SnapMirror relationships hang in transferring on ONTAP 9.4 and later
Applies to
- ONTAP 9.4 and later
- SnapMirror
Issue
SnapMirror relationships hang in a transferring state after moving a small amount of data. This issue occurs when ALL of the following are true:
- The SnapMirror source cluster is running ONTAP 9.4 or later.
- SnapMirror global replication throttles were enabled on the SnapMirror source cluster when the issue began. SnapMirror global replication throttles can be viewed through the
options replication.*command on the source cluster. - This issue impacts specific SnapMirror source nodes. The hung / stuck SnapMirrors may have a specific source node in-common while SnapMirrors from other source nodes in the same cluster may behave normally. In some cases, all source nodes in the cluster can be impacted.
In the SnapMirror source node's EMS logs, located in /etc/log/EMS, messages such as the following example will appear:
Tue May 01 00:00:00 [ClusterA-01: repl_Handle_reg: repl.xfer.term.timed.out:error]: A component of a replication transfer with destination volume DSID '1234' or SnapMirror(R) tape ndmp session '' is taking a long time to terminate since it is busy performing some operation.
In the SnapMirror source node's SKTrace logs, located in /etc/log/mlog/sktrace.log, messages such as the following example will appear:
2019-05-01T00:00:00Z 34159448880650904 [5:0] REPL_0: repl_sender::NodeFactory::abortOrphanSenderNodes(): | Clearing orphan sender node with transfer uniq: 012345678-0123-0123-0123-0123456789ab
Before applying the solution in this KB, ensure that all of the above criteria are met, and that matching error messages have been found. There are many issues, manifesting in various ways, that can result in SnapMirrors appearing to hang. If a similar issue is encountered, but the cause is unclear, open a case with NetApp Technical Support.