[FLINK-31187] Standalone HA mode does not work if dynamic properties are supplied

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: kubernetes-operator-1.4.0
• Fix Version/s: kubernetes-operator-1.5.0
• Component/s: Kubernetes Operator
• Labels: pull-request-available
• Assignee: Mate Czagany
• Reporter: Mate Czagany
• Created: 22/Feb/23 18:09
• Updated: 18/Jun/23 15:29
• Resolved: 01/Mar/23 15:26

Description

With FLINK-30518 '--host $(POD_IP)' has been added to the arguments of the JMs which fixes the issue with HA on standalone mode, but it always gets appended to the end of the final JM arguments: https://github.com/usamj/flink-kubernetes-operator/blob/72ec9d384def3091ce50c2a3e2a06cded3b572e6/flink-kubernetes-standalone/src/main/java/org/apache/flink/kubernetes/operator/kubeclient/decorators/CmdStandaloneJobManagerDecorator.java#L107

But this will not be parsed properly in case any dynamic properties were set in the arguments, e.g.:

 Program Arguments:
   --configDir
   /opt/flink/conf
   -D
   jobmanager.memory.off-heap.size=134217728b
   -D
   jobmanager.memory.jvm-overhead.min=201326592b
   -D
   jobmanager.memory.jvm-metaspace.size=268435456b
   -D
   jobmanager.memory.heap.size=469762048b
   -D
   jobmanager.memory.jvm-overhead.max=201326592b
   --job-classname
   org.apache.flink.streaming.examples.statemachine.StateMachineExample
   --test
   test
   --host
   172.17.0.11

You can verify this bug by using the YAML I've attached and in the JM logs you can see this line: 

Remoting started; listening on addresses :[akka.tcp://[e-mail]:6123]

Without any program arguments supplied this would correctly be:

Remoting started; listening on addresses :[akka.tcp://flink@172.17.0.8:6123]

I believe this could be easily fixed by appending the --host parameter before JobSpec.args and if a committer can assign this ticket to me I can create a PR for this.

Issue Links

• is related to: FLINK-30518 [flink-operator] Kubernetes HA Service not working with standalone mode
• links to: GitHub Pull Request #534