[GitHub] menew commented on issue #105: invoker-0 stuck in CrashLoopBackOff loop probably due to issues creating a kafka topic

[GitBox <gi...@apache.org>]

menew commented on issue #105: invoker-0 stuck in CrashLoopBackOff loop probably due to issues creating a kafka topic
URL: https://github.com/apache/incubator-openwhisk-deploy-kube/issues/105#issuecomment-349925831
 
 
   Hi, I've experienced almost same issue on controller deployment. Controller pods seem to be started, but suddenly shutdown with error message `ensureTopic for completed0 failed` after 120 seconds of silence (I removed `livenessProve` key to debug the pod).
   
   After investigating, I found that this failure is caused by misconfiguration of Kafka pod. I simply deployed Kafka by `kafka.yml` from this repository, then it looked booted up correctly and port 9092 was reachable from other pods on TCP layer. But, without kube-dns, it seems that consumers/producers cannot reach the Kafka pod on application layer because name `kafka.openwhisk` set in `KAFKA_ADVERTISED_HOSTNAME` is not resolvable.
   
   So, I replaced value `kafka.openwhisk` by `$(KAFKA_SERVICE_HOST)` in deployment YAML file, then controller pods started correctly. They responded to `curl` on `/ping` and some `wsk` commands.
   
   Of course my issue can be different from this one, but sanity check on your Kafka deployment might help (for example, I followed "Usage Quick Start" steps in https://github.com/ches/docker-kafka README with `docker run -it --net host --rm ches/kafka bash` container).
   

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


With regards,
Apache Git Services