New Agents and Proxies not working until Zabbix Server is restarted

[landsverka]

Hello,

We recently migrated our Zabbix database from MySQL to PostgreSQL 15 with TimescaleDB 2.10. Everything had appeared to go successfully but we are noticing a new issue. We are not able to collect data from new agents (Agent 2, haven't tried 1) or Zabbix Proxies until after we restart the Zabbix Server process.

Here is what we see in the log on the agent2:

2023/06/02 15:51:28.985314 [101] cannot connect to [192.168.5.15:10051]: 40C67FC2397F0000:error:0A000410:SSL routines:ssl3_read_bytes:sslv3 alert handshake failure:ssl/record/rec_layer_s3.c:1600:SSL alert number 40
2023/06/02 15:51:28.985363 [101] sending of heartbeat message for [dev-zabbix-proxy-4.dev.internal] started to fail​

After a restart of the Zabbix Server, these newly added agents and proxies work the way they are supposed to.

We are wondering if there is some Zabbix Server configuration parameter that maybe we messed up or something we can check in the database?

I appreciate any assistance! Thanks!

EDIT: Some additional information

This is the log we see on the server:

 4082:20230602:173742.611 cannot find requested PSK identity "PSK : dev-zabbix-proxy-4.dev.internal"
 4082:20230602:173742.611 failed to accept an incoming connection: from 192.168.5.44: TLS handshake set result code to 1: file ssl/t1_lib.c line 2827: error:14201076:SSL routines:tls_choose_sigalg:no suitable signature algorithm: TLS write fatal alert "handshake failure"​

I have inspected the database manually and I see the PSK in the host table that it says it cannot find.

[landsverka]

I should also mention, we notice the problem on both Zabbix 6.2.9 and 6.4.3 (in our dev and or production environments).

[LenR]

What is your CacheUpdateFrequency? You often don't "see" changes until the config cache is updated at this interval or by zabbix_server -R config_cache_reload. Same for proxies.

[landsverka]

What is your CacheUpdateFrequency? You often don't "see" changes until the config cache is updated at this interval or by zabbix_server -R config_cache_reload. Same for proxies.

Thank you for your response! CacheUpdateFrequency is unset, so it should be at it's default? I have also tried "zabbix_server -R config_cache_reload" to no effect. The only "fix" is to restart zabbix-server and then the new proxies/agents start working.

[landsverka]

Additionally we have noticed it with new SNMP based devices as well. It appears that the host table on the proxy is only updated immediately after a zabbix-server restart.

[landsverka]

Following back up here, sorry for triple posting. We have resolved this issue, as it turns out, we were missing a bunch of database triggers for the changelog table after our postgres migration. We have since added them and everything is working as it should. Shout out to the Zabbix Developers in Jira for pointing us in the right direction!!

[Voxie]

Following back up here, sorry for triple posting. We have resolved this issue, as it turns out, we were missing a bunch of database triggers for the changelog table after our postgres migration. We have since added them and everything is working as it should. Shout out to the Zabbix Developers in Jira for pointing us in the right direction!!

Can you point us in the right direction ? Or maybe the right Jira request ?

[Voxie]

Found it,