[MXS-1544] master_down split into lost_master and server_down Created: 2017-11-27 Updated: 2018-01-16 Resolved: 2018-01-16 |
|
| Status: | Closed |
| Project: | MariaDB MaxScale |
| Component/s: | N/A |
| Affects Version/s: | None |
| Fix Version/s: | N/A |
| Type: | Bug | Priority: | Minor |
| Reporter: | markus makela | Assignee: | Unassigned |
| Resolution: | Incomplete | Votes: | 0 |
| Labels: | None | ||
| Description |
|
If a part of the queries to a server are successful but the other part fail, a master_down or slave_down event can be wrongly interpreted as two separate events: lost_master and server_down. This causes the generation of master_down events to be non-deterministic. |
| Comments |
| Comment by markus makela [ 2017-12-27 ] |
|
This has not been reproduced and was merely a deduction based on the code. |
| Comment by markus makela [ 2018-01-02 ] |
|
Lowering to minor as this has not been reproduced or even estimated. |
| Comment by markus makela [ 2018-01-16 ] |
|
Cannot reproduce. |