My Oracle Support Banner

MySQL Shell / InnoDB Cluster Cannot Rejoin Instance To Running Cluster (Doc ID 2311491.1)

Last updated on DECEMBER 17, 2020

Applies to:

MySQL Server - Version 5.7 and later
Information in this document applies to any platform.
If one of the InnoDB cluster member has been restored with MEB, and if it's added back to the cluster using - cluster.rejoinInstance() , then the status for this added node will be shown as - (MISSING).

However, the Group Replication would be running fine after rejoining. Only the shell doesn't detect this newly added instance.

Checking through the shell log suggests that, there has been change in UUID of the instance after MEB restore and that's obvious. However, in 'instances' metadata table, the obsolete server UUID still persists.

The rejoined instances won't be recognized as it's old member, because of the changed UUID and it can't be added as new instance due to it's same hostname-port number.

In the meantime, Group replication was able to identify this change in UUID and it accepted the old member. It's the Shell that denies to accept the change!

Goal

To explain how MySQL Shell ( up to version 8.0.23 ) had an known defect which prevented a node from being rejoined back to the running Cluster after being removed.

Solution

To view full details, sign in with your My Oracle Support account.

Don't have a My Oracle Support account? Click to get started!


In this Document
Goal
Solution
 Remove Instance
 super_read_only
 Add Instance
References


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