High Connection Response Times over HTTPS

[MRedbourne]

Hey Folks,

I'm hoping someone here can help me. I'm not much of an apache (httpd) expert, and not very familiar with the performance tuning process. Some of our staff are reporting high connection wait times to the Apache webserver for Zabbix. (I've also experienced it, but always put it down to geographical distance and poor performance. However, even the local staff are having issues with connection times. I ran a couple tests, which I'll highlight the results of below:

1) Burpsuite (over VPN): Avg load time: 5500ms
2) Burpsuite (over Internet): Avg load time: 5500ms
3) ApacheBench (ZAB02 -> ZAB01 via Internet): See screenshot.
4) ApacheBench (ZAB02 -> ZAB01 via LAN): See screenshot.
5) Enabled "server-status", which shows request times nearing 6000ms.

I don't know what's wrong with the Apache installation. This is Zabbix 7.0.4 on RHEL9. The high connection times even over the LAN (ZAB01 and ZAB02 are adjacent to each other. Eg: 10.0.0.2 and 10.0.0.3 with no firewall) point to a config issue. This is RHEL 9.4 using Zabbix 7.0.4. I'm not sure what's wrong with it. Any ideas? It's a pretty normal set up with some traditional HTTPS hardening enabled.

[MRedbourne]

This has been resolved.

Cause: long DNS query times. Zabbix seems to open a new DB connection each time you click on something in the Web UI that re-queries the DB. Eg: Clicking a drop down is local, but clicking from "Data Collection > Hosts" to "Data Collection > Templates" causes a DB query. In our case of DB high availability, we query by DNS name, which gets conditionally forwarded to a set of external DNS servers (Azure), and doesn't appear to be cached at any point either.

Might be worth putting in a bug or feature request to re-use the active DB connection as long as it hasn't been terminated.