[bitnami/kafka] Achieve consistent broker ids during cluster restart/rebuild

[ghShu]

It is observed that current Kafka Chart leads to increasing broker ids after each restart/rebuild of
cluster when Persistent Volumes (PVs) are not deleted during restart/rebuild. This is
often the case since PVs are usually used to persist the log messages.

This commit provides a way to have deterministic broker ids after each cluster
restart/rebuild. Please refer to this note for detailed about the issue and explanation on the solution.

This solution assumes a deployment that uses this Docker image published in Dockerhub. Alternative solutions may exist with full access to Docker images. But this consistent broker ids behavior is believed to be helpful especially when PVs are used. This can help reduce the surprise in normal operation.

[miguelaeh]

Hi @ghShu ,
Thank you very much for the PR!!!
Please bump the patch version in the Chart.yaml file

[ghShu]

Thank you @miguelaeh, Kafka Chart version is bumped to 7.3.5.

[miguelaeh]

LGTM!!

[yanjunli76]

Hi there,
There are two command if external access is enabled.

charts/bitnami/kafka/templates/statefulset.yaml

Lines 62 to 65 in 5622f31

	{{- if .Values.externalAccess.enabled }}
	command:
	- /scripts/setup.sh
	{{- end }}

charts/bitnami/kafka/templates/statefulset.yaml

Lines 207 to 210 in 5622f31

	command: ["/bin/bash", "-c"]
	args:
	- export KAFKA_CFG_BROKER_ID=${MY_POD_NAME##*-};
	exec /entrypoint.sh /run.sh;

[miguelaeh]

Hi @yanjunli76 ,
Thank you very much for reporting this, you are right. I missed that part.
Fortunately, there is already a PR fixing it: #2098