Skip to:
Bug:
replication user is used in cross-site replication feature described here: https://www.percona.com/doc/kubernetes-operator-for-pxc/replication.html
Changing the password for this user affects the database, but not the replication.
I need to manually stop the replication, change the password and start it back.
Steps to reproduce:
Setup the replication
Change the password on source and replica by changing it in Secret
Get the following error in mysql
Expected behavior:
Operator should automatically change the password by executing:
is right.
I added the following on the replica side:
Got quite a quick turnaround:
I'm not sure about the timeouts here.
My setup
one k8s cluster
two namespaces
two PXC clusters in these namespaces - one main, another is replica
Replica cluster has the following settings:
My current Master_Host is:
What is happening and expectations
If the Master_Host is not reachable, I expect that the replication would switch to another host in 14 seconds.
In reality I see it is happening 67 seconds. See the log below:
13:23:09 is the first try
13:24:16 is the second try
I tries the same with 3 retries and 5 seconds and the switch happens in 65+65 seconds:
13:08:12
13:09:17
13:10:22
Any idea what is happening here? Is it network timeout causes 60 second delay?
Bug:
replication user is used in cross-site replication feature described here: https://www.percona.com/doc/kubernetes-operator-for-pxc/replication.html
Changing the password for this user affects the database, but not the replication.
I need to manually stop the replication, change the password and start it back.
Steps to reproduce:
Setup the replication
Change the password on source and replica by changing it in Secret
Get the following error in mysql
Expected behavior:
Operator should automatically change the password by executing: