[MXS-4122] Fast Global causal_reads Created: 2022-05-09 Updated: 2022-09-15 Resolved: 2022-09-15 |
|
| Status: | Closed |
| Project: | MariaDB MaxScale |
| Component/s: | readwritesplit |
| Affects Version/s: | None |
| Fix Version/s: | 22.08.2 |
| Type: | New Feature | Priority: | Major |
| Reporter: | Faisal Saeed (Inactive) | Assignee: | markus makela |
| Resolution: | Fixed | Votes: | 0 |
| Labels: | None | ||
| Issue Links: |
|
||||||||
| Sprint: | MXS-SPRINT-166 | ||||||||
| Description |
|
Request to implement a `causal_reads=fast_global` to have global causal reads scope but no wait for the timeout before routing the SELECT to the primary node. The use-case covers the majority of cases where the user would want to have a fast query response globally without the wait for replication synchronization. |
| Comments |
| Comment by markus makela [ 2022-07-18 ] |
|
More often that not this will devolve into routing all traffic onto the primary node of the cluster and you'd be better off just using max_slave_connections=0. The only case where this might be useful is when you have long periods of completely read-only traffic with short bouts of write-intensive loads. Even a single writes will end up redirecting all traffic onto the primary for roughly the duration of monitor_interval. |