[jira] [Commented] (FLINK-31860) FlinkDeployments never finalize when namespace is deleted
[
https://issues.apache.org/jira/browse/FLINK-31860?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17825414#comment-17825414
]
Gyula Fora commented on FLINK-31860:
I don’t really know how that would be possible but I welcome any
recommendation/ solution .
> FlinkDeployments never finalize when namespace is deleted
> -
>
> Key: FLINK-31860
> URL: https://issues.apache.org/jira/browse/FLINK-31860
> Project: Flink
> Issue Type: Bug
> Components: Kubernetes Operator
>Affects Versions: kubernetes-operator-1.3.1
> Environment: Apache Flink Kubernetes Operator 1.3.1
> Kubernetes 1.24.9
>Reporter: Jayme Howard
>Assignee: Jayme Howard
>Priority: Major
> Labels: pull-request-available, stale-assigned
>
> This appears to be a pretty straightforward issue, but I don't know the
> codebase well enough to propose a fix. When a FlinkDeployment is present in
> a namespace, and the namespace is deleted, the FlinkDeployment never
> reconciles and fails to complete its finalizer. This leads to the namespace
> being blocked from deletion indefinitely, requiring manual manipulation to
> remove the finalizer on the FlinkDeployment.
>
> Namespace conditions:
> {code:java}
> conditions:
> - lastTransitionTime: '2023-04-18T22:17:48Z'
> message: All resources successfully discovered
> reason: ResourcesDiscovered
> status: 'False'
> type: NamespaceDeletionDiscoveryFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: All legacy kube types successfully parsed
> reason: ParsedGroupVersions
> status: 'False'
> type: NamespaceDeletionGroupVersionParsingFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: All content successfully deleted, may be waiting on finalization
> reason: ContentDeleted
> status: 'False'
> type: NamespaceDeletionContentFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: 'Some resources are remaining: flinkdeployments.flink.apache.org
> has 2
> resource instances'
> reason: SomeResourcesRemain
> status: 'True'
> type: NamespaceContentRemaining
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: 'Some content in the namespace has finalizers remaining:
> flinkdeployments.flink.apache.org/finalizer
> in 2 resource instances'
> reason: SomeFinalizersRemain
> status: 'True'
> type: NamespaceFinalizersRemaining
> phase: Terminating {code}
> FlinkDeployment example (some fields redacted):
> {code:java}
> apiVersion: flink.apache.org/v1beta1
> kind: FlinkDeployment
> metadata:
> creationTimestamp: '2023-03-23T18:27:02Z'
> deletionGracePeriodSeconds: 0
> deletionTimestamp: '2023-03-23T18:27:35Z'
> finalizers:
> - flinkdeployments.flink.apache.org/finalizer
> generation: 3
> name:
> namespace:
> resourceVersion: '10565277081'
> uid: e50d2683-6c0c-467e-b10c-fe0f4e404692
> spec:
> flinkConfiguration:
> taskmanager.numberOfTaskSlots: '2'
> flinkVersion: v1_16
> image:
> job:
> args: []
> entryClass:
> jarURI:
> parallelism: 2
> state: running
> upgradeMode: stateless
> jobManager:
> replicas: 1
> resource:
> cpu: 1
> memory: 2048m
> logConfiguration:
> log4j-console.properties: '# This affects logging for both user code and
> Flink rootLogger.level = INFO rootLogger.appenderRef.console.ref =
> ConsoleAppender rootLogger.appenderRef.rolling.ref = RollingFileAppender
> # Uncomment this if you want to _only_ change Flink''s logging
> #logger.flink.name = org.apache.flink #logger.flink.level = INFO #
> The following lines keep the log level of common libraries/connectors on
> # log level INFO. The root logger does not override this. You have to
> manually # change the log levels here. logger.akka.name = akka
> logger.akka.level = INFO logger.kafka.name= org.apache.kafka
> logger.kafka.level = INFO logger.hadoop.name = org.apache.hadoop
> logger.hadoop.level = INFO logger.zookeeper.name = org.apache.zookeeper
> logger.zookeeper.level = INFO # Log all infos to the console
> appender.console.name = ConsoleAppender appender.console.type = CONSOLE
> appender.console.layout.type = PatternLayout
> appender.console.layout.pattern = %d{-MM-dd HH:mm:ss,SSS} %-5p %-60c %x
> - %m%n # Log all infos in the given rolling file
> appender.rolling.name = RollingFileAppender appender.rolling.type =
> RollingFile appender.rolling.append = false
> appender.rolling.fileName = ${sys:log.file} appender.rolling.filePattern
> = ${sys:log.file}.%i appender.rolling.layout.type = PatternLayout
> appender.rolling.layout.pattern = %d{-MM-dd
[jira] [Commented] (FLINK-31860) FlinkDeployments never finalize when namespace is deleted
[
https://issues.apache.org/jira/browse/FLINK-31860?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17825409#comment-17825409
]
Kevin Lam commented on FLINK-31860:
---
Thanks [~gyfora]! Is there any consideration for putting a solution in place in
the Flink Kubernetes Operator, if there will not be one from the kubernetes /
josdk side?
> FlinkDeployments never finalize when namespace is deleted
> -
>
> Key: FLINK-31860
> URL: https://issues.apache.org/jira/browse/FLINK-31860
> Project: Flink
> Issue Type: Bug
> Components: Kubernetes Operator
>Affects Versions: kubernetes-operator-1.3.1
> Environment: Apache Flink Kubernetes Operator 1.3.1
> Kubernetes 1.24.9
>Reporter: Jayme Howard
>Assignee: Jayme Howard
>Priority: Major
> Labels: pull-request-available, stale-assigned
>
> This appears to be a pretty straightforward issue, but I don't know the
> codebase well enough to propose a fix. When a FlinkDeployment is present in
> a namespace, and the namespace is deleted, the FlinkDeployment never
> reconciles and fails to complete its finalizer. This leads to the namespace
> being blocked from deletion indefinitely, requiring manual manipulation to
> remove the finalizer on the FlinkDeployment.
