Milind L created FLINK-40164:
--------------------------------
Summary: FlinkSessionJob's lifecycleState stays "UPGRADING" even
after it's suspended successfully
Key: FLINK-40164
URL: https://issues.apache.org/jira/browse/FLINK-40164
Project: Flink
Issue Type: Bug
Components: Kubernetes Operator
Affects Versions: 1.15.0
Reporter: Milind L
While running a FlinkSessionJob, if I change spec.job.state = suspended, the
job is successfully suspended, but the status.lifecycleState says "UPGRADING"
indefinitely. The Flink Web UI shows the job as canceled.
Files to reproduce this issue:
The FlinkDeployment:
{code:java}
apiVersion: flink.apache.org/v1beta1
kind: FlinkDeployment
metadata:
name: sessioncluster
namespace: default
spec:
image: idle-job:1.0
imagePullPolicy: Never
flinkVersion: v2_1
serviceAccount: flink
flinkConfiguration:
taskmanager.numberOfTaskSlots: "2"
jobManager:
resource:
cpu: 0.5
memory: 1024m
taskManager:
resource:
cpu: 0.5
memory: 1024m
{code}
The FlinkSessionJob and the code
{code:yaml}
apiVersion: flink.apache.org/v1beta1
kind: FlinkSessionJob
metadata: name: idle-job
spec:
deploymentName: sessioncluster
job:
entryClass: myclasspath.IdleJob
parallelism: 1
upgradeMode: stateless
state: running # later, suspended
{code}
{code:java}
package myclasspath;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
import org.apache.flink.streaming.api.functions.source.legacy.SourceFunction;
public class IdleJob {
public static void main(String[] args) throws Exception {
StreamExecutionEnvironment env =
StreamExecutionEnvironment.getExecutionEnvironment();
env.addSource(new IdleSource()).print();
env.execute("IdleJob");
}
public static class IdleSource implements SourceFunction<String> {
private volatile boolean running = true;
@Override
public void run(SourceContext<String> ctx) throws Exception {
while (running) {
Thread.sleep(10000);
}
}
@Override
public void cancel() {
running = false;
}
}
}
{code}
The sequence of events:
1. Create the session cluster and the session job, wait for it to start running.
2. Change the state of the job to 'suspended', apply, and wait. It will show as
canceled in the web UI, but kubectl get flinksessionjobs will not move from
"UPGRADING". Here's what it looks like for me:
{code:java}
{
"apiVersion": "flink.apache.org/v1beta1",
"kind": "FlinkSessionJob",
"metadata": {
"annotations": {
"kubectl.kubernetes.io/last-applied-configuration":
"{\"apiVersion\":\"flink.apache.org/v1beta1\",\"kind\":\"FlinkSessionJob\",\"metadata\":{\"annotations\":{},\"name\":\"idle-job\",\"namespace\":\"default\"},\"spec\":{\"deploymentName\":\"sessioncluster\",\"job\":{\"entryClass\":\"myclasspath.IdleJob\",\"parallelism\":1,\"state\":\"suspended\",\"upgradeMode\":\"stateless\"}}}\n"
},
"creationTimestamp": "2026-07-17T12:11:05Z",
"finalizers": [
"flinksessionjobs.flink.apache.org/finalizer"
],
"generation": 2,
"labels": {
"target.session": "sessioncluster"
},
"name": "idle-job",
"namespace": "default",
"resourceVersion": "26301",
"uid": "aa1fdb02-3c70-486c-adf8-e0f557594b3c"
},
"spec": {
"deploymentName": "sessioncluster",
"flinkConfiguration": {},
"job": {
"args": [],
"entryClass": "myclasspath.IdleJob",
"parallelism": 1,
"state": "suspended",
"upgradeMode": "stateless"
}
},
"status": {
"jobStatus": {
"checkpointInfo": {
"lastPeriodicCheckpointTimestamp": 0
},
"jobId": "a610d66b1a3ec486b638ff617a02e819",
"jobName": "IdleJob",
"savepointInfo": {
"lastPeriodicSavepointTimestamp": 0,
"savepointHistory": []
},
"startTime": "1784290267673",
"state": "CANCELED",
"updateTime": "1784290339023"
},
"lifecycleState": "UPGRADING",
"observedGeneration": 1,
"reconciliationStatus": {
"lastReconciledSpec":
"{\"spec\":{\"job\":{\"jarURI\":null,\"parallelism\":1,\"entryClass\":\"myclasspath.IdleJob\",\"args\":[],\"state\":\"suspended\",\"savepointTriggerNonce\":null,\"initialSavepointPath\":null,\"checkpointTriggerNonce\":null,\"upgradeMode\":\"stateless\",\"allowNonRestoredState\":null,\"savepointRedeployNonce\":null,\"autoscalerResetNonce\":null},\"restartNonce\":null,\"flinkConfiguration\":{},\"deploymentName\":\"sessioncluster\"},\"resource_metadata\":{\"apiVersion\":\"flink.apache.org/v1beta1\",\"firstDeployment\":false}}",
"lastStableSpec":
"{\"spec\":{\"job\":{\"jarURI\":null,\"parallelism\":1,\"entryClass\":\"myclasspath.IdleJob\",\"args\":[],\"state\":\"running\",\"savepointTriggerNonce\":null,\"initialSavepointPath\":null,\"checkpointTriggerNonce\":null,\"upgradeMode\":\"stateless\",\"allowNonRestoredState\":null,\"savepointRedeployNonce\":null,\"autoscalerResetNonce\":null},\"restartNonce\":null,\"flinkConfiguration\":{},\"deploymentName\":\"sessioncluster\"},\"resource_metadata\":{\"apiVersion\":\"flink.apache.org/v1beta1\",\"firstDeployment\":true}}",
"reconciliationTimestamp": 1784290269845,
"state": "UPGRADING"
}
}
}
{code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
