Closed tonyhutter closed 5 days ago
I think this problem goes from L2ARC rebuild not being stopped by spa_export_common()
before setting spa_final_txg
, which with some luck may log its completion between that time and spa_l2cache_drop()
call by spa_unload()
.
I've intercepted the log message that causes it, confirming my guess:
VERIFY3(tx->tx_txg <= spa_final_dirty_txg(spa)) failed (64 <= 63) L2ARC rebuild successful, restored %llu blocks
System information
Describe the problem you're observing
QEMU Almalinux 9 runner reported
l2arc_mfuonly_pos
as KILLED with this panic:Describe how to reproduce the problem
Saw this as a test result while testing an unrelated PR https://github.com/tonyhutter/zfs/actions/runs/11598165017/job/32293419930?pr=3
Include any warning/errors/backtraces from the system logs