[MXS-3898] Add a way to define and switch different timeouts for primary and replicate during failover Created: 2021-12-03  Updated: 2023-03-10  Resolved: 2022-09-26

Status: Closed
Project: MariaDB MaxScale
Component/s: Documentation, failover
Affects Version/s: None
Fix Version/s: 6.4.2, 22.08.1

Type: New Feature Priority: Major
Reporter: Valerii Kravchuk Assignee: Esa Korhonen
Resolution: Fixed Votes: 0
Labels: None

Sprint: MXS-SPRINT-166

 Description   

There are good reasons to have different settings for various timeouts like:

  • max_statement_time
  • wait_timeout
  • lock_wait_timeout
  • innodb_lock_wait_timeout
  • etc

for primary/master server vs slave/replica. When failover happens these settings should be changed and assuming Maxscale does and controls failover process, it should get options to do this. I mean separate settings for "primary" and "replica" timeouts or maybe any/all global variables to enforce on a node that switches to that status.



 Comments   
Comment by markus makela [ 2022-03-08 ]

Changing this to a New Feature since that's what it is.

Comment by markus makela [ 2022-03-14 ]

This should already be possible using the promotion and demotion SQL files: https://mariadb.com/kb/en/mariadb-maxscale-25-mariadb-monitor/#promotion_sql_file-and-demotion_sql_file

valerii do you think this would be adequate or would some more refined behavior be required?

Comment by Valerii Kravchuk [ 2022-03-14 ]

Yes, for any global dynamic variables these files seem to provide a way to solve the problem. But I'd appreciate some clear examples on how to use them to set timeouts in the KB and/or Enterprise documentation.

Comment by markus makela [ 2022-03-14 ]

Could you give some example values that one would use with the given variables you had? We can then add these as examples into the documentation for these parameters.

Generated at Thu Feb 08 04:24:45 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.