Any idea where to start to research ? Does it look like some background job is taking all those pollers ?
This is Zabbix 6.4 running in HA cluster (2 nodes)
Attached Files
-
Any idea where to start to research ? Does it look like some background job is taking all those pollers ?
This is Zabbix 6.4 running in HA cluster (2 nodes)
-
A 15 hour cycle is a little strange.
Are the servers or the database(s) VMs that might be in a storage pool with some other system that is using heavy I/O at the times when these issues happen?
Have you increased the setting for "StartPollers" in /etc/zabbix/zabbix_server.conf (or some local config file that gets included by zabbix_server.conf)? If you haven't and it's still using the default, considering increasing it by a small amount. Small changes, observe, and then potentially additional small changes only if needed.
Are you running the latest point release in the 6.4.x series, or are you a few minor updates behind? -
Yes, we did. Pollers ar eat ~60% except those strange peaks. The majority of our hosts are behind active proxies so if I understand correctly, these are processed by trappers. We have some passive hosts that we pull data directly from but their amount is quite low.
What else could pollers be doing to stay so busy?
We are running Zabbix 6.4.19
Comment
Comment