Corrupt BTRFS Files Systems Under / For Multiple VMs In The Same Time Period (Doc ID 2590129.1)

Last updated on JULY 20, 2024

Applies to:

Symptoms

• Oracle Database Appliance HA
• Any ODAVP Hardware Type
• VM found running on both nodes using a shared file system repository
• Corruption of the file systems detected after the VM is found running on both nodes at the same time.
  Corruption included System.img files of the available snapshot backups for these VMs
• The same symptom may be found on one or more of the VMs
• Files systems used can include:
  BTRFS mounted under /
  Corruption included System.img files of the available snapshot backups for these VMs
  XFS mounted under /u01 found corruption of U01.img
• Possibly related:
  online snapshot backup mechanism
  ACFS/ASM

Corrupt BTRFS files systems under / for multiple VMs in the same time period

• The XFS corruption seemed initially not present in the U01.img file of the available snapshot backup of the relevant VM.
• The issues with the corrupted BTRFS filesystems were actually fixed by using System.img files of comparable VMs on the other ODA.
• The issue with the corrupted XFS files system was solved by using the U01.img of the available snapshot.
• The first restore of that seemed initially OK but got corrupted beyond repair after the first write actions on /u01.
• Simple linux commands such using " ls " which previously had worked, now failed with

"ls: cannot open directory .: Input/output error"

• After a new restore of the U01.img file, the XFS filesystem was immediately checked for corruption using 'xfs_repair -n' and then actually repaired.
• After that the filesystem showed no problems

Changes

A VM was offline but restarted resulted in both nodes with the same VM running

Cause

My Oracle Support provides customers with access to over a million knowledge articles and a vibrant support community of peers and Oracle experts.