Old replica configuration is not purged if a channel is renamed in cr.yml

Description

The issue:
--------------
If you are not adding, but simply renaming a channel in cr.yml, the slave configuration for the old channel name stays in place and one more configuration is created. It looks a bit confusing because only one configuration is described in cr.yml

How to reproduce:
--------------------------
I have this in my cr.yml

And this is the configuration in the DB:

If I update the cr.yml and change the channel name

I have the updated info in the replication_asynchronous_connection_failover table:

But I still have the old slave configuration:

It's not a big problem in itself, but maybe confusing, since only one configuration is described in cr.yml and only one configuration supports connection failover in this case

Possible solution:
------------------
The following command will clean up the configuration for the specified channel:

Environment

None

Smart Checklist

Activity

Show:

Slava Sarzhan October 18, 2021 at 10:06 AM

In case of cross-region we do not know if it is old channel or just replication was temporary disabled or something else happened. That is why we need to keep it to have possibility to continue replication as soon as we can do it.

Sergey Pronin September 17, 2021 at 8:17 AM

 is it expected behavior?

Not a Bug

Details

Assignee

Reporter

Affects versions

Priority

Smart Checklist

Created August 18, 2021 at 4:46 PM
Updated March 5, 2024 at 5:46 PM
Resolved October 18, 2021 at 10:09 AM