grafana-operator: [Bug] 'deployment not ready' is reported and hang 'Phase: reconciling' status for grafana instance using v4.4.1 operator
Describe the bug ‘deployment not ready’ is reported in grafana instance and event using v4.4.1 operator, after operator is installed and grafana cr instance is created.
Version v4.4.1
To Reproduce Not a full reproduction but this is how I found it:
- Installed operator v4.4.1 to a single namespace
grafana
in OCP operator hub. - Created CR for an instance
- check instance, found
Warning ProcessingError 105m (x4188 over 14h) GrafanaDashboard deployment not ready
- in OCP console web ui, go to Networking > Route and click
grafana-route
, grafana UI loading failed.
Expected behavior No error found and all cr up.
Runtime (please complete the following information):
- OS: Linux
- Grafana Operator Version v4.4.1
- Environment: OCP 4.10.3
- Deployment type: Installed operator v4.4.1 to a single namespace
grafana
in OCP operator hub.
Additional context
$ oc get pod
NAME READY STATUS RESTARTS AGE
grafana-deployment-6c6d7cc447-dx6kh 1/1 Running 0 104m
grafana-operator-controller-manager-6444c498b5-dgzrx 2/2 Running 0 14h
$ oc get grafanas.integreatly.org
NAME AGE
example-grafana 14h
$ oc describe grafanas.integreatly.org example-grafana
...
Status:
Message: success
Phase: reconciling
Previous Service Name: grafana-service
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Warning ProcessingError 105m (x4188 over 14h) GrafanaDashboard deployment not ready
About this issue
- Original URL
- State: closed
- Created 2 years ago
- Reactions: 2
- Comments: 25
We are seeing this on Grafana Operator v4.4.1 in OpenShift.
Seems like the API Schema for GrafanaDashboard has the “labels” object, while the GrafanaDataSource and GrafanaNotificationChannel API Schema does not on integreatly.org | v1alpha1
A notificationchannel deployment with label app=grafana would result in instant reconcilling of object.
Redeploying the deployment without the label app=grafana would be successful.
We need the app=grafana was that is matchlabel for our operator to find the monitoring object for multi-namespaces.
I don’t see how this cannot be resolvable if we simply add the label object into all the Grafana Objects API Schema.
My bad. Did not noticed that setting dashboardLabelSelector is mandatory. After that dashboards and notifications are loaded successfully.