Corrupted xfs
Hi, one of our (E3 Samsung 850 pro Lsi Raid 10) ESXi host servers has:
1 x Cpanel Cloudlinux VM
Centos 6 Ext4 LVM
1 x Cpanel VM
Centos 7 Xfs
Twice this week we had an issue with Raid dropping two drives (the wrong two!) so the VM's crashed. I changed the drives from UBAD to UGOOD etc to get them back online until replacement.
The VM with ext4 file format ran fsck and then came back up ok
The VM with xfs format corrupted and after hours of stress, there was no way xfs_repair would repair the corruption, so we restored from backups.
Has anyone else experienced this? I'm considering rebuilding the Centos 7 VM's and future VM's as ext4 because until now I didn't realise fsck did not work with xfs.
It seems ext4 is more tolerant to Raid and crash recovery.
-
Howdy, In our testing we've seen very similar outcomes. If you want fast and you'll make backups XFS all the way. If you're going to need recovery you'll have better chances with EXT4 but performance will suffer. For us it means developer sandboxes which could have lots of changes each day between backups get EXT4. QA crash test vm's get XFS. For the record I always recommend backups! 0
Please sign in to leave a comment.
Comments
1 comment