CONTAP-118215: IO disruption or high latency following SMas Planned Failover
Issue
In certain ONTAP systems employing SnapMirror active sync (SMas) feature, some SMas source consistency groups (CGs) and volumes may experience I/O disruptions following a planned failover operation on a different SMas relationship. This may also show up as sudden high latency in the range of 20 sec.
Illustration
cg1_src ----- cg1_dst
- cg1_vol_src ----- cg1_vol_dst
cg2_src ----- cg2_dst
- cg2_vol_src ----- cg2_vol_dst
When snapmirror failover command is executed on cg1_dst, cg2_vol_src which belongs to cg2 can potentially face an IO disruption if cg1_vol_src and cg2_vol_src are located on the same node.
Volumes Affected
- The defect is specific to SMas volumes and typically limited to a specific volume co-located with source endpoints of the CG undergoing planned failover.
- Non-SMas volumes (for example, volumes in SM-S or SM-Async relationships) are not affected.
Symptoms
- Hosts may notice high latency in reading or writing data — often a consistent latency of about 20 seconds.
- Hosts may encounter other I/O disruptions or completely inaccessible LUNs associated with the consistency groups impacted.