[WAS-580/WAS-575] - Upgraded strimzi Azure

[devopswolfram]

• Upgraded strimzi version from 0.23 to 0.29
• Upgraded kafka version from 2.8.0 to 3.0.0
• Upgraded kubernetes version 1.24 to 1.25
• Removed old ingress controller and replaced with helm

[ghost]

This PR is missing a github actions pipeline to validate the PR builds successfully (you want verify / test deployment - you may consider using kind)

[ghost]

Are you aware of the repercussion(s) removing this?

[devopswolfram]

Right, fixed.

[ghost]

Which storageclass are you planning to use? What if there is no default? (The deployment will fail)

[devopswolfram]

There is always a default sc in this deployment. I removed it for some reason, will check.

[ghost]

As per strimzi source code / recommendation this value is memory: 384Mi - why has it been changed to 2Gi ? how was this increased value chosen?

[devopswolfram]

For some reason kafka-persistent-kafka pods are exited with OOMKilled when initializing and it caused failure of the script. I increased the limits to 1Gi, the result was same. So I increased it to 2Gi.

Yes, in default they're 384Mi in the helm chart.
https://github.com/strimzi/strimzi-kafka-operator/blob/0.29.0/helm-charts/helm3/strimzi-kafka-operator/values.yaml#L139-L142

[devopswolfram]

We faced OOMKilled issue when testing kafka.

[(⎈|WAS-aks:kafka)]$ k top pods
NAME                                       CPU(cores)   MEMORY(bytes)   
kafka-bridge-d9958bc9c-ttcl7               113m         208Mi           
kafka-persistent-kafka-0                   404m         706Mi           
kafka-persistent-kafka-1                   736m         870Mi           
kafka-persistent-kafka-2                   385m         716Mi           
kafka-persistent-zookeeper-0               29m          475Mi           
kafka-persistent-zookeeper-1               130m         597Mi           
kafka-persistent-zookeeper-2               57m          462Mi           
strimzi-cluster-operator-d5b5c5bcb-4ntfq   123m         497Mi 

Increased limits.memory to 2Gi