Uploaded image for project: 'Flink'
  1. Flink
  2. FLINK-25963

FLIP-212: Introduce Flink Kubernetes Operator

    XMLWordPrintableJSON

Details

    Attachments

      1.
      		 Initial Kubernetes Operator Prototype contribution Sub-task Closed Gyula Fora  
      2.
      		 Separate job and deployment errors in FlinkDeployment status Sub-task Closed Gyula Fora  
      3.
      		 Implement shared validation logic for FlinkDeployment objects Sub-task Closed Gyula Fora  
      4.
      		 Create webhook REST api test Sub-task Closed Nicholas Jiang  
      5.
      		 Create controller test for stateful/stateless upgrade flow Sub-task Closed Gyula Fora  
      6.
      		 Improve JobStatus tracking and handle different job states Sub-task Closed Unassigned  
      7.
      		 Support last-state upgrade mode Sub-task Closed Yang Wang  
      8.
      		 Integrate flink-kubernetes-operator repo with CI/CD Sub-task Closed Yang Wang  
      9.
      		 Containers Should Not Run As Root Sub-task Closed Matyas Orhidi  
      10.
      		 Refactor FlinkUtils#getEffectiveConfig into smaller pieces Sub-task Closed Biao Geng  
      11.
      		 Use the same enum for expected and observed jobstate (JobState / JobStatus.state) Sub-task Closed Biao Geng  
      12.
      		 Support manual savepoint triggering in the operator Sub-task Closed Matyas Orhidi  
      13.
      		 Support kubernetes-operator metrics using the Flink metric system Sub-task Closed Matyas Orhidi  
      14.
      		 Commit generated CRD into Git repo Sub-task Closed Matyas Orhidi  
      15.
      		 Publish Kubernetes operator to container registry Sub-task Closed Márton Balassi  
      16.
      		 Make 'replicas' work in JobManager Spec Sub-task Closed Biao Geng  
      17.
      		 Track and cap retries in ReconciliationStatus Sub-task Closed Matyas Orhidi  
      18.
      		 Document metrics configuration for Prometheus Sub-task Closed Matyas Orhidi  
      19.
      		 Support watching specific namespace for FlinkDeployments Sub-task Closed Gyula Fora  
      20.
      		 Reconciliation should try to start job when not already started or move to permanent error Sub-task Closed Thomas Weise  
      21.
      		 Revisit serviceAccount and taskSlots direct fields in CRD Sub-task Closed Thomas Weise  
      22.
      		 Remove need for flink-operator clusterrole Sub-task Closed Márton Balassi  
      23.
      		 Control Logging Behavior in Flink Deployments Sub-task Closed Matyas Orhidi  
      24.
      		 Move the Operator Env to the common Utils Sub-task Closed WenJun Min  
      25.
      		 Deletion should remove HA related configmaps also Sub-task Closed Gyula Fora  
      26.
      		 Avoid load flink conf each reconcile loop Sub-task Closed WenJun Min  
      27.
      		 Introduce the webhook config to free the environment options Sub-task Closed Unassigned  
      28.
      		 Revisit the create of RestClusterClient Sub-task Closed WenJun Min  
      29.
      		 Move sanity check in FlinkService#cancelJob to DefaultDeploymentValidator Sub-task Closed Nicholas Jiang  
      30.
      		 Make Flink cluster communication asynchronous Sub-task Closed Sandor Kelemen  
      31.
      		 Extract Reconciler interface Sub-task Closed WenJun Min  
      32.
      		 Make some option of operator configurable Sub-task Closed WenJun Min  
      33.
      		 Add validation check of num of JM replica Sub-task Closed Biao Geng  
      34.
      		 Cleanly separate validator, observer and reconciler modules Sub-task Closed Gyula Fora  
      35.
      		 SharedIndexInformer should respect watched namespaces Sub-task Closed Gyula Fora  
      36.
      		 Clean up webhook jar and dependency management Sub-task Closed Nicholas Jiang  
      37.
      		 Improve operator logging Sub-task Closed Gyula Fora  
      38.
      		 Deleting the operator while jobs are running causes the jobs to fail Sub-task Closed Márton Balassi  
      39.
