Make the helm chart runable on any k8s cluster #702
Description
Is your feature request related to a problem? Please describe.
The current helm chart is very opinionated. You cannot install it on any random cluster. If you try you will find these issues:
- Hardcoded namespaces (in a multi purpose cluster nothing should run in a default ns)
- Non secure pod settings which rely on host mounts
- Pods which cannot be moved from one host to another host because pvcs are only used for rabbitmq and thanos but not for minio, clickhouse, mariodb (they all use host mounts)
- Some tools are installed using the helm chart which are most likely not needed at all like filebeat (because a typical cluster already has a log file collector and prometheus monitoring up and running)
- Some pods have very strange image pull settings produces ErrImageNeverPull errors
- Hardcoded storage classes for some pods like rabbitmq (the default storage class should be used by default and should be changeable)
- Hardcoded ingress class (some clusters might use Traefik or ALB ingress)
- No security settings like securityContext
- Components cannot be replaced by PaaS offerings (e.g. replace rabbitmq or mariadb with some cloud provider service like AmazonMQ or RDS)
Describe the solution you'd like
I would like to install this tool like any other tool using helm. The helm chart should work with default settings without much customization. It should run just fine on a typical EKS bottlerocket cluster.
Describe alternatives you've considered
I have tried to highlight and fix individual issues first like #701 #700 #691 #692
but the problem seems to be much larger than each individual issue.
Additional context