Hello everyone, I have been dealing with a tricky monitoring issue on our Zabbix 7.0 LTS server and could really use some perspective from the community. We recently set up a collection of web scenarios and HTTP agent items to track uptime and response latency for our external download endpoints. However, during busy traffic periods our automated step checks against this resource regularly trip curl error 28 operation timeouts, even though manual checks from the terminal immediately after report acceptable response times. In connection with these timeouts, we are dealing with continuous trigger flapping that resolves itself within one polling cycle, creating excessive noise in our incident logs and alert channels. Furthermore, the response time graphs show unexpected spikes during the transfer phase that do not match the CPU load or network interface traffic reported by the Zabbix agent on the host itself. I would appreciate any insight into whether this points to contention within the web poller worker processes or if there are specific timeout and retry parameters we should adjust to avoid these false alarms without missing genuine downtime.
Comment