-
Improvement
-
Resolution: Unresolved
-
Minor
-
None
-
Future Dev
We have come across performance issues in cleaning up the session caches which are somewhat resolved in MDL-82511. But the whole design of how this works is kinda horrible. It would be much better if each cache store could self manage how it cleans up its own data. Specifically for redis, it would just add an expire for each piece of data and let redis do the hard work and cache_cron_task should do nothing.
- Discovered while testing
-
MDL-82511 cache_cron_task causes a failover on redis/sentinel clusters
-
- Waiting for integration review
-