[MXS-3755] Handle multiple replication sources on rejoin Created: 2021-09-03 Updated: 2021-09-07 Resolved: 2021-09-07 |
|
| Status: | Closed |
| Project: | MariaDB MaxScale |
| Component/s: | mariadbmon |
| Affects Version/s: | 6 |
| Fix Version/s: | 6.2.0 |
| Type: | New Feature | Priority: | Major |
| Reporter: | markus makela | Assignee: | markus makela |
| Resolution: | Fixed | Votes: | 0 |
| Labels: | None | ||
| Issue Links: |
|
||||||||
| Description |
|
When mariadbmon performs a failover, it currently merges all the replication connections from the old master to the newly promoted master. This poses a problem: if a rejoining server has more than one replication stream configured, the rejoin won't be attempted:
To fully support an infinite amount of failovers, the rejoining process must clear out all existing replication streams from the server to be rejoined. As this is an irreversible process, it should only be attempted when enforce_simple_topology is enabled. |
| Comments |
| Comment by markus makela [ 2021-09-07 ] |
|
The failover part already worked correctly in version 6, the only part that did not work was the rejoin when there was more than one configured replication source. In 6.2.0, a failover from one node to the other and back will preserve all configured replication streams. If a node rejoins the cluster and it has more than one configured replication source, the first one is modified to point to the cluster and the others are removed. |