wkulhanek
commented
6 years ago Hi @cake99 Are you mixing this container and the prometheus pod from OpenShift? Your mention of 'openshift-metrics' seems to indicate that. While you can install a Prometheus automatically into OCP it will do so in a Pod that uses oauth-proxies to secure both Prometheus and Alertmanager. The bigger problem then is that Grafana can't connect to Prometheus behind the proxy. :-(
As for the error message: that means that the user that's being used to run the pod is not authorized to access the openshift-metrics project.
If you are using the template from this repo it will set up a clusterresourcebinding that will give prometheus the correct permissions.
Hello,
I have followed the instructions and it seems overall ok (i have three nodes running the node_explorer containers, only missing the one on the master).
It constantly restarting. I have this in the logs:
FailedSync error determining status: rpc error: code = 2 desc = Error response from daemon: {"message":"devmapper: Unknown device 83da9c1200eae4ae9ac90f1c8c03fa5087b588f1e1d82fdde00cc4df5638d385"}
In the alert-proxy pod, I have this:
2018/03/16 12:22:43 oauthproxy.go:657: 10.131.0.44:50208 Cookie "_oauth_proxy" not present
| 2018/03/16 12:22:43 provider.go:347: authorizer reason: User "system:anonymous" cannot get namespaces in project "openshift-metrics"
Any ideas what I did wrong?
Thanks a lot!