[MXS-4814] GTIDs used by causal_reads=global cannot be reset without restarting MaxScale Created: 2023-10-17 Updated: 2023-10-25 Resolved: 2023-10-20 |
|
| Status: | Closed |
| Project: | MariaDB MaxScale |
| Component/s: | readwritesplit |
| Affects Version/s: | 6.4.10, 22.08.8, 23.02.4, 23.08.1 |
| Fix Version/s: | 6.4.11, 22.08.9, 23.02.5, 23.08.2 |
| Type: | Bug | Priority: | Major |
| Reporter: | markus makela | Assignee: | markus makela |
| Resolution: | Fixed | Votes: | 0 |
| Labels: | None | ||
| Description |
|
When causal_reads=global is used in readwritesplit, there's no way to reset the global GTID state without either restarting MaxScale or destroying the service. One example where this could happen is if a cluster is restored from a backup and the GTID goes back to an earlier state. If extra GTIDs or diverged domains were recorded in readwritesplit, the only way to remedy the situation is to restart MaxScale. A simple solution to this is to provide a module command in readwritesplit that resets the global GTID state. |
| Comments |
| Comment by markus makela [ 2023-10-20 ] |
|
Added a reset-gtid command to readwritesplit that can be used to reset the GTID state without restarting MaxScale. |