Hi there Trying to deploy flink as standalone clu...
# troubleshooting
e
Hi there Trying to deploy flink as standalone cluster on k8s WITHOUT operator (unfortunately I cannot install the required CRDs on my k8s cluster). I’m failing to deploy the cluster in high availability mode. I have read the documentation and added the needed configuration (running the job manager with the args [“standalone-job”, “—host”, “$POD_IP”] and with setting the pod ip env variable dinamicly of course ) (read deployment/resource-providers/standalone/kubernetes/#applicatiob-cluster-resource-definitions). When reading the generated high availability ConfigMap it seems like the job managers registered themselves with their localhost address- 127.0.0.1, instead of using the pod ip as needed. It also matches with the TaskManagers trying to connect to a localhost address… I’m wondering if the job managers using the —host property, and if not - what should I do to make them work? TLDR - trying to configure high availability on k8s WITHOUT operator failing - I would love to see examples for a working setup.
e
I have already followed this documentation and it did not work.
It just looks like the taskmanagers ignore the high availability setup and only look at the jobmanager.rpc.address configuration (even though that by the documentation the should ignore it) (if it’s not set they look at 127.0.0.1) I think that the reason HA setups works with the operator is because the operator actually changes this value in the config.yaml configuration file - and makes it directly point to the leader JM at any given time.
p
Hey there, Elad! I did manage to do it in the past. If it helps, I can jump in a call 30min trying to help if it makes sense.
e
I would love that. Contact me privately please 🙏🏽