Open myskaludek opened 1 year ago
We can work around this by defining a custom health check override using config-cm :)
# helm values.yaml
configs:
cm:
resource.customizations: |
flink.apache.org/FlinkDeployment:
health.lua: |
health_status = {}
if obj.status ~= nil and obj.status.jobManagerDeploymentStatus ~= nil then
if obj.status.jobManagerDeploymentStatus == "READY" then
health_status.status = "Healthy"
return health_status
end
if obj.status.jobManagerDeploymentStatus == "DEPLOYED_NOT_READY" or obj.status.jobManagerDeploymentStatus == "DEPLOYING" then
health_status.status = "Progressing"
health_status.message = "Waiting for deploying"
return health_status
end
if obj.status.jobManagerDeploymentStatus == "ERROR" then
health_status.status = "Degraded"
health_status.message = obj.status.reconciliationStatus.error
return health_status
end
end
health_status.status = "Progressing"
health_status.message = "Waiting for Flink operator"
return health_status
I add costumization to configmap and i work as expected but will be nice to have good healthcheck in upstream :-)
Reopening in case someone wants to pick up the PR and resolve unit test issues.
I think https://github.com/argoproj/argo-cd/pull/15065 should also close this issue as well.
Checklist:
argocd version
.Describe the bug
App Health stale on
Progressing
even if FlinkDeployment is managed by operator and running properly. Problem is in Custom Health Check https://github.com/argoproj/argo-cd/blob/e05298b9c6ab8610104271fa8491f019fee3c587/resource_customizations/flink.apache.org/FlinkDeployment/health.lua#L4Apache flink kubernetes operator not reporting
status.reconciliationStatus.success
instead is reportingstatus.jobManagerDeploymentStatus: READY
To Reproduce
Just deploy FlinkDeployment with Apache Flink Kubernetes Operator v 1.2.0 and above
Expected behavior
Healt is
Healty
when flinkJobDeployment is runningScreenshots
Version
Logs