[MXS-1941] Auto READ request routing to MASTER if slaves are lagging behind Created: 2018-06-26 Updated: 2018-11-27 Resolved: 2018-11-27 |
|
| Status: | Closed |
| Project: | MariaDB MaxScale |
| Component/s: | readwritesplit |
| Affects Version/s: | None |
| Fix Version/s: | 2.3.1 |
| Type: | New Feature | Priority: | Major |
| Reporter: | Faisal Saeed (Inactive) | Assignee: | markus makela |
| Resolution: | Fixed | Votes: | 0 |
| Labels: | None | ||
| Description |
|
The customer expects MaxScale to take care of the routing of queries internally which it does, also if all the slaves are lagging behind or the GTIDs are not yet caught up to Master, The "READ" queries should automatically be routed to The Master DB node. This is to simplify the application design so that app team doesn't need to worry about getting outdated data reads from SLAVES. |
| Comments |
| Comment by markus makela [ 2018-11-08 ] |
|
Possibly fixed by causal_reads in 2.3.1: it will route the query to the master if the slaves are too far behind. |
| Comment by markus makela [ 2018-11-27 ] |
|
This is fixed by the changes to causal_reads that now retries the query on the master. |