-
Change Request
-
Resolution: Unresolved
-
Medium
-
None
-
6.4.11rc1
-
None
When HA was introduced Zabbix was still doing a full config cache reload every time and it made sense to have the standby node not do this. On large installations this could cause additional load on the DB.
But this causes a problem that when failover happens, the standby node starts working only after config cache has been reloaded. On large installations this could take some time, sometimes 10 minutes or even more. This causes a very slow failover.
Now that we have much more efficient config reload schema, we might consider to let standby node populate configuration cache on startup, so that it's ready for work as soon as failover happens.
Periodic reload of changes could maybe happen less often than on active node.
Ideally this could be configurable.