[MXS-2110] swithover fails with error Unknown error'. Query: 'FLUSH LOGS Created: 2018-10-24 Updated: 2019-01-23 Resolved: 2019-01-23 |
|
| Status: | Closed |
| Project: | MariaDB MaxScale |
| Component/s: | failover |
| Affects Version/s: | 2.2.4 |
| Fix Version/s: | 2.2.4 |
| Type: | Bug | Priority: | Major |
| Reporter: | Aftab Khan | Assignee: | Esa Korhonen |
| Resolution: | Not a Bug | Votes: | 0 |
| Labels: | Maxscale | ||
| Environment: |
test |
||
| Attachments: |
|
| Sprint: | MXS-SPRINT-71, MXS-SPRINT-72, MXS-SPRINT-73, MXS-SPRINT-74 |
| Description |
|
We are testing MaxScale with a simple Master-Slave servers running MariaDB 10.1. Automatic failover works fine. We have two servers: server1 = INFDMYS001.example.gov.uk We have successfully tested following event series:
But switchover does not work:
Monitor user has all required privileges:
I have attached maxscale log and config files. |
| Comments |
| Comment by Esa Korhonen [ 2018-11-28 ] |
|
Tried reproducing this but no luck. Does this happen with more recent MaxScale/Server versions as well? The error message means that the monitor tried to "FLUSH LOGS;" on the current master when demoting it, however the query failed. Then, the monitor queried the connector driver for an error description and got the "Unknown error". This is a valid server error code but doesn't explain anything. Is this a repeating occurrence, happening every time? What if you log into the server using a client program and the monitor user and try to "FLUSH LOGS;"? |
| Comment by Aftab Khan [ 2018-12-11 ] |
|
>What if you log into the server using a client program and the monitor user and try to "FLUSH LOGS;"? |
| Comment by Esa Korhonen [ 2019-01-08 ] |
|
I suggest trying with a more recent version of MaxScale. Alternatively, you can add the monitor settings "backend_read_timeout" and "backend_write_timeout" and set them to e.g. 3. In 2.2.4 the default values for these settings are somewhat strict and may cause connection errors for the monitor. |
| Comment by Esa Korhonen [ 2019-01-23 ] |
|
Closing for now. Reopen if issue is unsolved. |