App Health on FlinkDeployment stale on progressing

myskaludek commented 1 year ago

Checklist:

[x] I've searched in the docs and FAQ for my answer: https://bit.ly/argocd-faq.
[x] I've included steps to reproduce the bug.
[x] I've pasted the output of 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#L4

Apache flink kubernetes operator not reporting status.reconciliationStatus.success instead is reporting status.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 running

Screenshots

Version

argocd: v2.6.6+unknown
  BuildDate: 2023-03-17T11:36:14Z
  GitCommit: 
  GitTreeState: 
  GitTag: 2.6.6
  GoVersion: go1.20.2
  Compiler: gc
  Platform: linux/amd64
argocd-server: v2.6.3+e05298b
  BuildDate: 2023-02-27T14:40:19Z
  GitCommit: e05298b9c6ab8610104271fa8491f019fee3c587
  GitTreeState: clean
  GoVersion: go1.18.10
  Compiler: gc
  Platform: linux/amd64
  Kustomize Version: v4.5.5 2022-05-20T20:25:40Z
  Helm Version: v3.10.3+g835b733
  Kubectl Version: v0.24.2
  Jsonnet Version: v0.19.1

Logs

zezaeoh commented 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

myskaludek commented 1 year ago

I add costumization to configmap and i work as expected but will be nice to have good healthcheck in upstream :-)

crenshaw-dev commented 1 year ago

Reopening in case someone wants to pick up the PR and resolve unit test issues.

dlemfh commented 11 months ago

I think https://github.com/argoproj/argo-cd/pull/15065 should also close this issue as well.

argoproj / argo-cd

App Health on FlinkDeployment stale on progressing #12993