[MXS-4737] Graceful shutdown of MaxScale when using more than one MaxScale instance Created: 2023-08-31  Updated: 2023-12-15

Status: Open
Project: MariaDB MaxScale
Component/s: failover
Affects Version/s: None
Fix Version/s: Icebox

Type: New Feature Priority: Major
Reporter: Michael Widenius Assignee: Joe Cotellese
Resolution: Unresolved Votes: 1
Labels: None

Issue Links:
Relates
relates to MDEV-32053 New features requested by customer on... Open

 Description   

The customer has to restart MaxScale in case of MaxScale upgrade or MaxScale maintenance (for example if MaxScale would start to take too many resources).

The task is to implement a graceful shutdown of MaxScale without causing any notable interruptions to users that uses MaxScale to connect to MariaDB server.

The logic would be as follows when requesting a graceful shutdown:

  • Move all connections to the other MaxScale
  • Shutdown MaxScale
  • One restart, move half of the connections from the other MaxScale to the current one.

The customer can then repeat the above procedure with the other MaxScale if needed.

 Comments   
Comment by Michael Widenius [ 2023-09-14 ]

Yes, MXS-4153 is the same thing, but this MXS is more explicit how it would work from the customers point of view.
I suggest that someone takes that example from above and adds it to MXS-4153, after which one can declare this as a duplicate.
Note that the resulting task MUST also be linked from https://jira.mariadb.org/browse/MDEV-32053 so that the customer can easily follow the progress of the task they think are important for them.

Generated at Thu Feb 08 04:30:47 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.