search

sleepka / zabbix-kubernetes-monitoring

Zabbix Kubernetes monitoring
GNU General Public License v3.0
57 stars 45 forks source link

Info about Zabbix monitoring of Kubernetes #9

Open lslamp opened 3 years ago

lslamp commented 3 years ago

I am very new to Kubernetes. I have followed various web pages that explain and provide details on how to set up a cluster with minecraft and monitoring.

After very much frustration and misdirection and not understanding. I eventually managed to get the cluster up and running. No matter what I did I could not understand how to get the monitoring working. I then found your description on how to add monitoring. After importing the template, I saw that there was a number of entries under the discovery tab. All are enabled apart from the one for the cluster probe.

image

After a while the discovery process seemed to add very many entries under the zabbix server. Looking at the config for the zabbix server, under the Applications tab, I see 8 Applications with a number of items next to each entry.

image

The list of items is attached. list.txt

Looking at the items, I see no monitor that provides me information that I think is useful, that might just be my ignorance.

Can you please tell me if I am missing something. I am interested in understanding what and how the cluster is set up. Individual RAM and CPU on each node. Combined RAM and CPU of all nodes together. utilization of RAM and CPU application load and distribution across the cluster.

Any help/guidance is very much apprecuated. Thanks Lawrence

jpuris commented 3 years ago

Amount of Memory and CPU is not within the scope of the items this project collects. What it does collect is more of what you can see when you describe a kubernetes "object".

For example, if you'd kubectl describe pod <pod name> -n <namespace> you'd see bunch of metrics about it, such as

Memory pressure: False
Name: <pod name>
Namespace: <namespace>
Restart count: 0
...

These are the metrics it collects via kubernetes API.

If you want to monitor the cpu / memory metrics, then I suggest you to use something that retrieves metrics from cadvisor or metrics server

I've seen this done easily with Prometheus + Grafana setups. Take a look at https://www.fosstechnix.com/kubernetes-cluster-monitoring-with-prometheus-and-grafana/ or any other of the many guides out there.

For grafana dashboards you can check Grafana dashboards or make your own.