Closed Aakcht closed 4 years ago
You are welcome to submit a PR with the proposed change. This is open source software after all ;)
@gerhard @dcorbacho is the regex we use at the moment intentional?
And to be clear, this plugin or its Grafana dashboards are not specific to this peer discovery example. It would be nice to support some common naming patterns out of the box but the example is just that, an example of peer discovery on Kubernetes with an explicit production non-suitability disclaimer.
@mkuratczyk is there a common node naming scheme you'd recommend?
Well assuming StatefulSet is used in Kubernetes, the node naming scheme will always look somewhat similar to the peer discovery example. I'll submit a PR then, I just want to be sure that it won't break something in non-kubernetes deployments.
After closer inspecion it turned out that the problem is only with -
in pods names in statefulsets, and I think that it is also should be possible to have .
in a nodename before pod number, so I added a PR that changes ^rabbit@\w+0
to ^rabbit@[\w.-]+0
.
Hello, the documentation at https://www.rabbitmq.com/prometheus.html says:
According to the documentation, node naming scheme used in K8S deployment example should work. The nodename there is
rabbit@$(MY_POD_NAME).$(K8S_SERVICE_NAME).$(MY_POD_NAMESPACE).svc.cluster.local
, so assuming$MY_POD_NAMESPACE
doesn't have any numbers in it , the colour labelling should work fine. But actually it doesn't work, because the regex used in the dashboards is /^rabbit@\w+0/ for zero node. Maybe it's possible to change the regex to something like/^rabbit@.*?0.*/
,/^rabbit@.*?1.*/
, etc ? With the change color labelling in the dashboards should work for with kubernetes deployments without additional changes in the dashboards.