cluster: Add possibility to configure custom pg_hba

[mjtrangoni]

Hi,

This is a BREAKING CHANGE, and fixes #308.

IMHO, we should go with this kind of fixes now and taking into account that postgresql has a lot of configurations defined here, not only parameters, this fixes the way it will be handled.

Thank you!

PS: I had to remove one trailing space for the the CI to success, but in a separate commit so that you can remove it easily

[itay-grudev]

I agree with this change. The advantage of running with version < 0.x,x is that breaking changes can happen while evaluating the initial prototype in the wild.

I will release it as 0.1.0 and add the relevant changelog entry.

[itay-grudev]

@mjtrangoni Could you please run make schema? It's why the tests are failing, I think.

[mjtrangoni]

Hi @itay-grudev,

Thanks for your prompt reply. It seems I can't run it on my end, because the schema-gen does not support darwin, and is also not maintained anymore. Any hints on how to override this?

See,

$make schema
Error: unknown command "schema-gen" for "helm"
Run 'helm --help' for usage.
Error: unknown command "schema-gen" for "helm"
Run 'helm --help' for usage.
$ helm plugin install https://github.com/karuppiah7890/helm-schema-gen.git
karuppiah7890/helm-schema-gen crit platform darwin/arm64 is not supported.  Make sure this script is up-to-date and file request at https://github.com/karuppiah7890/helm-schema-gen/issues/new
Error: plugin install hook for "schema-gen" exited with error

[mjtrangoni]

@itay-grudev tests are green now. pg_hba should be an array and not an object.

[itay-grudev]

I was just going through the documentation and found, that we've documented the postgresql: as a proxy to the full configuration.

So this means that this is no longer a simply a braking change, but also a bug fix.