grafana-operator
grafana-operator copied to clipboard
Grafana Operator Internal Dashboard
Is your feature request related to a problem? Please describe.
No
Describe the solution you'd like
After some searching for a public dashboard for the grafana operator internal metrics, i was not able to find one.
Is there any already existing?
Describe alternatives you've considered
Building one (when there is some spare time)
We haven't built any dashboard for the grafana-operator. But we would love PR:s in this area :)
I was playing around with potential dashboards here and came up with the following.
I imported this dashboard for controller-runtime metrics https://grafana.com/grafana/dashboards/12122
It didn't show anything initially, but with a change to the 'instance' variable and the queries to use the service
label instead, it might give a good starting point. (label_values(controller_runtime_active_workers, service)
)
Note that there's nothing grafana operator specific in the dashboard.
It took quite a bit of setup to get to this point from zero, so there's probably some readme updates I can make for giving permission to prometheus to scrape grafana operator metrics (via the rbac proxy), setting up the data source, and an example servicemonitor.
Will wait for feedback before creating a PR.
I really like this approach, did you test this with multiple grafana operators in the same cluster?
In my use case i provide several grafana instances with their own operator (unfortunatly no multiple instances support from Operator).
Since i have workwise now a bit more time i will try to take this approach and combine it with the grafana metrics board.
did you test this with multiple grafana operators in the same cluster
No, I haven't. That might require another template variable for namespace
Do follow up with how your approach goes
Hey @1naboki1 , did you get some time to combine this with the grafana metrics board?
We are currently working on a dashboard for the Operator. We would like to define some goals and requirements that are useful for assessing the operator state.
Goals:
-
assess the health of the Operator (not the Operand): 1.1. figure out if other processes (or users) are interfering with the Operator (this could cause resources to constantly change and restore). 1.2. Memory and CPU usage over time. 1.3. Number of garbage collections happening. If this number goes up, it could indicate that the Operator is struggling for memory. 1.4. Network traffic from the Operator to Grafana and failed requests. This could indicate that Grafana itself is not healthy.
-
Overview of managed resources 2.1. get an overview of all Operands (how many Grafana instances, dashboards, data sources and folders is the operator managing and where are they located). This might require new metrics or additional labels in existing metrics.
Does that sounds reasonable?
ping @ThirumlaDevi