High CIFS latency on FlexGroup constituents from heavy RENAME workload
Applies to
- ONTAP
- FlexGroup
- CIFS
Issue
Users may experience high latency in FlexGroup volumes, manifesting as:
- Elevated CPU D-blade latency.
- High CIFS or NFS latency affecting multiple constituents across nodes.
- Hung NFS RPC calls or stalled application processes dependent on file system operations.
Example latency output from a cluster:
cluster01::*> system node run -node node-01 -command "wafltop show -v latency -i 10 -n 10"
Latency
Application Latency ms
----------- ----------
aggr01:VOLUME_FLEXGROUP__0010:cifs: 885.569
aggr01:VOLUME_FLEXGROUP__0008:cifs: 58.957
aggr01:VOLUME_FLEXGROUP__0009:cifs: 53.140
aggr01:VOLUME_FLEXGROUP__0004:cifs: 52.011
aggr01:VOLUME_FLEXGROUP__0006:cifs: 38.911
aggr01:VOLUME_FLEXGROUP__0011:cifs: 37.636
aggr01:VOLUME_FLEXGROUP__0001:cifs: 34.990
- The FlexGroup CVs had 100% remote placement.
- Latency spikes may be isolated to specific constituents, with coefficient of variation (CV) exceeding 400% and activity levels indicating heavy workload imbalances.
- What do "flexgroup show" target and urgency and other columns mean?
