SVM DR fails when performed on a mixed-version cluster
Applies to
- ONTAP 9
- Upgrade
- Mixed-version cluster
- Volumes belonging to an SVM exists on both upgraded and non-upgraded nodes
- SVM DR
Issue
- Manual upgrade was performed on a cluster consisting of 12 nodes.
- After upgrading four nodes, the remaining nodes were scheduled for upgrade at a later date, resulting in a mixed-version cluster state.
- During the scheduled update on that day, SVM DR failed for most SVMs except for a few.
- Individual SnapMirror operations and SnapMirror operations between constituents volumes succeeded, but only the overall SVM DR process failed.
- For the failed SVM DRs, the following logs are recorded in
mgwd.log.
ERR: vserverdr_snapmirror: TxID: 7610464723183607811 src/tables/vserverdr/svmdr/vserverdr_snapmirror.cc:3950 handle_marker_imp Dst Vserver:(DST SVM) : Failed to queue operation in process for Vserver: (SVM UUID). Reason: An invalid PID of 7 was received for table vserverdr_process_bynode_peek_and_applyvserverdr_snapmirror:Wed Feb 25 01:20:02 2026:src/tables/vserverdr_log_trace.cc:35 (thr_id:0x82466ad00) TxID: 7610464723183607811 src/tables/vserverdr/svmdr/vserverdr_snapmirror.cc:3950 handle_marker_imp Dst Vserver:(DST SVM) : Failed to queue operation in process for Vserver: (SVM UUID). Reason: An invalid PID of 7 was received for table vserverdr_process_bynode_peek_and_apply
- Successful SVM DRs have volumes concentrated on a single node.
- Failed SVM DRs have volumes distributed across multiple nodes and they are on mixed version.
