e4defrag led to kernel reporting split-lock and 2 corrupted RAID arrays

Bug #2063006 reported by Jaromír Cápík
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
e2fsprogs (Ubuntu)
New
Undecided
Unassigned

Bug Description

I was curious how e4defrag works and how it could optimize a speed and loudness of 48TB mdraid array with four 16TB drives in RAID5 and tried the defrag on the whole mountpoint. After a couple of minutes the mountpoint became inaccessible and I found two of the drives marked as faulty even when S.M.A.R.T. is healthy. The dmesg log showed many messages related to "split-lock", unfortunately I don't remeber the exact wording. The whole computer became unstable and I decided to reboot. The boot failed as well, as it also damaged data on a second RAID1 array with system that I didn't touch with the e4defrag at all. So, one array kicked out 2 drives and one array ended up with a filesystem corruption and fsck fixing many issues. I could reassemble the RAID with "assume-clean" and it seems to be ok now, but the whole incident makes me believe there could be something unhealthy in the kernel's ioctl or deeper that needs attention.

The kernel version is 5.15.0-102-generic #112-Ubuntu SMP
The e2fsprogs version is 1.46.5-2ubuntu1.1

description: updated
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.