardware (still running on pretty old i7-3770 hardware), but the disks work fine as individual XFS disks.

i think it would be better to setup both disks as single disk cache pool, and clone the contents of the cache disk to the other disk.

if the cache disk fails i can quickly point my shares to the cloned cache pool.

i have not experienced a single ssd failure in all my years so i shouldn't have to do this very often. taking this into consideration it just doesn't make much sense to use BTRFS pool for me now.

 

is there a plugin that can clone or mirror my cache disk contents so i can simply point my shares to the new disk?

i really like ca backup restore appdata, it saved me many times. but the backup is a TAR.

[JorgeB]

You can use the mover to move all the data from the pool to the array, create new pool(s) and move data back:

 

https://forums.unraid.net/topic/46802-faq-for-unraid-v6/?do=findComment&comment=511923

 

[limawaken]

38 minutes ago, JorgeB said:

You can use the mover to move all the data from the pool to the array, create new pool(s) and move data back:

 

https://forums.unraid.net/topic/46802-faq-for-unraid-v6/?do=findComment&comment=511923

 

i actually hadn't thought of doing it this way... i was able to copy some files off the cache disk so i guess all the data was intact, but would mover still work since the filesystem was already corrupted and the cache had become read-only?

[JorgeB]

If it's read-only mover might not work, you can still move the data manually using your favorite tool.

 

P.S. if btrfs keeps getting corrupt you likely have some hardware issue, like bad RAM, xfs is more resilient to hardware issues, but if there's a problem data will still get corrupt, you just won't know about it.

[limawaken]

16 hours ago, JorgeB said:

if btrfs keeps getting corrupt you likely have some hardware issue

maybe for me having 2 disks in a pool kinda doubles the probability of corruption. could be some poor connection between my mobo sata controller and the disks (sata cable, drive cage). last memtest i did a couple months ago went well.

 

incidentally there was a lightning storm last month, knocked out my server... had to replace my server power supply, UPS and network switches. although that was a couple week ago maybe that sudden unclean shutdown contributed to the btrfs corruption.