[FLINK-30222] Suspended a job in last-state mode bug

• Type: Bug
• Status: Closed
• Priority: Blocker
• Resolution: Fixed
• Affects Version/s: 1.16.0, kubernetes-operator-1.2.0
• Fix Version/s: kubernetes-operator-1.3.0
• Component/s: Kubernetes Operator
• Labels: pull-request-available
• Assignee: Peter Vary
• Reporter: tanjialiang
• Created: 27/Nov/22 08:54
• Updated: 29/Nov/22 11:32
• Resolved: 29/Nov/22 11:32

Description

In flink 1.16.0, it support set kubernetes HA with options 'kubernetes', such as 'high-availability: kubernetes'. But in kubernetes operator 1.2.0, I try to suspended a job in last-state mode, it validate fail, because of 'Job could not be upgraded with last-state while Kubernetes HA disabled'.

 

I try to use kubectl patch to supsended a job with last-state

kubectl -nbigdata-flink patch flinkdeployments.flink.apache.org/streaming-638223bf650ac869689faa62-flink --type=merge -p '{"spec": {"job":
{"state": "suspended", "upgradeMode": "last-state"}

it found an error, because my kubernetes HA is disabled

Error from server: admission webhook "flinkoperator.flink.apache.org" denied the request: Job could not be upgraded with last-state while Kubernetes HA disabled 

but i enabled kubernetes HA with this follow options:

kubernetes.cluster-id: <cluster-id>
high-availability: kubernetes
high-availability.storageDir: hdfs:///flink/recovery 

and i found flink kubernetes operator 1.2.0 validate the kubernetes HA in the old options:

high-availability: org.apache.flink.kubernetes.highavailability.KubernetesHaServicesFactory 

it may be in the org.apache.flink.kubernetes.operator.utils.FlinkUtils#isKubernetesHAActivated to judge.

Issue Links

• links to: GitHub Pull Request #456