STOP GROUP_REPLICATION Hangs On A Secondary Group Replication Member (Doc ID 2925682.1)

Last updated on FEBRUARY 02, 2023

Applies to:

Symptoms

On : 8.0 version, Group Replication

ACTUAL BEHAVIOR
---------------
STOP GROUP_REPLICATION hung on a secondary GR member.
While removing (cluster.removeInstance()) a secondary member from a GR InnoDB cluster (All members running 8.0.31), the command STOP GROUP_REPLICATION hung and we had to kill the mysqld process to be able to add the member back to the group.

EXPECTED BEHAVIOR
-----------------------
When removing the instance, there should only be a small timeout for replication to be stopped correctly, not hang for a long time.

STEPS
-----------------------
The issue can be reproduced at will with the following steps:
1. Set up an InnoDB Cluster 8.0.31 with default communication stack.
2. Confirm cluster is working with cluster.status()
3. Do cluster.removeInstance() to remove a secondary node.

BUSINESS IMPACT
-----------------------
The issue has the following business impact:
Due to this issue, users cannot easily manage the addition/removal of cluster nodes as would be expected.

Changes

Cause

My Oracle Support provides customers with access to over a million knowledge articles and a vibrant support community of peers and Oracle experts.