NFS clients take about 1 minute to recover writing during takeover when using NFSv4
Applies to
- ONTAP 9
- NFSv4
Issue
- NFS clients take about 1 minute to resume I/O during takeover when using NFSv4
- Aggregate relocation due to Takeover/Giveback starts the lock reclaim grace period as indicated in EMS /
event log show
:
Example:
[?] Wed Jul 06 09:32:36 +0800 [node2: cf_takeover: cf.fm.takeoverComplete:notice]: Failover monitor: takeover completed
[?] Wed Jul 06 09:32:36 +0800 [node2: cf_takeover: cf.fm.takeoverDuration:info]: Failover monitor: takeover duration time is 3 seconds.
[?] Wed Jul 06 09:32:03 +0800 [node2: lmgr_ng_aggr_grace_worker: lmgr.reclaim.start.grace:debug]: Lock reclaim grace period started on aggregate 'node1_Aggr_Data01' with file system id '651312343'.
[?] Wed Jul 06 09:32:48 +0800 [node2: wafl_exempt01: lmgr.reclaim.stop.grace:debug]: Lock reclaim grace period stopped on aggregate 'node1_Aggr_Data01' with file system id '651312343'.
Note: This affects all NFS versions
- LIF migration starts separate grace period to allow NFSv4 clients to reclaim locks (default 45s)
Example:
nblade2: Nblade.graceBegin:debug]: NFS server grace state has begun for Vserver "SVM1", LIF ID "1028", LIP IP address "10.1.1.1"
nblade2: Nblade.graceEnd:debug]: NFS server grace state has ended for Vserver "SVM1", LIF ID "1028", LIF IP address "10.1.11.1".