>
> Namespace conditions:
> {code:java}
> conditions:
> - lastTransitionTime: '2023-04-18T22:17:48Z'
> message: All resources successfully discovered
> reason: ResourcesDiscovered
> status: 'False'
> type: NamespaceDeletionDiscoveryFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: All legacy kube types successfully parsed
> reason: ParsedGroupVersions
> status: 'False'
> type: NamespaceDeletionGroupVersionParsingFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: All content successfully deleted, may be waiting on finalization
> reason: ContentDeleted
> status: 'False'
> type: NamespaceDeletionContentFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: 'Some resources are remaining: flinkdeployments.flink.apache.org
> has 2
> resource instances'
> reason: SomeResourcesRemain
> status: 'True'
> type: NamespaceContentRemaining
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: 'Some content in the namespace has finalizers remaining:
> flinkdeployments.flink.apache.org/finalizer
> in 2 resource instances'
> reason: SomeFinalizersRemain
> status: 'True'
> type: NamespaceFinalizersRemaining
> phase: Terminating {code}
> FlinkDeployment example (some fields redacted):
> {code:java}
> apiVersion: flink.apache.org/v1beta1
> kind: FlinkDeployment
> metadata:
> creationTimestamp: '2023-03-23T18:27:02Z'
> deletionGracePeriodSeconds: 0
> deletionTimestamp: '2023-03-23T18:27:35Z'
> finalizers:
> - flinkdeployments.flink.apache.org/finalizer
> generation: 3
> name:
> namespace:
> resourceVersion: '10565277081'
> uid: e50d2683-6c0c-467e-b10c-fe0f4e404692
> spec:
> flinkConfiguration:
> taskmanager.numberOfTaskSlots: '2'
> flinkVersion: v1_16
> image:
> job:
> args: []
> entryClass:
> jarURI:
> parallelism: 2
> state: running
> upgradeMode: stateless
> jobManager:
> replicas: 1
> resource:
> cpu: 1
> memory: 2048m
> logConfiguration:
> log4j-console.properties: '# This affects logging for both user code and
> Flink rootLogger.level = INFO rootLogger.appenderRef.console.ref =
> ConsoleAppender rootLogger.appenderRef.rolling.ref = RollingFileAppender
> # Uncomment this if you want to _only_ change Flink''s logging
> #logger.flink.name = org.apache.flink #logger.flink.level = INFO #
> The following lines keep the log level of common libraries/connectors on
> # log level INFO. The root logger does not override this. You have to
> manually # change the log levels here. logger.akka.name = akka
> logger.akka.level = INFO logger.kafka.name= org.apache.kafka
> logger.kafka.level = INFO logger.hadoop.name = org.apache.hadoop
> logger.hadoop.level = INFO logger.zookeeper.name = org.apache.zookeeper
> logger.zookeeper.level = INFO # Log all infos to the console
> appender.console.name = ConsoleAppender appender.console.type = CONSOLE
> appender.console.layout.type = PatternLayout
> appender.console.layout.pattern = %d{-MM-dd HH:mm:ss,SSS} %-5p %-60c %x
> - %m%n # Log all infos in the given rolling file
> appender.rolling.name = RollingFileAppender appender.rolling.type =
> RollingFile appender.rolling.append = false
> appender.rolling.fileName = ${sys:log.file} appender.rolling.filePattern
> = ${sys:log.file}.%i appender.rolling.layout.type
[jira] [Commented] (FLINK-31860) FlinkDeployments never finalize when namespace is deleted
[
https://issues.apache.org/jira/browse/FLINK-31860?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17825379#comment-17825379
]
Gyula Fora commented on FLINK-31860:
I am not aware of any solution from the kubernetes / josdk side [~klam-shop]
> FlinkDeployments never finalize when namespace is deleted
> -
>
> Key: FLINK-31860
> URL: https://issues.apache.org/jira/browse/FLINK-31860
> Project: Flink
> Issue Type: Bug
> Components: Kubernetes Operator
>Affects Versions: kubernetes-operator-1.3.1
> Environment: Apache Flink Kubernetes Operator 1.3.1
> Kubernetes 1.24.9
>Reporter: Jayme Howard
>Assignee: Jayme Howard
>Priority: Major
> Labels: pull-request-available, stale-assigned
>
> This appears to be a pretty straightforward issue, but I don't know the
> codebase well enough to propose a fix. When a FlinkDeployment is present in
> a namespace, and the namespace is deleted, the FlinkDeployment never
> reconciles and fails to complete its finalizer. This leads to the namespace
> being blocked from deletion indefinitely, requiring manual manipulation to
> remove the finalizer on the FlinkDeployment.
>
> Namespace conditions:
> {code:java}
> conditions:
> - lastTransitionTime: '2023-04-18T22:17:48Z'
> message: All resources successfully discovered
> reason: ResourcesDiscovered
> status: 'False'
> type: NamespaceDeletionDiscoveryFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: All legacy kube types successfully parsed
> reason: ParsedGroupVersions
> status: 'False'
> type: NamespaceDeletionGroupVersionParsingFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: All content successfully deleted, may be waiting on finalization
> reason: ContentDeleted
> status: 'False'
> type: NamespaceDeletionContentFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: 'Some resources are remaining: flinkdeployments.flink.apache.org
> has 2
> resource instances'
> reason: SomeResourcesRemain
> status: 'True'
> type: NamespaceContentRemaining
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: 'Some content in the namespace has finalizers remaining:
> flinkdeployments.flink.apache.org/finalizer
> in 2 resource instances'
> reason: SomeFinalizersRemain
> status: 'True'
> type: NamespaceFinalizersRemaining
> phase: Terminating {code}
> FlinkDeployment example (some fields redacted):
> {code:java}
> apiVersion: flink.apache.org/v1beta1
> kind: FlinkDeployment
> metadata:
> creationTimestamp: '2023-03-23T18:27:02Z'
> deletionGracePeriodSeconds: 0
> deletionTimestamp: '2023-03-23T18:27:35Z'
> finalizers:
> - flinkdeployments.flink.apache.org/finalizer
> generation: 3
> name:
> namespace:
> resourceVersion: '10565277081'
> uid: e50d2683-6c0c-467e-b10c-fe0f4e404692
> spec:
> flinkConfiguration:
> taskmanager.numberOfTaskSlots: '2'
> flinkVersion: v1_16
> image:
> job:
> args: []
> entryClass:
> jarURI:
> parallelism: 2
> state: running
> upgradeMode: stateless
> jobManager:
> replicas: 1
> resource:
> cpu: 1
> memory: 2048m
> logConfiguration:
> log4j-console.properties: '# This affects logging for both user code and
> Flink rootLogger.level = INFO rootLogger.appenderRef.console.ref =
> ConsoleAppender rootLogger.appenderRef.rolling.ref = RollingFileAppender
> # Uncomment this if you want to _only_ change Flink''s logging
> #logger.flink.name = org.apache.flink #logger.flink.level = INFO #
> The following lines keep the log level of common libraries/connectors on
> # log level INFO. The root logger does not override this. You have to
> manually # change the log levels here. logger.akka.name = akka
> logger.akka.level = INFO logger.kafka.name= org.apache.kafka
> logger.kafka.level = INFO logger.hadoop.name = org.apache.hadoop
> logger.hadoop.level = INFO logger.zookeeper.name = org.apache.zookeeper
> logger.zookeeper.level = INFO # Log all infos to the console
> appender.console.name = ConsoleAppender appender.console.type = CONSOLE
> appender.console.layout.type = PatternLayout
> appender.console.layout.pattern = %d{-MM-dd HH:mm:ss,SSS} %-5p %-60c %x
> - %m%n # Log all infos in the given rolling file
> appender.rolling.name = RollingFileAppender appender.rolling.type =
> RollingFile appender.rolling.append = false
> appender.rolling.fileName = ${sys:log.file} appender.rolling.filePattern
> = ${sys:log.file}.%i appender.rolling.layout.type = PatternLayout
> appender.rolling.layout.pattern = %d{-MM-dd HH:mm:ss,SSS} %-5p
[jira] [Commented] (FLINK-31860) FlinkDeployments never finalize when namespace is deleted
[
https://issues.apache.org/jira/browse/FLINK-31860?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17824083#comment-17824083
]
Kevin Lam commented on FLINK-31860:
---
Hi, any updates on this issue?
It seems that [https://github.com/kubernetes/kubernetes/issues/115070] was
closed without any changes made.
> FlinkDeployments never finalize when namespace is deleted
> -
>
> Key: FLINK-31860
> URL: https://issues.apache.org/jira/browse/FLINK-31860
> Project: Flink
> Issue Type: Bug
> Components: Kubernetes Operator
>Affects Versions: kubernetes-operator-1.3.1
> Environment: Apache Flink Kubernetes Operator 1.3.1
> Kubernetes 1.24.9
>Reporter: Jayme Howard
>Assignee: Jayme Howard
>Priority: Blocker
> Labels: pull-request-available, stale-assigned
>
> This appears to be a pretty straightforward issue, but I don't know the
> codebase well enough to propose a fix. When a FlinkDeployment is present in
> a namespace, and the namespace is deleted, the FlinkDeployment never
> reconciles and fails to complete its finalizer. This leads to the namespace
> being blocked from deletion indefinitely, requiring manual manipulation to
> remove the finalizer on the FlinkDeployment.
