-- SOLVED IN ZABBIX 5.4.5 ---
Hi,
I recently upgraded a Zabbix environment running in docker containers (centos - mysql) from 5.2.2 to 5.4.0 (and 5.4.3).
Since the upgrade, I have issues with hosts that are shutdown at night.
When the hosts are started again, They are marked as unavailable in the gui, with errors like "interrupted system call zabbix/No route to host":
the agents connect to the zabbix server via a zabbix proxy. From within the proxy container, I can ping and telnet the monitored host.
When I perform a test execution of the zabbix agent ping check, I get result 1 (available), and from the moment I click "Execute now" on this check, the host becomes available again. This is also visible in the zabbix proxy logs.
So even though the host is reachable, Zabbix somehow doesn't figure it out automatically since the upgrade.
After I performed some more tests, the problem seems to be the Zabbix proxy. When the issue happens, a zabbix proxy restart resolves it.
All Unreachable/unavailable parameters are set to the default values, but as this happens randomly, it doesn't seem to be a configuration problem.
# UnavailableDelay=60
# UnreachablePeriod=45
# UnreachableDelay=15
Hoping another upgrade could solve the issue, I upgraded from 5.4.0 to 5.4.3 (ubuntu containers), issue still persists. Zabbix proxy doesn't notice the servers are back up until I restart the proxy, or manually execute the zabbix agent ping check.
As the only change is the Zabbix version, this seems to be a bug starting in Zabbix 5.4.0 and higher.
--- EDIT ---
The issue is solved in zabbix 5.4.5!
Hi,
I recently upgraded a Zabbix environment running in docker containers (centos - mysql) from 5.2.2 to 5.4.0 (and 5.4.3).
Since the upgrade, I have issues with hosts that are shutdown at night.
When the hosts are started again, They are marked as unavailable in the gui, with errors like "interrupted system call zabbix/No route to host":
the agents connect to the zabbix server via a zabbix proxy. From within the proxy container, I can ping and telnet the monitored host.
When I perform a test execution of the zabbix agent ping check, I get result 1 (available), and from the moment I click "Execute now" on this check, the host becomes available again. This is also visible in the zabbix proxy logs.
So even though the host is reachable, Zabbix somehow doesn't figure it out automatically since the upgrade.
After I performed some more tests, the problem seems to be the Zabbix proxy. When the issue happens, a zabbix proxy restart resolves it.
All Unreachable/unavailable parameters are set to the default values, but as this happens randomly, it doesn't seem to be a configuration problem.
# UnavailableDelay=60
# UnreachablePeriod=45
# UnreachableDelay=15
Hoping another upgrade could solve the issue, I upgraded from 5.4.0 to 5.4.3 (ubuntu containers), issue still persists. Zabbix proxy doesn't notice the servers are back up until I restart the proxy, or manually execute the zabbix agent ping check.
As the only change is the Zabbix version, this seems to be a bug starting in Zabbix 5.4.0 and higher.
--- EDIT ---
The issue is solved in zabbix 5.4.5!