程序包 org.apache.flink.kubernetes.operator.reconciler

类 ReconciliationUtils

  • public class ReconciliationUtils
extends java.lang.Object
    Reconciliation utilities.

    • 方法概要

      所有方法 静态方法 具体方法 
      修饰符和类型 方法 说明
      static <SPEC extends org.apache.flink.kubernetes.operator.api.spec.AbstractFlinkSpec>
      boolean      		 applyValidationErrorAndResetSpec​(FlinkResourceContext<? extends org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<SPEC,​?>> ctx, java.lang.String validationError)
      In case of validation errors we need to (temporarily) reset the old spec so that we can reconcile other outstanding changes, instead of simply blocking.
      static void checkAndUpdateStableSpec​(org.apache.flink.kubernetes.operator.api.status.CommonStatus<?> status)
      Checks the status and if the corresponding Flink job/application is in stable running state, it updates the last stable spec.
      static void clearLastReconciledSpecIfFirstDeploy​(org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<?,​?> resource)
      Clear last reconciled spec if that corresponds to the first deployment.
      static <T> T clone​(T object)  
      static <SPEC extends org.apache.flink.kubernetes.operator.api.spec.AbstractFlinkSpec>
      SPEC      		 getDeployedSpec​(org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<SPEC,​?> deployment)  
      static java.lang.Long getUpgradeTargetGeneration​(org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<?,​?> resource)
      Get spec generation for the current in progress upgrade.
      static boolean isJobInTerminalState​(org.apache.flink.kubernetes.operator.api.status.CommonStatus<?> status)  
      static boolean isJobRunning​(org.apache.flink.kubernetes.operator.api.status.CommonStatus<?> status)  
      static boolean isUpgradeModeChangedToLastStateAndHADisabledPreviously​(org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<?,​?> flinkApp, org.apache.flink.configuration.Configuration observeConfig)  
      static java.time.Duration rescheduleAfter​(org.apache.flink.kubernetes.operator.api.status.JobManagerDeploymentStatus status, org.apache.flink.kubernetes.operator.api.FlinkDeployment flinkDeployment, FlinkOperatorConfiguration operatorConfiguration)  
      static <STATUS extends org.apache.flink.kubernetes.operator.api.status.CommonStatus<?>,​R extends org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<?,​STATUS>>
      io.javaoperatorsdk.operator.api.reconciler.ErrorStatusUpdateControl<R>      		 toErrorStatusUpdateControl​(FlinkResourceContext<R> ctx, java.lang.Exception e, StatusRecorder<R,​STATUS> statusRecorder)
      Update the resource error status and metrics when the operator encountered an exception during reconciliation.
      static <SPEC extends org.apache.flink.kubernetes.operator.api.spec.AbstractFlinkSpec,​STATUS extends org.apache.flink.kubernetes.operator.api.status.CommonStatus<SPEC>,​R extends io.fabric8.kubernetes.client.CustomResource<SPEC,​STATUS>>
      io.javaoperatorsdk.operator.api.reconciler.UpdateControl<R>      		 toUpdateControl​(FlinkOperatorConfiguration operatorConfiguration, R current, R previous, boolean reschedule)  
      static <SPEC extends org.apache.flink.kubernetes.operator.api.spec.AbstractFlinkSpec>
      void      		 updateAfterScaleUp​(org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<SPEC,​?> target, org.apache.flink.configuration.Configuration deployConfig, java.time.Clock clock, FlinkService.ScalingResult scalingResult)  
      static void updateForReconciliationError​(FlinkResourceContext ctx, java.lang.Throwable error)  
      static <SPEC extends org.apache.flink.kubernetes.operator.api.spec.AbstractFlinkSpec>
      void      		 updateLastReconciledSavepointTriggerNonce​(org.apache.flink.kubernetes.operator.api.status.SavepointInfo savepointInfo, org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<SPEC,​?> target)  
      static <SPEC extends org.apache.flink.kubernetes.operator.api.spec.AbstractFlinkSpec>
      void      		 updateReconciliationMetadata​(org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<SPEC,​?> resource)  
      static <SPEC extends org.apache.flink.kubernetes.operator.api.spec.AbstractFlinkSpec>
      void      		 updateStatusBeforeDeploymentAttempt​(org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<SPEC,​?> target, org.apache.flink.configuration.Configuration conf)  
      static <SPEC extends org.apache.flink.kubernetes.operator.api.spec.AbstractFlinkSpec>
      void      		 updateStatusBeforeDeploymentAttempt​(org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<SPEC,​?> target, org.apache.flink.configuration.Configuration conf, java.time.Clock clock)
      Update status before deployment attempt of a new resource spec.
      static void updateStatusForAlreadyUpgraded​(org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<?,​?> resource)
      Updates status in cases where a previously successful deployment wasn't recorded for any reason.
      static <SPEC extends org.apache.flink.kubernetes.operator.api.spec.AbstractFlinkSpec>
      void      		 updateStatusForDeployedSpec​(org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<SPEC,​?> target, org.apache.flink.configuration.Configuration conf)  
      static <SPEC extends org.apache.flink.kubernetes.operator.api.spec.AbstractFlinkSpec>
      void      		 updateStatusForDeployedSpec​(org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<SPEC,​?> target, org.apache.flink.configuration.Configuration conf, java.time.Clock clock)
      Update status after successful deployment of a new resource spec.