>
> Namespace conditions:
> {code:java}
> conditions:
> - lastTransitionTime: '2023-04-18T22:17:48Z'
> message: All resources successfully discovered
> reason: ResourcesDiscovered
> status: 'False'
> type: NamespaceDeletionDiscoveryFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: All legacy kube types successfully parsed
> reason: ParsedGroupVersions
> status: 'False'
> type: NamespaceDeletionGroupVersionParsingFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: All content successfully deleted, may be waiting on finalization
> reason: ContentDeleted
> status: 'False'
> type: NamespaceDeletionContentFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: 'Some resources are remaining: flinkdeployments.flink.apache.org
> has 2
> resource instances'
> reason: SomeResourcesRemain
> status: 'True'
> type: NamespaceContentRemaining
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: 'Some content in the namespace has finalizers remaining:
> flinkdeployments.flink.apache.org/finalizer
> in 2 resource instances'
> reason: SomeFinalizersRemain
> status: 'True'
> type: NamespaceFinalizersRemaining
> phase: Terminating {code}
> FlinkDeployment example (some fields redacted):
> {code:java}
> apiVersion: flink.apache.org/v1beta1
> kind: FlinkDeployment
> metadata:
> creationTimestamp: '2023-03-23T18:27:02Z'
> deletionGracePeriodSeconds: 0
> deletionTimestamp: '2023-03-23T18:27:35Z'
> finalizers:
> - flinkdeployments.flink.apache.org/finalizer
> generation: 3
> name:
> namespace:
> resourceVersion: '10565277081'
> uid: e50d2683-6c0c-467e-b10c-fe0f4e404692
> spec:
> flinkConfiguration:
> taskmanager.numberOfTaskSlots: '2'
> flinkVersion: v1_16
> image:
> job:
> args: []
> entryClass:
> jarURI:
> parallelism: 2
> state: running
> upgradeMode: stateless
> jobManager:
> replicas: 1
> resource:
> cpu: 1
> memory: 2048m
> logConfiguration:
> log4j-console.properties: '# This affects logging for both user code and
> Flink rootLogger.level = INFO rootLogger.appenderRef.console.ref =
> ConsoleAppender rootLogger.appenderRef.rolling.ref = RollingFileAppender
> # Uncomment this if you want to _only_ change Flink''s logging
> #logger.flink.name = org.apache.flink #logger.flink.level = INFO #
> The following lines keep the log level of common libraries/connectors on
> # log level INFO. The root logger does not override this. You have to
> manually # change the log levels here. logger.akka.name = akka
> logger.akka.level = INFO logger.kafka.name= org.apache.kafka
> logger.kafka.level = INFO logger.hadoop.name = org.apache.hadoop
> logger.hadoop.level = INFO logger.zookeeper.name = org.apache.zookeeper
> logger.zookeeper.level = INFO # Log all infos to the console
> appender.console.name = ConsoleAppender appender.console.type = CONSOLE
> appender.console.layout.type = PatternLayout
> appender.console.layout.pattern = %d{-MM-dd HH:mm:ss,SSS} %-5p %-60c %x
> - %m%n # Log all infos in the given rolling file
> appender.rolling.name = RollingFileAppender appender.rolling.type =
> RollingFile appender.rolling.append = false
> appender.rolling.fileName = ${sys:log.file} appender.rolling.filePattern
> = ${sys:log.file}.%i appender.rolling.layout.type = PatternLayout
>
[jira] [Commented] (FLINK-31860) FlinkDeployments never finalize when namespace is deleted
[
https://issues.apache.org/jira/browse/FLINK-31860?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17758437#comment-17758437
]
Clement Chevalier commented on FLINK-31860:
---
Hi Folks,
we're also facing this and all our dev process relies on namespace deletion so
it's quite annoying for us.
I didn't have the opportunity to test the fix proposed by [~rmetzger] , but it
sounded like a reasonable workaround, until the
[issue|https://github.com/operator-framework/java-operator-sdk/issues/1876] is
fixed on java-operator-sdk/kubernetes.
[~gyfora] did you test this simple fix with cluster-wide permissions ?
> FlinkDeployments never finalize when namespace is deleted
> -
>
> Key: FLINK-31860
> URL: https://issues.apache.org/jira/browse/FLINK-31860
> Project: Flink
> Issue Type: Bug
> Components: Kubernetes Operator
>Affects Versions: kubernetes-operator-1.3.1
> Environment: Apache Flink Kubernetes Operator 1.3.1
> Kubernetes 1.24.9
>Reporter: Jayme Howard
>Assignee: Jayme Howard
>Priority: Blocker
> Labels: pull-request-available
>
> This appears to be a pretty straightforward issue, but I don't know the
> codebase well enough to propose a fix. When a FlinkDeployment is present in
> a namespace, and the namespace is deleted, the FlinkDeployment never
> reconciles and fails to complete its finalizer. This leads to the namespace
> being blocked from deletion indefinitely, requiring manual manipulation to
> remove the finalizer on the FlinkDeployment.
>
> Namespace conditions:
> {code:java}
> conditions:
> - lastTransitionTime: '2023-04-18T22:17:48Z'
> message: All resources successfully discovered
> reason: ResourcesDiscovered
> status: 'False'
> type: NamespaceDeletionDiscoveryFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: All legacy kube types successfully parsed
> reason: ParsedGroupVersions
> status: 'False'
> type: NamespaceDeletionGroupVersionParsingFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: All content successfully deleted, may be waiting on finalization
> reason: ContentDeleted
> status: 'False'
> type: NamespaceDeletionContentFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: 'Some resources are remaining: flinkdeployments.flink.apache.org
> has 2
> resource instances'
> reason: SomeResourcesRemain
> status: 'True'
> type: NamespaceContentRemaining
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: 'Some content in the namespace has finalizers remaining:
> flinkdeployments.flink.apache.org/finalizer
> in 2 resource instances'
> reason: SomeFinalizersRemain
> status: 'True'
> type: NamespaceFinalizersRemaining
> phase: Terminating {code}
> FlinkDeployment example (some fields redacted):
> {code:java}
> apiVersion: flink.apache.org/v1beta1
> kind: FlinkDeployment
> metadata:
> creationTimestamp: '2023-03-23T18:27:02Z'
> deletionGracePeriodSeconds: 0
> deletionTimestamp: '2023-03-23T18:27:35Z'
> finalizers:
> - flinkdeployments.flink.apache.org/finalizer
> generation: 3
> name:
> namespace:
> resourceVersion: '10565277081'
> uid: e50d2683-6c0c-467e-b10c-fe0f4e404692
> spec:
> flinkConfiguration:
> taskmanager.numberOfTaskSlots: '2'
> flinkVersion: v1_16
> image:
> job:
> args: []
> entryClass:
> jarURI:
> parallelism: 2
> state: running
> upgradeMode: stateless
> jobManager:
> replicas: 1
> resource:
> cpu: 1
> memory: 2048m
> logConfiguration:
> log4j-console.properties: '# This affects logging for both user code and
> Flink rootLogger.level = INFO rootLogger.appenderRef.console.ref =
> ConsoleAppender rootLogger.appenderRef.rolling.ref = RollingFileAppender
> # Uncomment this if you want to _only_ change Flink''s logging
> #logger.flink.name = org.apache.flink #logger.flink.level = INFO #
> The following lines keep the log level of common libraries/connectors on
> # log level INFO. The root logger does not override this. You have to
> manually # change the log levels here. logger.akka.name = akka
> logger.akka.level = INFO logger.kafka.name= org.apache.kafka
> logger.kafka.level = INFO logger.hadoop.name = org.apache.hadoop
> logger.hadoop.level = INFO logger.zookeeper.name = org.apache.zookeeper
> logger.zookeeper.level = INFO # Log all infos to the console
> appender.console.name = ConsoleAppender appender.console.type = CONSOLE
> appender.console.layout.type = PatternLayout
> appender.console.layout.pattern = %d{-MM-dd HH:mm:ss,SSS} %-5p %-60c %x
> - %m%n # Log all infos in the given rolling file
[jira] [Commented] (FLINK-31860) FlinkDeployments never finalize when namespace is deleted
[
https://issues.apache.org/jira/browse/FLINK-31860?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17720981#comment-17720981
]
Jayme Howard commented on FLINK-31860:
--
I *think* that only didn't work because of per-namespace permissions. If
cluster-wide permissions are in place, I would expect this to still work?
