This is a manual base of these three commits.
https://github.com/qraft-technologies/pulsar-helm-chart/commit/e53b746a5972701574adb0b4aabb7913d8b6e927
https://github.com/qraft-technologies/pulsar-helm-chart/commit/0960ca1ddaf6d5ff853168769792a364a2fb0725
https://github.com/qraft-technologies/pulsar-helm-chart/commit/24fdb82e02e4e418a04a96e933d7b4279c03f0b8

As inference container belongs to worker pod which needs two ports
exposed from a pod, existing scrape configuration was not enough. This
commit adds additional scrape config rules for quick and dirty patch to
scrape both manager and inference container metrics.

…tream

Feature/rebase onto upstream

We decided we should turn journaling off for now, as proposed and
discussed in https://qraftec.atlassian.net/wiki/spaces/AXE/pages/126648341.

Note possible data loss when machine failure or ledger truncation.

Though it is not the cleanest, it gets the jobs done. Further changes to
axe-rest-api deployment configuration should be synced to
prometheus-deployment.yaml.

Though note that this tries to scrape ip address of cluster host, so if
your axe-rest-api server is running on baremetal but your cluster is
running on vm, it would not be able to scrape metrics. Deploy cluster on
baremetal or run your server on top of vm.

* F: kubernetes version update to v1.22

* F: remain nodePort only when it's NodePort type