      • 从类继承的方法 java.lang.Object

        clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

    • 构造器详细资料

      • ReconciliationUtils

        public ReconciliationUtils()

    • 方法详细资料

      • updateStatusForDeployedSpec

        public static <SPEC extends org.apache.flink.kubernetes.operator.api.spec.AbstractFlinkSpec> void updateStatusForDeployedSpec​(org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<SPEC,​?> target,
                                                                                                                              org.apache.flink.configuration.Configuration conf,
                                                                                                                              java.time.Clock clock)
        Update status after successful deployment of a new resource spec. Existing reconciliation errors will be cleared, lastReconciled spec will be updated and for suspended jobs it will also be marked stable.

        For Application deployments TaskManager info will also be updated.

        类型参数:
        SPEC - Spec type.
        参数:
        target - Target Flink resource.
        conf - Deployment configuration.
        clock - Clock for getting time.

      • updateStatusForDeployedSpec

        @VisibleForTesting
public static <SPEC extends org.apache.flink.kubernetes.operator.api.spec.AbstractFlinkSpec> void updateStatusForDeployedSpec​(org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<SPEC,​?> target,
                                                                                                                              org.apache.flink.configuration.Configuration conf)

      • updateStatusBeforeDeploymentAttempt

        public static <SPEC extends org.apache.flink.kubernetes.operator.api.spec.AbstractFlinkSpec> void updateStatusBeforeDeploymentAttempt​(org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<SPEC,​?> target,
                                                                                                                                      org.apache.flink.configuration.Configuration conf,
                                                                                                                                      java.time.Clock clock)
        Update status before deployment attempt of a new resource spec. Existing reconciliation errors will be cleared, lastReconciled spec will be updated and reconciliation status marked UPGRADING.

        For Application deployments TaskManager info will also be updated.

        类型参数:
        SPEC - Spec type.
        参数:
        target - Target Flink resource.
        conf - Deployment configuration.
        clock - Clock for getting the current time.

      • updateStatusBeforeDeploymentAttempt

        @VisibleForTesting
public static <SPEC extends org.apache.flink.kubernetes.operator.api.spec.AbstractFlinkSpec> void updateStatusBeforeDeploymentAttempt​(org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<SPEC,​?> target,
                                                                                                                                      org.apache.flink.configuration.Configuration conf)

      • updateAfterScaleUp

        public static <SPEC extends org.apache.flink.kubernetes.operator.api.spec.AbstractFlinkSpec> void updateAfterScaleUp​(org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<SPEC,​?> target,
                                                                                                                     org.apache.flink.configuration.Configuration deployConfig,
                                                                                                                     java.time.Clock clock,
                                                                                                                     FlinkService.ScalingResult scalingResult)

      • updateLastReconciledSavepointTriggerNonce

        public static <SPEC extends org.apache.flink.kubernetes.operator.api.spec.AbstractFlinkSpec> void updateLastReconciledSavepointTriggerNonce​(org.apache.flink.kubernetes.operator.api.status.SavepointInfo savepointInfo,
                                                                                                                                            org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<SPEC,​?> target)

      • updateForReconciliationError

        public static void updateForReconciliationError​(FlinkResourceContext ctx,
                                                java.lang.Throwable error)

      • clone

        public static <T> T clone​(T object)

      • toUpdateControl

        public static <SPEC extends org.apache.flink.kubernetes.operator.api.spec.AbstractFlinkSpec,​STATUS extends org.apache.flink.kubernetes.operator.api.status.CommonStatus<SPEC>,​R extends io.fabric8.kubernetes.client.CustomResource<SPEC,​STATUS>> io.javaoperatorsdk.operator.api.reconciler.UpdateControl<R> toUpdateControl​(FlinkOperatorConfiguration operatorConfiguration,
                                                                                                                                                                                                                                                                                                                                                R current,
                                                                                                                                                                                                                                                                                                                                                R previous,
                                                                                                                                                                                                                                                                                                                                                boolean reschedule)