> FlinkDeployments never finalize when namespace is deleted
> -
>
> Key: FLINK-31860
> URL: https://issues.apache.org/jira/browse/FLINK-31860
> Project: Flink
> Issue Type: Bug
> Components: Kubernetes Operator
>Affects Versions: kubernetes-operator-1.3.1
> Environment: Apache Flink Kubernetes Operator 1.3.1
> Kubernetes 1.24.9
>Reporter: Jayme Howard
>Assignee: Jayme Howard
>Priority: Blocker
> Labels: pull-request-available
>
> This appears to be a pretty straightforward issue, but I don't know the
> codebase well enough to propose a fix. When a FlinkDeployment is present in
> a namespace, and the namespace is deleted, the FlinkDeployment never
> reconciles and fails to complete its finalizer. This leads to the namespace
> being blocked from deletion indefinitely, requiring manual manipulation to
> remove the finalizer on the FlinkDeployment.
>
> Namespace conditions:
> {code:java}
> conditions:
> - lastTransitionTime: '2023-04-18T22:17:48Z'
> message: All resources successfully discovered
> reason: ResourcesDiscovered
> status: 'False'
> type: NamespaceDeletionDiscoveryFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: All legacy kube types successfully parsed
> reason: ParsedGroupVersions
> status: 'False'
> type: NamespaceDeletionGroupVersionParsingFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: All content successfully deleted, may be waiting on finalization
> reason: ContentDeleted
> status: 'False'
> type: NamespaceDeletionContentFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: 'Some resources are remaining: flinkdeployments.flink.apache.org
> has 2
> resource instances'
> reason: SomeResourcesRemain
> status: 'True'
> type: NamespaceContentRemaining
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: 'Some content in the namespace has finalizers remaining:
> flinkdeployments.flink.apache.org/finalizer
> in 2 resource instances'
> reason: SomeFinalizersRemain
> status: 'True'
> type: NamespaceFinalizersRemaining
> phase: Terminating {code}
> FlinkDeployment example (some fields redacted):
> {code:java}
> apiVersion: flink.apache.org/v1beta1
> kind: FlinkDeployment
> metadata:
> creationTimestamp: '2023-03-23T18:27:02Z'
> deletionGracePeriodSeconds: 0
> deletionTimestamp: '2023-03-23T18:27:35Z'
> finalizers:
> - flinkdeployments.flink.apache.org/finalizer
> generation: 3
> name:
> namespace:
> resourceVersion: '10565277081'
> uid: e50d2683-6c0c-467e-b10c-fe0f4e404692
> spec:
> flinkConfiguration:
> taskmanager.numberOfTaskSlots: '2'
> flinkVersion: v1_16
> image:
> job:
> args: []
> entryClass:
> jarURI:
> parallelism: 2
> state: running
> upgradeMode: stateless
> jobManager:
> replicas: 1
> resource:
> cpu: 1
> memory: 2048m
> logConfiguration:
> log4j-console.properties: '# This affects logging for both user code and
> Flink rootLogger.level = INFO rootLogger.appenderRef.console.ref =
> ConsoleAppender rootLogger.appenderRef.rolling.ref = RollingFileAppender
> # Uncomment this if you want to _only_ change Flink''s logging
> #logger.flink.name = org.apache.flink #logger.flink.level = INFO #
> The following lines keep the log level of common libraries/connectors on
> # log level INFO. The root logger does not override this. You have to
> manually # change the log levels here. logger.akka.name = akka
> logger.akka.level = INFO logger.kafka.name= org.apache.kafka
> logger.kafka.level = INFO logger.hadoop.name = org.apache.hadoop
> logger.hadoop.level = INFO logger.zookeeper.name = org.apache.zookeeper
> logger.zookeeper.level = INFO # Log all infos to the console
> appender.console.name = ConsoleAppender appender.console.type = CONSOLE
> appender.console.layout.type = PatternLayout
> appender.console.layout.pattern = %d{-MM-dd HH:mm:ss,SSS} %-5p %-60c %x
> - %m%n # Log all infos in the given rolling file
> appender.rolling.name = RollingFileAppender appender.rolling.type =
> RollingFile appender.rolling.append = false
> appender.rolling.fileName = ${sys:log.file} appender.rolling.filePattern
> = ${sys:log.file}.%i appender.rolling.layout.type = PatternLayout
>
[jira] [Commented] (FLINK-31860) FlinkDeployments never finalize when namespace is deleted
[
https://issues.apache.org/jira/browse/FLINK-31860?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17720973#comment-17720973
]
Gyula Fora commented on FLINK-31860:
[~isugimpy] we can definitely add this improvement but it didn't fix it for me
completely as I was getting an error from the Java Operator SDK itself as it
cannot finish deletion of the CR itself (cannot remove the finalizer)
> FlinkDeployments never finalize when namespace is deleted
> -
>
> Key: FLINK-31860
> URL: https://issues.apache.org/jira/browse/FLINK-31860
> Project: Flink
> Issue Type: Bug
> Components: Kubernetes Operator
>Affects Versions: kubernetes-operator-1.3.1
> Environment: Apache Flink Kubernetes Operator 1.3.1
> Kubernetes 1.24.9
>Reporter: Jayme Howard
>Assignee: Gyula Fora
>Priority: Blocker
> Labels: pull-request-available
>
> This appears to be a pretty straightforward issue, but I don't know the
> codebase well enough to propose a fix. When a FlinkDeployment is present in
> a namespace, and the namespace is deleted, the FlinkDeployment never
> reconciles and fails to complete its finalizer. This leads to the namespace
> being blocked from deletion indefinitely, requiring manual manipulation to
> remove the finalizer on the FlinkDeployment.
