Few days ago our host with Zabbix server (also containing MySQL DB used by Zabbix proxy service located on other host -- I khow this stupid) was shut down (by 'shutdown' command, so DB is ok) for a few hours.
Proxy service also died after a while (no surprise).
After Zabbix server get back online & proxy service was restarted on proxy machine, proxy stops to send updates for monitored items to Zabbix server. It collects data from monitored hosts, it puts them into DB, it even read configuration changes from main Zabbix server, but status for hosts/items monitored via proxy are not being updated anymore.
Direct checks by Zabbix server itself are fine. Logs are clean. tcpdump shows traffic between both proxy & server and proxy & monitored hosts. From strace on zabbix_proxy I see regular heartbeat and configuration changes queries from proxy to server.
I tried to restart both proxy and server, I tried to restart all agents, I tried to delete & add back monitored hosts, I tried to manually update 'nextcheck' value for items monitored via proxy to future date (it stuck in queue time where server was shut down) -- no effect at all except now there are no timed out checks in queue.
Any thoughts?
Proxy service also died after a while (no surprise).
After Zabbix server get back online & proxy service was restarted on proxy machine, proxy stops to send updates for monitored items to Zabbix server. It collects data from monitored hosts, it puts them into DB, it even read configuration changes from main Zabbix server, but status for hosts/items monitored via proxy are not being updated anymore.
Direct checks by Zabbix server itself are fine. Logs are clean. tcpdump shows traffic between both proxy & server and proxy & monitored hosts. From strace on zabbix_proxy I see regular heartbeat and configuration changes queries from proxy to server.
I tried to restart both proxy and server, I tried to restart all agents, I tried to delete & add back monitored hosts, I tried to manually update 'nextcheck' value for items monitored via proxy to future date (it stuck in queue time where server was shut down) -- no effect at all except now there are no timed out checks in queue.
Any thoughts?
Comment