Is it just me or is the learning curve a lot greater with Zabbix? The error messages seem extremely vague or completely useless. The web GUI fields don’t have proper validation. They moved a lot of things around in the 6.4 version and now googling a solution gives me out of date info. The template network sensors are picking up about 20 ethernet interfaces on windows VMs in HyperV and I cant select just the one that I want to monitor (I guess I have to write my own sensor for that?).
I was demo’d Zabbix by a friend who has 39k+ sensors working on less hardware than my 1000 sensors use in PRTG, and the price difference is huge… So I really want this to work for me but I spent the whole day today feeling uneasy about it.
What are your guys thoughts?
PRTG feels old and clunky. We are getting spammed by a sensor up alert on snmp to a windows server and the sensor never goes down in the first place. The graphs are low quality. It uses so many system resources and warns you against using too many WMI or PS/Bash sensors. It struggles to guess when its overloaded and checking CPU/RAM/disk queue does nothing to help you guess. We have been needing to move from PRTG for a while.
For my needs I just want icmp, http, https, tcp, ssh. But IT requires server/switch/router resource metrics as well. Zabbix is mature, has better dashboard and graphs, has a dark theme, and very efficient with the system resources. We have multiple vendors that use it for monitoring and my friend has replaced PRTG with it for all their networks that they monitor and they look at far more than what we need to monitor. We’ve also looked at n-able but I’d prefer open source where possible.
I’m going to keep trying to get comfortable with it and when it comes to December I’m hoping we can opt to not renew our PRTG subscription.