>
> Namespace conditions:
> {code:java}
> conditions:
> - lastTransitionTime: '2023-04-18T22:17:48Z'
> message: All resources successfully discovered
> reason: ResourcesDiscovered
> status: 'False'
> type: NamespaceDeletionDiscoveryFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: All legacy kube types successfully parsed
> reason: ParsedGroupVersions
> status: 'False'
> type: NamespaceDeletionGroupVersionParsingFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: All content successfully deleted, may be waiting on finalization
> reason: ContentDeleted
> status: 'False'
> type: NamespaceDeletionContentFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: 'Some resources are remaining: flinkdeployments.flink.apache.org
> has 2
> resource instances'
> reason: SomeResourcesRemain
> status: 'True'
> type: NamespaceContentRemaining
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: 'Some content in the namespace has finalizers remaining:
> flinkdeployments.flink.apache.org/finalizer
> in 2 resource instances'
> reason: SomeFinalizersRemain
> status: 'True'
> type: NamespaceFinalizersRemaining
> phase: Terminating {code}
> FlinkDeployment example (some fields redacted):
> {code:java}
> apiVersion: flink.apache.org/v1beta1
> kind: FlinkDeployment
> metadata:
> creationTimestamp: '2023-03-23T18:27:02Z'
> deletionGracePeriodSeconds: 0
> deletionTimestamp: '2023-03-23T18:27:35Z'
> finalizers:
> - flinkdeployments.flink.apache.org/finalizer
> generation: 3
> name:
> namespace:
> resourceVersion: '10565277081'
> uid: e50d2683-6c0c-467e-b10c-fe0f4e404692
> spec:
> flinkConfiguration:
> taskmanager.numberOfTaskSlots: '2'
> flinkVersion: v1_16
> image:
> job:
> args: []
> entryClass:
> jarURI:
> parallelism: 2
> state: running
> upgradeMode: stateless
> jobManager:
> replicas: 1
> resource:
> cpu: 1
> memory: 2048m
> logConfiguration:
> log4j-console.properties: '# This affects logging for both user code and
> Flink rootLogger.level = INFO rootLogger.appenderRef.console.ref =
> ConsoleAppender rootLogger.appenderRef.rolling.ref = RollingFileAppender
> # Uncomment this if you want to _only_ change Flink''s logging
> #logger.flink.name = org.apache.flink #logger.flink.level = INFO #
> The following lines keep the log level of common libraries/connectors on
> # log level INFO. The root logger does not override this. You have to
> manually # change the log levels here. logger.akka.name = akka
> logger.akka.level = INFO logger.kafka.name= org.apache.kafka
> logger.kafka.level = INFO logger.hadoop.name = org.apache.hadoop
> logger.hadoop.level = INFO logger.zookeeper.name = org.apache.zookeeper
> logger.zookeeper.level = INFO # Log all infos to the console
> appender.console.name = ConsoleAppender appender.console.type = CONSOLE
> appender.console.layout.type = PatternLayout
> appender.console.layout.pattern = %d{-MM-dd HH:mm:ss,SSS} %-5p %-60c %x
> - %m%n # Log all infos in the given rolling file
> appender.rolling.name = RollingFileAppender appender.rolling.type =
> RollingFile appender.rolling.append = false
> appender.rolling.fileName = ${sys:log.file} appender.rolling.filePattern
> =
[jira] [Commented] (FLINK-31860) FlinkDeployments never finalize when namespace is deleted
[
https://issues.apache.org/jira/browse/FLINK-31860?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17720654#comment-17720654
]
Jayme Howard commented on FLINK-31860:
--
[~rmetzger] That looks awesome! Thank you for sharing it!
[~gyfora] The above looks like it'd accomplish a fix for the immediate problem
very easily and prevent this from impacting production clusters. If I opened a
PR with that, would that be something considered acceptable as a workaround
until a more proper fix could be implemented? I'd prefer to not have to
maintain an internal fork of the operator when the community as a whole could
benefit from this.
> FlinkDeployments never finalize when namespace is deleted
> -
>
> Key: FLINK-31860
> URL: https://issues.apache.org/jira/browse/FLINK-31860
> Project: Flink
> Issue Type: Bug
> Components: Kubernetes Operator
>Affects Versions: kubernetes-operator-1.3.1
> Environment: Apache Flink Kubernetes Operator 1.3.1
> Kubernetes 1.24.9
>Reporter: Jayme Howard
>Assignee: Gyula Fora
>Priority: Blocker
> Labels: pull-request-available
>
> This appears to be a pretty straightforward issue, but I don't know the
> codebase well enough to propose a fix. When a FlinkDeployment is present in
> a namespace, and the namespace is deleted, the FlinkDeployment never
> reconciles and fails to complete its finalizer. This leads to the namespace
> being blocked from deletion indefinitely, requiring manual manipulation to
> remove the finalizer on the FlinkDeployment.
>
> Namespace conditions:
> {code:java}
> conditions:
> - lastTransitionTime: '2023-04-18T22:17:48Z'
> message: All resources successfully discovered
> reason: ResourcesDiscovered
> status: 'False'
> type: NamespaceDeletionDiscoveryFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: All legacy kube types successfully parsed
> reason: ParsedGroupVersions
> status: 'False'
> type: NamespaceDeletionGroupVersionParsingFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: All content successfully deleted, may be waiting on finalization
> reason: ContentDeleted
> status: 'False'
> type: NamespaceDeletionContentFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: 'Some resources are remaining: flinkdeployments.flink.apache.org
> has 2
> resource instances'
> reason: SomeResourcesRemain
> status: 'True'
> type: NamespaceContentRemaining
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: 'Some content in the namespace has finalizers remaining:
> flinkdeployments.flink.apache.org/finalizer
> in 2 resource instances'
> reason: SomeFinalizersRemain
> status: 'True'
> type: NamespaceFinalizersRemaining
> phase: Terminating {code}
> FlinkDeployment example (some fields redacted):
> {code:java}
> apiVersion: flink.apache.org/v1beta1
> kind: FlinkDeployment
> metadata:
> creationTimestamp: '2023-03-23T18:27:02Z'
> deletionGracePeriodSeconds: 0
> deletionTimestamp: '2023-03-23T18:27:35Z'
> finalizers:
> - flinkdeployments.flink.apache.org/finalizer
> generation: 3
> name:
> namespace:
> resourceVersion: '10565277081'
> uid: e50d2683-6c0c-467e-b10c-fe0f4e404692
> spec:
> flinkConfiguration:
> taskmanager.numberOfTaskSlots: '2'
> flinkVersion: v1_16
> image:
> job:
> args: []
> entryClass:
> jarURI:
> parallelism: 2
> state: running
> upgradeMode: stateless
> jobManager:
> replicas: 1
> resource:
> cpu: 1
> memory: 2048m
> logConfiguration:
> log4j-console.properties: '# This affects logging for both user code and
> Flink rootLogger.level = INFO rootLogger.appenderRef.console.ref =
> ConsoleAppender rootLogger.appenderRef.rolling.ref = RollingFileAppender
> # Uncomment this if you want to _only_ change Flink''s logging
> #logger.flink.name = org.apache.flink #logger.flink.level = INFO #
> The following lines keep the log level of common libraries/connectors on
> # log level INFO. The root logger does not override this. You have to
> manually # change the log levels here. logger.akka.name = akka
> logger.akka.level = INFO logger.kafka.name= org.apache.kafka
> logger.kafka.level = INFO logger.hadoop.name = org.apache.hadoop
> logger.hadoop.level = INFO logger.zookeeper.name = org.apache.zookeeper
> logger.zookeeper.level = INFO # Log all infos to the console
> appender.console.name = ConsoleAppender appender.console.type = CONSOLE
> appender.console.layout.type = PatternLayout
> appender.console.layout.pattern = %d{-MM-dd HH:mm:ss,SSS} %-5p %-60c %x
> - %m%n # Log all infos in the given
[jira] [Commented] (FLINK-31860) FlinkDeployments never finalize when namespace is deleted
[
https://issues.apache.org/jira/browse/FLINK-31860?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17719694#comment-17719694
]
Robert Metzger commented on FLINK-31860:
We were also facing this problem, and we've solved it for now in this hacky way:
{code}
---
.../kubernetes/operator/utils/EventUtils.java | 24 ++-
.../templates/rbac.yaml | 1 +
2 files changed, 19 insertions(+), 6 deletions(-)
diff --git
a/flink-kubernetes-operator/src/main/java/org/apache/flink/kubernetes/operator/utils/EventUtils.java
b/flink-kubernetes-operator/src/main/java/org/apache/flink/kubernetes/operator/utils/EventUtils.java
index d993de2..36c49a4 100644
---
a/flink-kubernetes-operator/src/main/java/org/apache/flink/kubernetes/operator/utils/EventUtils.java
