search

KjellWolf / CW-Zabbix-Templates

Zabbix Templates for MinIO
2 stars 0 forks source link

Separate templates for cluster and node #3

Open GhaziTriki opened 2 months ago

GhaziTriki commented 2 months ago

Hello,

Like many other platforms, monitoring of cluster and node metrics are separate. The node metrics should be available only on the node and the cluster metrics should be available in the load balancer.

KjellWolf commented 2 months ago

Hey,

what about this: I rename the current template "Single_node_Template[...]" for testclusters and such.

And then I try to seperate cluster from node metrics.

Do you have any ideas how I should categorize between cluster metrics and node metrics to make this consistend?

If you look in the RAW data you see metrics that are x times there depending on the Node number other data thats just once there, but is reportet as the node the LB has chosen for the request from zabbix.

Data like How many drives are online or offline for example are Cluster metrics, but are reported by every node in the Prometheus data.

GhaziTriki commented 2 months ago

Hello,

Good but hard question. Need to dig further into exposed metrics and how prometheus/grafana handles it. Isn't it a good start?

KjellWolf commented 3 weeks ago

Yea, how other projects handle the separation would be ideal.

Maybe I get some prework done here too in the next few days.