      • rescheduleAfter

        public static java.time.Duration rescheduleAfter​(org.apache.flink.kubernetes.operator.api.status.JobManagerDeploymentStatus status,
                                                 org.apache.flink.kubernetes.operator.api.FlinkDeployment flinkDeployment,
                                                 FlinkOperatorConfiguration operatorConfiguration)

      • isUpgradeModeChangedToLastStateAndHADisabledPreviously

        public static boolean isUpgradeModeChangedToLastStateAndHADisabledPreviously​(org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<?,​?> flinkApp,
                                                                             org.apache.flink.configuration.Configuration observeConfig)

      • getDeployedSpec

        public static <SPEC extends org.apache.flink.kubernetes.operator.api.spec.AbstractFlinkSpec> SPEC getDeployedSpec​(org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<SPEC,​?> deployment)

      • isJobInTerminalState

        public static boolean isJobInTerminalState​(org.apache.flink.kubernetes.operator.api.status.CommonStatus<?> status)

      • isJobRunning

        public static boolean isJobRunning​(org.apache.flink.kubernetes.operator.api.status.CommonStatus<?> status)

      • applyValidationErrorAndResetSpec

        public static <SPEC extends org.apache.flink.kubernetes.operator.api.spec.AbstractFlinkSpec> boolean applyValidationErrorAndResetSpec​(FlinkResourceContext<? extends org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<SPEC,​?>> ctx,
                                                                                                                                      java.lang.String validationError)
        In case of validation errors we need to (temporarily) reset the old spec so that we can reconcile other outstanding changes, instead of simply blocking.

        This is only possible if we have a previously reconciled spec.

        For in-flight application upgrades we need extra logic to set the desired job state to running

        类型参数:
        SPEC - Spec type.
        参数:
        ctx - The current deployment context
        validationError - Validation error encountered for the current spec
        返回:
        True if the spec was reset and reconciliation can continue. False if nothing to reconcile.

      • toErrorStatusUpdateControl

        public static <STATUS extends org.apache.flink.kubernetes.operator.api.status.CommonStatus<?>,​R extends org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<?,​STATUS>> io.javaoperatorsdk.operator.api.reconciler.ErrorStatusUpdateControl<R> toErrorStatusUpdateControl​(FlinkResourceContext<R> ctx,
                                                                                                                                                                                                                                                                                                java.lang.Exception e,
                                                                                                                                                                                                                                                                                                StatusRecorder<R,​STATUS> statusRecorder)
        Update the resource error status and metrics when the operator encountered an exception during reconciliation.
        类型参数:
        STATUS - Status type.
        R - Resource type.
        参数:
        ctx - Flink Resource context
        e - Exception that caused the retry
        statusRecorder - statusRecorder object for patching status
        返回:
        This always returns Empty optional currently, due to the status update logic

      • getUpgradeTargetGeneration

        public static java.lang.Long getUpgradeTargetGeneration​(org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<?,​?> resource)
        Get spec generation for the current in progress upgrade.
        参数:
        resource - Flink resource.
        返回:
        The spec generation for the upgrade.

      • clearLastReconciledSpecIfFirstDeploy

        public static void clearLastReconciledSpecIfFirstDeploy​(org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<?,​?> resource)
        Clear last reconciled spec if that corresponds to the first deployment. This is important in cases where the first deployment fails.
        参数:
        resource - Flink resource.

      • checkAndUpdateStableSpec

        public static void checkAndUpdateStableSpec​(org.apache.flink.kubernetes.operator.api.status.CommonStatus<?> status)
        Checks the status and if the corresponding Flink job/application is in stable running state, it updates the last stable spec.
        参数:
        status - Status to be updated.

      • updateStatusForAlreadyUpgraded

        public static void updateStatusForAlreadyUpgraded​(org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<?,​?> resource)
        Updates status in cases where a previously successful deployment wasn't recorded for any reason. We simply change the job status from SUSPENDED to RUNNING and ReconciliationState to DEPLOYED while keeping the metadata.
        参数:
        resource - Flink resource to be updated.

      • updateReconciliationMetadata

        public static <SPEC extends org.apache.flink.kubernetes.operator.api.spec.AbstractFlinkSpec> void updateReconciliationMetadata​(org.apache.flink.kubernetes.operator.api.AbstractFlinkResource<SPEC,​?> resource)