+++
b/flink-kubernetes-operator/src/main/java/org/apache/flink/kubernetes/operator/utils/EventUtils.java
@@ -22,6 +22,8 @@
import io.fabric8.kubernetes.api.model.HasMetadata;
import io.fabric8.kubernetes.api.model.ObjectReferenceBuilder;
import io.fabric8.kubernetes.client.KubernetesClient;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
import java.time.Instant;
import java.util.function.Consumer;
@@ -31,6 +33,7 @@
*
https://github.com/EnMasseProject/enmasse/blob/master/k8s-api/src/main/java/io/enmasse/k8s/api/KubeEventLogger.java
*/
public class EventUtils {
+private static final Logger LOG =
LoggerFactory.getLogger(EventUtils.class);
public static String generateEventName(
HasMetadata target,
@@ -58,14 +61,14 @@ public static boolean createOrUpdateEvent(
String message,
EventRecorder.Component component,
Consumer eventListener) {
+var namespace = target.getMetadata().getNamespace();
+if (isNamespaceMarkedForDeletion(client, namespace)) {
+LOG.info("Ignoring event because namespace is marked for
deletion");
+return true;
+}
var eventName = generateEventName(target, type, reason, message,
component);
-var existing =
-client.v1()
-.events()
-.inNamespace(target.getMetadata().getNamespace())
-.withName(eventName)
-.get();
+var existing =
client.v1().events().inNamespace(namespace).withName(eventName).get();
if (existing != null
&& existing.getType().equals(type.name())
@@ -109,4 +112,13 @@ public static boolean createOrUpdateEvent(
return true;
}
}
+
+private static boolean isNamespaceMarkedForDeletion(KubernetesClient
client, String namespace) {
+try {
+return
client.namespaces().withName(namespace).get().isMarkedForDeletion();
+} catch (Exception e) {
+LOG.warn("Error while checking namespace status", e);
+return false;
+}
+}
}
diff --git a/helm/flink-kubernetes-operator/templates/rbac.yaml
b/helm/flink-kubernetes-operator/templates/rbac.yaml
index f50852e..21d7071 100644
--- a/helm/flink-kubernetes-operator/templates/rbac.yaml
+++ b/helm/flink-kubernetes-operator/templates/rbac.yaml
@@ -29,6 +29,7 @@ rules:
- events
- configmaps
- secrets
+ - namespaces
verbs:
- "*"
{{- if .Values.rbac.nodesRule.create }}
{code}
> FlinkDeployments never finalize when namespace is deleted
> -
>
> Key: FLINK-31860
> URL: https://issues.apache.org/jira/browse/FLINK-31860
> Project: Flink
> Issue Type: Bug
> Components: Kubernetes Operator
>Affects Versions: kubernetes-operator-1.3.1
> Environment: Apache Flink Kubernetes Operator 1.3.1
> Kubernetes 1.24.9
>Reporter: Jayme Howard
>Assignee: Gyula Fora
>Priority: Blocker
> Labels: pull-request-available
>
> This appears to be a pretty straightforward issue, but I don't know the
> codebase well enough to propose a fix. When a FlinkDeployment is present in
> a namespace, and the namespace is deleted, the FlinkDeployment never
> reconciles and fails to complete its finalizer. This leads to the namespace
> being blocked from deletion indefinitely, requiring manual manipulation to
> remove the finalizer on the FlinkDeployment.
>
> Namespace conditions:
> {code:java}
> conditions:
> - lastTransitionTime: '2023-04-18T22:17:48Z'
> message: All resources successfully discovered
> reason: ResourcesDiscovered
> status: 'False'
> type: NamespaceDeletionDiscoveryFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: All legacy kube types successfully parsed
> reason: ParsedGroupVersions
> status: 'False'
> type: NamespaceDeletionGroupVersionParsingFailure
> - lastTransitionTime:
[jira] [Commented] (FLINK-31860) FlinkDeployments never finalize when namespace is deleted
[
https://issues.apache.org/jira/browse/FLINK-31860?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17717055#comment-17717055
]
Gyula Fora commented on FLINK-31860:
I have to put this work on hold for now because any effort on the flink
operator side cannot fix a bigger underlying problem. We are investigating if
there is a workaround in kubernetes itself somehow.
More details here:
[java-operator-sdk/java-operator-sdk#1876|https://github.com/java-operator-sdk/java-operator-sdk/issues/1876]
> FlinkDeployments never finalize when namespace is deleted
> -
>
> Key: FLINK-31860
> URL: https://issues.apache.org/jira/browse/FLINK-31860
> Project: Flink
> Issue Type: Bug
> Components: Kubernetes Operator
>Affects Versions: kubernetes-operator-1.3.1
> Environment: Apache Flink Kubernetes Operator 1.3.1
> Kubernetes 1.24.9
>Reporter: Jayme Howard
>Assignee: Gyula Fora
>Priority: Blocker
> Labels: pull-request-available
>
> This appears to be a pretty straightforward issue, but I don't know the
> codebase well enough to propose a fix. When a FlinkDeployment is present in
> a namespace, and the namespace is deleted, the FlinkDeployment never
> reconciles and fails to complete its finalizer. This leads to the namespace
> being blocked from deletion indefinitely, requiring manual manipulation to
> remove the finalizer on the FlinkDeployment.
>
> Namespace conditions:
> {code:java}
> conditions:
> - lastTransitionTime: '2023-04-18T22:17:48Z'
> message: All resources successfully discovered
> reason: ResourcesDiscovered
> status: 'False'
> type: NamespaceDeletionDiscoveryFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: All legacy kube types successfully parsed
> reason: ParsedGroupVersions
> status: 'False'
> type: NamespaceDeletionGroupVersionParsingFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: All content successfully deleted, may be waiting on finalization
> reason: ContentDeleted
> status: 'False'
> type: NamespaceDeletionContentFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: 'Some resources are remaining: flinkdeployments.flink.apache.org
> has 2
> resource instances'
> reason: SomeResourcesRemain
> status: 'True'
> type: NamespaceContentRemaining
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: 'Some content in the namespace has finalizers remaining:
> flinkdeployments.flink.apache.org/finalizer
> in 2 resource instances'
> reason: SomeFinalizersRemain
> status: 'True'
> type: NamespaceFinalizersRemaining
> phase: Terminating {code}
> FlinkDeployment example (some fields redacted):
> {code:java}
> apiVersion: flink.apache.org/v1beta1
> kind: FlinkDeployment
> metadata:
> creationTimestamp: '2023-03-23T18:27:02Z'
> deletionGracePeriodSeconds: 0
> deletionTimestamp: '2023-03-23T18:27:35Z'
> finalizers:
> - flinkdeployments.flink.apache.org/finalizer
> generation: 3
> name:
> namespace:
> resourceVersion: '10565277081'
> uid: e50d2683-6c0c-467e-b10c-fe0f4e404692
> spec:
> flinkConfiguration:
> taskmanager.numberOfTaskSlots: '2'
> flinkVersion: v1_16
> image:
> job:
> args: []
> entryClass:
> jarURI:
> parallelism: 2
> state: running
> upgradeMode: stateless
> jobManager:
> replicas: 1
> resource:
> cpu: 1
> memory: 2048m
> logConfiguration:
> log4j-console.properties: '# This affects logging for both user code and
> Flink rootLogger.level = INFO rootLogger.appenderRef.console.ref =
> ConsoleAppender rootLogger.appenderRef.rolling.ref = RollingFileAppender
> # Uncomment this if you want to _only_ change Flink''s logging
> #logger.flink.name = org.apache.flink #logger.flink.level = INFO #
> The following lines keep the log level of common libraries/connectors on
> # log level INFO. The root logger does not override this. You have to
> manually # change the log levels here. logger.akka.name = akka
> logger.akka.level = INFO logger.kafka.name= org.apache.kafka
> logger.kafka.level = INFO logger.hadoop.name = org.apache.hadoop
> logger.hadoop.level = INFO logger.zookeeper.name = org.apache.zookeeper
> logger.zookeeper.level = INFO # Log all infos to the console
> appender.console.name = ConsoleAppender appender.console.type = CONSOLE
> appender.console.layout.type = PatternLayout
> appender.console.layout.pattern = %d{-MM-dd HH:mm:ss,SSS} %-5p %-60c %x
> - %m%n # Log all infos in the given rolling file
> appender.rolling.name = RollingFileAppender appender.rolling.type =
> RollingFile appender.rolling.append =
[jira] [Commented] (FLINK-31860) FlinkDeployments never finalize when namespace is deleted
[
https://issues.apache.org/jira/browse/FLINK-31860?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17716832#comment-17716832
]
Jayme Howard commented on FLINK-31860:
--
This looks very reasonable to me as written, greatly appreciate being tagged in
on it! I'll see if I can get a test of it done, but may not be able to
reasonably do so until next week when I'm back to work.
