New configuration on server not pushed to proxy?

[vanessa]

Hello, first post.

I have a proxy (192.168.0.5) that's receiving connections from an active zabbix agent2 (192.168.0.13).

I have added this agent to the zabbix-server in the ZabbixUI, where the name (plants-agent) and ip (192.168.0.13) of the agent definitely is set correctly (agent interface, default port). I'm using various templates such as ICMP ping and linux active agent2.

The proxy-ip is correct on the zabbix-agent2 conf, where Server=192.168.0.5 and ServerActive=192.168.0.5 is set to the ip of the proxy. Hostname is set to plants-agent.

Historically, doing this has worked out. No problem at all. I'm really certain that every ascii character and number is correct here.

I have tripple-checked that everything is correct (and others have checked too).

However, recently, the logs on the proxies have started to show the following every time I try to set up a new agent (it has never been a problem before). This problem seems to be for all proxies and agents and new hosts in general:

/var/log/zabbix/zabbix_proxy.log (on ) logs the following:
cannot send list of active checks to "192.168.0.13": host [plants-agent] not found.

I've tried to force the proxy (through the zabbix UI) to refresh the config. I've restarted the zabbix-agent and proxy. The only thing that helps here is restarting the server.

Any help to figure out what the problem is would be very much appreciated!

Br, V

[cyber]

Did you also set this host to be monitored by proxy in GUI?

[vanessa]

Did you also set this host to be monitored by proxy in GUI?

Yes I did.

I did everything from scratch. Uninstalled everything. Still same problem.
I have a database cluster, but it is consistent and working (long story short).

Two redundant zabbix-servers installed according to the documentation.

The problem has reappeared.

[vanessa]

The host table is populated on the zabbix_server table, but not on the proxy_table unless the server is rebooted (which means that the secondary HA zabbix-server takes over)... then the host table on the zabbix_proxy gets populated again.

I shut down the secondary server (so using the main again), and it's working again.. but most likely temprarily (again).

There are a lot of new bug reports regarding similar problems.

[vanessa]

Possibly a bug somewhere regarding the config cache where only restarting the service helps.

[cyber]

Ok, you did not mention HA servers... Are both of them nicely defined in proxy config? https://www.zabbix.com/documentation...x_proxy#server

[ISiroshtan]

Did you by chance try to manually reload configuration cache of Zabbix server instead of restarting? "zabbix_server -R config_cache_reload"

When you force refreshing proxy configuration from UI, did you check proxy logs? It should have 2 lines:
"force reloading of the configuration cache" - should indicate the request was received by proxy and it will attempt to get and load new configuration
"received configuration data from server at 'server_name', datalen xxxxxx" - this should indicate that configuration was recieved. If datalen is 0 - means there is no difference in current proxy configuration from currently cached zabbix_server configuration
(at least that was log flow in 6.4)

I know people don't usually play with CacheUpdateFrequency parameter in zabbix_server.conf, but behaviour seems similar to what would be happening if it was set to some relatively big value (we have it set to 1 hour in our claster and I can see similarities to what I experienced when starting to work with it)

P.S. not tried to use Zabbix 7, so might indeed be some bug