Hey Team, released the new `flink-k8s-operator` ve...
# troubleshooting
v
Hey Team, released the new
flink-k8s-operator
version
1.12
and we notice a warning "Job Not Found" in the k8s events of the target namespace. The operator is able to see the correct jobId. I can see it in the logs. But for some reason it does use it, but instead uses some other value. Any idea what is going on ? Relevant logs here:
Copy code
flink-kubernetes-operator-569944464f-4dd8c flink-kubernetes-operator 2025-08-04 21:24:39,941 o.a.f.r.r.RestClient           [DEBUG] Received response {"jobs":[{"jid":"000000002edab9a50000000000000000","name":"streaming-canary-job","start-time":1754342059808,"end-time":-1,"duration":620135,"state":"RUNNING","last-modification":1754342500559,"tasks":{"running":4,"canceling":0,"canceled":0,"total":4,"created":0,"scheduled":0,"deploying":0,"reconciling":0,"finished":0,"initializing":0,"failed":0}}]}.

// more logs here

flink-kubernetes-operator-569944464f-4dd8c flink-kubernetes-operator 2025-08-04 21:25:25,158 o.a.f.r.r.RestClient           [DEBUG] Sending request of class class org.apache.flink.runtime.rest.messages.EmptyRequestBody to flink-canary-rest.flink-canary:8081/v1/jobs/overview
flink-kubernetes-operator-569944464f-4dd8c flink-kubernetes-operator 2025-08-04 21:25:25,197 o.a.f.r.r.RestClient           [DEBUG] Sending request of class class org.apache.flink.runtime.rest.messages.EmptyRequestBody to flink-canary-rest.flink-canary:8081/v1/config
flink-kubernetes-operator-569944464f-4dd8c flink-kubernetes-operator 2025-08-04 21:25:25,202 o.a.f.r.r.RestClient           [DEBUG] Sending request of class class org.apache.flink.runtime.rest.messages.EmptyRequestBody to flink-canary-rest.flink-canary:8081/v1/taskmanagers
flink-kubernetes-operator-569944464f-4dd8c flink-kubernetes-operator 2025-08-04 21:25:25,210 o.a.f.r.r.RestClient           [DEBUG] Sending request of class class org.apache.flink.runtime.rest.messages.EmptyRequestBody to flink-canary-rest.flink-canary:8081/v1/jobs/9bfdca22ec0b09ca745d20cd99c5cb6a/checkpoints
This looks similar except the job is not stuck :https://issues.apache.org/jira/browse/FLINK-38106
@Matyas Orhidi @Gyula F贸ra Is this something you can help with ? This looks like a regression since 1.12.
It looks like if flink generates a new job id, then the operator has no way of figuring out what the new job id is.
g
Please open a jira ticket , I am on holiday right now
馃憤 1