      		 Introduce Savepoint object in JobStatus Sub-task Closed Matyas Orhidi  
      40.
      		 Observer should support JobManager deployment crashed or deleted externally Sub-task Closed Thomas Weise  
      41.
      		 Introduce flink-kubernetes-shaded to avoid overlapping classes Sub-task Closed Yang Wang  
      42.
      		 Last state upgrade mode should allow reconciliation regardless of job and deployment status Sub-task Closed Gyula Fora  
      43.
      		 Webhook should only validate on /validate endpoint end log errors for others Sub-task Closed Nicholas Jiang  
      44.
      		 Reconsider setting generationAwareEventProcessing = true Sub-task Closed Gyula Fora  
      45.
      		 Trigger the updateControl when the FlinkDeployment have changed Sub-task Closed WenJun Min  
      46.
      		 Ability to restart deployment w/o spec change Sub-task Closed WenJun Min  
      47.
      		 Extract Observer Interface Sub-task Closed WenJun Min  
      48.
      		 Try to use @EnableKubernetesMockClient(crud = true) in controller test Sub-task Closed Nicholas Jiang  
      49.
      		 Clean termination of FlinkDeployment Sub-task Closed Gyula Fora  
      50.
      		 Savepoint trigger/tracking improvements Sub-task Closed Matyas Orhidi  
      51.
      		 Re-schedule reconcile more often until job is in ready state Sub-task Closed Gyula Fora  
      52.
      		 Allow definining Operator configuration in Helm chart Values Sub-task Closed Gyula Fora  
      53.
      		 Avoid state loss when switching to last-state upgrade mode Sub-task Closed Yang Wang  
      54.
      		 Check if JM can serve rest api calls every time before reconcile Sub-task Closed Biao Geng  
      55.
      		 Mark CRD classes experimental Sub-task Closed Nicholas Jiang  
      56.
      		 Consider changing flinkVersion to enum type or removing it completely Sub-task Closed Gyula Fora  
      57.
      		 Add startTime in JobStatus Sub-task Closed Nicholas Jiang  
      58.
      		 Improve the observe logic in SessionObserver Sub-task Closed Biao Geng  
      59.
      		 Specify EventSource when watching multiple namespaces Sub-task Closed Gyula Fora  
      60.
      		 Set ClusterIP service type when watching specific namespaces Sub-task Closed Gyula Fora  
      61.
      		 Introduce Ingress URL templating Sub-task Closed Matyas Orhidi  
      62.
      		 Remove .sec from OperatorConfigOptions and use Duration type config instead Sub-task Closed Nicholas Jiang  
      63.
      		 E2E tests should cover different watchNamespace scenarios Sub-task Closed Márton Balassi  
      64.
      		 Rethink the default reschedule reconcile loop Sub-task Closed Unassigned  
      65.
      		 Individual upgradeMode change does not trigger a reconciliation Sub-task Closed Unassigned  
      66.
      		 Link operator doc site to Flink Website Sub-task Closed ZhengYu Chen
      67.
      		 The batch job not work well with Operator Sub-task Closed Unassigned  
      68.
      		 Move jobs to suspended state before upgrading Sub-task Closed Gyula Fora  
      69.
      		 Output more status info for JobObserver Sub-task Closed Biao Geng  
      70.
      		 Add sanity check for state.savepoints.dir when using savepoint upgrade mode Sub-task Closed Nicholas Jiang  
      71.
      		 Align the helm chart version with the flink operator Sub-task Closed Gyula Fora  

      Activity

        People

          thw Thomas Weise
          thw Thomas Weise
          Votes:
          0 Vote for this issue
          Watchers:
          21 Start watching this issue

          Dates

            Created:
            Updated:
            Resolved:

            Time Tracking

              Estimated:
              Original Estimate - Not Specified
              Not Specified
              Remaining:
              Time Spent - 2h Remaining Estimate - 2h
              2h
              Logged:
              Time Spent - 2h Remaining Estimate - 2h
              2h