We are currently testing Zabbix to see if it would be a good monitoring/graphing system for our infrastructure. I have the latest Zabbix (5.4.9) running, with a few switches and servers now being monitored. The problem we're running into is that whenever a switch is rebooted, has its firmware upgraded, or has its port configuration changed (e.g. a port going from 40G to 4x 10G), the problems and graphs no longer correspond to the original port. This also creates many bogus problem reports about ports going down, where in reality, Zabbix seems to have lost track of which port is which. As an example, see the included image: When I look at my problems list, and click on port swp7s1, it offers to show me the graph of the operational status of port swp6s2 instead! Note that all the problem reports listed here are bogus, and due to the internal bookkeeping in Zabbix getting confused by doing a port breakout on the switch in question.
Given that this is a significant issue with the usability and UI, I wonder if we're doing something wrong here, or how to fix this? We're also running monitoring on Cacti, and that doesn't mind at all in the same use cases.
Some more details: We are monitoring Mellanox SN2100 switches (running Cumulus), HP ProCurve switches (8212zl, 5412zl) and some SuperMicro blade switches and Linux hosts. All monitoring for the switches is done over SNMPv3. The Zabbix install is running on an up-to-date Debian Bullseye install (currently a VM as we are testing several monitoring platforms, and to learn the scaling requirements).
Given that this is a significant issue with the usability and UI, I wonder if we're doing something wrong here, or how to fix this? We're also running monitoring on Cacti, and that doesn't mind at all in the same use cases.
Some more details: We are monitoring Mellanox SN2100 switches (running Cumulus), HP ProCurve switches (8212zl, 5412zl) and some SuperMicro blade switches and Linux hosts. All monitoring for the switches is done over SNMPv3. The Zabbix install is running on an up-to-date Debian Bullseye install (currently a VM as we are testing several monitoring platforms, and to learn the scaling requirements).