Today I had a small problem with my Zabbix server. I saw this error: Utilization of unreachable poller processes over 75%. The problem happens when Zabbix tries to connect to “dead” (offline) hosts, but it doesn’t have enough “workers” (processes) to monitor multiple hosts simultaneously. By default, Zabbix has only one worker for this task..
To fix this you need to add more workers in the configuration:
- Open the file: /etc/zabbix/zabbix_server.conf
- Find this line: # StartPollersUnreachable=1
- Uncomment it and set value to 10
StartPollersUnreachable=10
Restart Zabbix
systemctl restart zabbix-server.
Now, the problem should gone away.