> FlinkDeployments never finalize when namespace is deleted
> -
>
> Key: FLINK-31860
> URL: https://issues.apache.org/jira/browse/FLINK-31860
> Project: Flink
> Issue Type: Bug
> Components: Kubernetes Operator
>Affects Versions: kubernetes-operator-1.3.1
> Environment: Apache Flink Kubernetes Operator 1.3.1
> Kubernetes 1.24.9
>Reporter: Jayme Howard
>Assignee: Gyula Fora
>Priority: Blocker
> Labels: pull-request-available
>
> This appears to be a pretty straightforward issue, but I don't know the
> codebase well enough to propose a fix. When a FlinkDeployment is present in
> a namespace, and the namespace is deleted, the FlinkDeployment never
> reconciles and fails to complete its finalizer. This leads to the namespace
> being blocked from deletion indefinitely, requiring manual manipulation to
> remove the finalizer on the FlinkDeployment.
>
> Namespace conditions:
> {code:java}
> conditions:
> - lastTransitionTime: '2023-04-18T22:17:48Z'
> message: All resources successfully discovered
> reason: ResourcesDiscovered
> status: 'False'
> type: NamespaceDeletionDiscoveryFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: All legacy kube types successfully parsed
> reason: ParsedGroupVersions
> status: 'False'
> type: NamespaceDeletionGroupVersionParsingFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: All content successfully deleted, may be waiting on finalization
> reason: ContentDeleted
> status: 'False'
> type: NamespaceDeletionContentFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: 'Some resources are remaining: flinkdeployments.flink.apache.org
> has 2
> resource instances'
> reason: SomeResourcesRemain
> status: 'True'
> type: NamespaceContentRemaining
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: 'Some content in the namespace has finalizers remaining:
> flinkdeployments.flink.apache.org/finalizer
> in 2 resource instances'
> reason: SomeFinalizersRemain
> status: 'True'
> type: NamespaceFinalizersRemaining
> phase: Terminating {code}
> FlinkDeployment example (some fields redacted):
> {code:java}
> apiVersion: flink.apache.org/v1beta1
> kind: FlinkDeployment
> metadata:
> creationTimestamp: '2023-03-23T18:27:02Z'
> deletionGracePeriodSeconds: 0
> deletionTimestamp: '2023-03-23T18:27:35Z'
> finalizers:
> - flinkdeployments.flink.apache.org/finalizer
> generation: 3
> name:
> namespace:
> resourceVersion: '10565277081'
> uid: e50d2683-6c0c-467e-b10c-fe0f4e404692
> spec:
> flinkConfiguration:
> taskmanager.numberOfTaskSlots: '2'
> flinkVersion: v1_16
> image:
> job:
> args: []
> entryClass:
> jarURI:
> parallelism: 2
> state: running
> upgradeMode: stateless
> jobManager:
> replicas: 1
> resource:
> cpu: 1
> memory: 2048m
> logConfiguration:
> log4j-console.properties: '# This affects logging for both user code and
> Flink rootLogger.level = INFO rootLogger.appenderRef.console.ref =
> ConsoleAppender rootLogger.appenderRef.rolling.ref = RollingFileAppender
> # Uncomment this if you want to _only_ change Flink''s logging
> #logger.flink.name = org.apache.flink #logger.flink.level = INFO #
> The following lines keep the log level of common libraries/connectors on
> # log level INFO. The root logger does not override this. You have to
> manually # change the log levels here. logger.akka.name = akka
> logger.akka.level = INFO logger.kafka.name= org.apache.kafka
> logger.kafka.level = INFO logger.hadoop.name = org.apache.hadoop
> logger.hadoop.level = INFO logger.zookeeper.name = org.apache.zookeeper
> logger.zookeeper.level = INFO # Log all infos to the console
> appender.console.name = ConsoleAppender appender.console.type = CONSOLE
> appender.console.layout.type = PatternLayout
> appender.console.layout.pattern = %d{-MM-dd HH:mm:ss,SSS} %-5p %-60c %x
> - %m%n # Log all infos in the given rolling file
> appender.rolling.name = RollingFileAppender appender.rolling.type =
> RollingFile appender.rolling.append = false
> appender.rolling.fileName = ${sys:log.file} appender.rolling.filePattern
> = ${sys:log.file}.%i
[jira] [Commented] (FLINK-31860) FlinkDeployments never finalize when namespace is deleted
[
https://issues.apache.org/jira/browse/FLINK-31860?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17716805#comment-17716805
]
Gyula Fora commented on FLINK-31860:
[~isugimpy] , the PR is ready, please help review/test this.
> FlinkDeployments never finalize when namespace is deleted
> -
>
> Key: FLINK-31860
> URL: https://issues.apache.org/jira/browse/FLINK-31860
> Project: Flink
> Issue Type: Bug
> Components: Kubernetes Operator
>Affects Versions: kubernetes-operator-1.3.1
> Environment: Apache Flink Kubernetes Operator 1.3.1
> Kubernetes 1.24.9
>Reporter: Jayme Howard
>Assignee: Gyula Fora
>Priority: Blocker
> Labels: pull-request-available
>
> This appears to be a pretty straightforward issue, but I don't know the
> codebase well enough to propose a fix. When a FlinkDeployment is present in
> a namespace, and the namespace is deleted, the FlinkDeployment never
> reconciles and fails to complete its finalizer. This leads to the namespace
> being blocked from deletion indefinitely, requiring manual manipulation to
> remove the finalizer on the FlinkDeployment.
