Looping lock manager can lead to NFS outage or system panic on ONTAP 9
Applies to
- ONTAP 9
- NFS
Description
- Clients see NFSv4 mount failures; SetclientID ops fail with nfs4err_delay.
- System panic with panic string similar to the following:
Panic string: process on cpu19 hung (wafl_exempt12) for 5001 milliseconds!
- A defect was introduced in the lock manager that can be triggered by qtree create operations, disabling oplocks operations, aggregate relocation, sendhome operations, volume moves, node shutdowns, and during ONTAP upgrade.
- This includes upgrading to an ONTAP version where the fix is present.
- The lock manager enters a lock recovery loop that starves the system of resources until:
- NFS service ceases to respond to clients or
- The system panics
- This is being tracked by BUG 1168841