maxscale.cnf
|
Relates |
|||
| relates to |
|
infinite memory usage, possible leak | Closed |
[MXS-2430] A possible memory leak in version Created: 2019-04-11 Updated: 2019-04-17 Resolved: 2019-04-17 |
|
| Status: | Closed |
| Project: | MariaDB MaxScale |
| Component/s: | N/A |
| Affects Version/s: | 2.3.4 |
| Fix Version/s: | N/A |
| Type: | Bug | Priority: | Major |
| Reporter: | Dmitry | Assignee: | markus makela |
| Resolution: | Not a Bug | Votes: | 0 |
| Labels: | None | ||
| Environment: |
Red Hat Enterprise Linux Server release 7.6 (Maipo) |
||
| Attachments: |
maxscale.cnf
|
||||||||
| Issue Links: |
|
||||||||
| Description |
|
I am observing a steady increase in memory utilization by the maxscale process while it is under load. The memory utilization seems to be stable while the process is idling or under a low load. Looking at the periodic snapshots of top while the process is under load, the memory utilization increases by 1 to 2G every 30 minutes. The process is configured to manage 2 database clusters. disable_sescmd_history is set to true The full config is attached |
| Comments |
| Comment by markus makela [ 2019-04-15 ] |
|
If you add query_classifier_cache_size=0 under the [maxscale] section in maxscale.cnf, does it solve the problem? |
| Comment by Dmitry [ 2019-04-15 ] |
|
Thank you for the suggestion. I have looked for cache related parameters, but somehow missed this one. I am testing the change. |
| Comment by Dmitry [ 2019-04-16 ] |
|
That did it. I apologize for misinterpreting a feature as a memory leak. Thank You. |
| Comment by markus makela [ 2019-04-16 ] |
|
To verify that it works correctly, please try it with a lower value e.g. query_classifier_cache_size=100M. |
| Comment by Dmitry [ 2019-04-17 ] |
|
Verified. Memory consumption with 100M is also stable. |