>
> Namespace conditions:
> {code:java}
> conditions:
> - lastTransitionTime: '2023-04-18T22:17:48Z'
> message: All resources successfully discovered
> reason: ResourcesDiscovered
> status: 'False'
> type: NamespaceDeletionDiscoveryFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: All legacy kube types successfully parsed
> reason: ParsedGroupVersions
> status: 'False'
> type: NamespaceDeletionGroupVersionParsingFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: All content successfully deleted, may be waiting on finalization
> reason: ContentDeleted
> status: 'False'
> type: NamespaceDeletionContentFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: 'Some resources are remaining: flinkdeployments.flink.apache.org
> has 2
> resource instances'
> reason: SomeResourcesRemain
> status: 'True'
> type: NamespaceContentRemaining
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: 'Some content in the namespace has finalizers remaining:
> flinkdeployments.flink.apache.org/finalizer
> in 2 resource instances'
> reason: SomeFinalizersRemain
> status: 'True'
> type: NamespaceFinalizersRemaining
> phase: Terminating {code}
> FlinkDeployment example (some fields redacted):
> {code:java}
> apiVersion: flink.apache.org/v1beta1
> kind: FlinkDeployment
> metadata:
> creationTimestamp: '2023-03-23T18:27:02Z'
> deletionGracePeriodSeconds: 0
> deletionTimestamp: '2023-03-23T18:27:35Z'
> finalizers:
> - flinkdeployments.flink.apache.org/finalizer
> generation: 3
> name:
> namespace:
> resourceVersion: '10565277081'
> uid: e50d2683-6c0c-467e-b10c-fe0f4e404692
> spec:
> flinkConfiguration:
> taskmanager.numberOfTaskSlots: '2'
> flinkVersion: v1_16
> image:
> job:
> args: []
> entryClass:
> jarURI:
> parallelism: 2
> state: running
> upgradeMode: stateless
> jobManager:
> replicas: 1
> resource:
> cpu: 1
> memory: 2048m
> logConfiguration:
> log4j-console.properties: '# This affects logging for both user code and
> Flink rootLogger.level = INFO rootLogger.appenderRef.console.ref =
> ConsoleAppender rootLogger.appenderRef.rolling.ref = RollingFileAppender
> # Uncomment this if you want to _only_ change Flink''s logging
> #logger.flink.name = org.apache.flink #logger.flink.level = INFO #
> The following lines keep the log level of common libraries/connectors on
> # log level INFO. The root logger does not override this. You have to
> manually # change the log levels here. logger.akka.name = akka
> logger.akka.level = INFO logger.kafka.name= org.apache.kafka
> logger.kafka.level = INFO logger.hadoop.name = org.apache.hadoop
> logger.hadoop.level = INFO logger.zookeeper.name = org.apache.zookeeper
> logger.zookeeper.level = INFO # Log all infos to the console
> appender.console.name = ConsoleAppender appender.console.type = CONSOLE
> appender.console.layout.type = PatternLayout
> appender.console.layout.pattern = %d{-MM-dd HH:mm:ss,SSS} %-5p %-60c %x
> - %m%n # Log all infos in the given rolling file
> appender.rolling.name = RollingFileAppender appender.rolling.type =
> RollingFile appender.rolling.append = false
> appender.rolling.fileName = ${sys:log.file} appender.rolling.filePattern
> = ${sys:log.file}.%i appender.rolling.layout.type = PatternLayout
> appender.rolling.layout.pattern = %d{-MM-dd HH:mm:ss,SSS} %-5p %-60c %x
> - %m%n
[jira] [Commented] (FLINK-31860) FlinkDeployments never finalize when namespace is deleted
[
https://issues.apache.org/jira/browse/FLINK-31860?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17714393#comment-17714393
]
Gyula Fora commented on FLINK-31860:
Thank you for the detailed analysis , event triggering seems to be the culprit
here , you are completely right.
We should avoid triggering events or sending status updates in the cleanup
step. We can work on the fix and I will cc you for the review so you can get
some familiarity with the code:)
> FlinkDeployments never finalize when namespace is deleted
> -
>
> Key: FLINK-31860
> URL: https://issues.apache.org/jira/browse/FLINK-31860
> Project: Flink
> Issue Type: Bug
> Components: Kubernetes Operator
>Affects Versions: kubernetes-operator-1.3.1
> Environment: Apache Flink Kubernetes Operator 1.3.1
> Kubernetes 1.24.9
>Reporter: Jayme Howard
>Priority: Minor
>
> This appears to be a pretty straightforward issue, but I don't know the
> codebase well enough to propose a fix. When a FlinkDeployment is present in
> a namespace, and the namespace is deleted, the FlinkDeployment never
> reconciles and fails to complete its finalizer. This leads to the namespace
> being blocked from deletion indefinitely, requiring manual manipulation to
> remove the finalizer on the FlinkDeployment.
>
> Namespace conditions:
> {code:java}
> conditions:
> - lastTransitionTime: '2023-04-18T22:17:48Z'
> message: All resources successfully discovered
> reason: ResourcesDiscovered
> status: 'False'
> type: NamespaceDeletionDiscoveryFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: All legacy kube types successfully parsed
> reason: ParsedGroupVersions
> status: 'False'
> type: NamespaceDeletionGroupVersionParsingFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: All content successfully deleted, may be waiting on finalization
> reason: ContentDeleted
> status: 'False'
> type: NamespaceDeletionContentFailure
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: 'Some resources are remaining: flinkdeployments.flink.apache.org
> has 2
> resource instances'
> reason: SomeResourcesRemain
> status: 'True'
> type: NamespaceContentRemaining
> - lastTransitionTime: '2023-03-23T18:27:37Z'
> message: 'Some content in the namespace has finalizers remaining:
> flinkdeployments.flink.apache.org/finalizer
> in 2 resource instances'
> reason: SomeFinalizersRemain
> status: 'True'
> type: NamespaceFinalizersRemaining
> phase: Terminating {code}
> FlinkDeployment example (some fields redacted):
> {code:java}
> apiVersion: flink.apache.org/v1beta1
> kind: FlinkDeployment
> metadata:
> creationTimestamp: '2023-03-23T18:27:02Z'
> deletionGracePeriodSeconds: 0
> deletionTimestamp: '2023-03-23T18:27:35Z'
> finalizers:
> - flinkdeployments.flink.apache.org/finalizer
> generation: 3
> name:
> namespace:
> resourceVersion: '10565277081'
> uid: e50d2683-6c0c-467e-b10c-fe0f4e404692
> spec:
> flinkConfiguration:
> taskmanager.numberOfTaskSlots: '2'
> flinkVersion: v1_16
> image:
> job:
> args: []
> entryClass:
> jarURI:
> parallelism: 2
> state: running
> upgradeMode: stateless
> jobManager:
> replicas: 1
> resource:
> cpu: 1
> memory: 2048m
> logConfiguration:
> log4j-console.properties: '# This affects logging for both user code and
> Flink rootLogger.level = INFO rootLogger.appenderRef.console.ref =
> ConsoleAppender rootLogger.appenderRef.rolling.ref = RollingFileAppender
> # Uncomment this if you want to _only_ change Flink''s logging
> #logger.flink.name = org.apache.flink #logger.flink.level = INFO #
> The following lines keep the log level of common libraries/connectors on
> # log level INFO. The root logger does not override this. You have to
> manually # change the log levels here. logger.akka.name = akka
> logger.akka.level = INFO logger.kafka.name= org.apache.kafka
> logger.kafka.level = INFO logger.hadoop.name = org.apache.hadoop
> logger.hadoop.level = INFO logger.zookeeper.name = org.apache.zookeeper
> logger.zookeeper.level = INFO # Log all infos to the console
> appender.console.name = ConsoleAppender appender.console.type = CONSOLE
> appender.console.layout.type = PatternLayout
> appender.console.layout.pattern = %d{-MM-dd HH:mm:ss,SSS} %-5p %-60c %x
> - %m%n # Log all infos in the given rolling file
> appender.rolling.name = RollingFileAppender appender.rolling.type =
> RollingFile appender.rolling.append = false
> appender.rolling.fileName = ${sys:log.file} appender.rolling.filePattern
> = ${sys:log.file}.%i
