Fix resource constraints

[dmotte]

The documentation about postgres_pod_resources says that you can disable the defaults by setting them to zero OR empty string:

Empty string or 0 disables the default.

Source:

postgres-operator/docs/reference/operator_parameters.md

Lines 570 to 604 in 2582b93

	## Kubernetes resource requests
	This group allows you to configure resource requests for the Postgres pods.
	Those parameters are grouped under the `postgres_pod_resources` key in a
	CRD-based configuration.
	* **default_cpu_request**
	CPU request value for the Postgres containers, unless overridden by
	cluster-specific settings. Empty string or `0` disables the default.
	* **default_memory_request**
	memory request value for the Postgres containers, unless overridden by
	cluster-specific settings. Empty string or `0` disables the default.
	* **default_cpu_limit**
	CPU limits for the Postgres containers, unless overridden by cluster-specific
	settings. Empty string or `0` disables the default.
	* **default_memory_limit**
	memory limits for the Postgres containers, unless overridden by cluster-specific
	settings. Empty string or `0` disables the default.
	* **max_cpu_request**
	optional upper boundary for CPU request
	* **max_memory_request**
	optional upper boundary for memory request
	* **min_cpu_limit**
	hard CPU minimum what we consider to be required to properly run Postgres
	clusters with Patroni on Kubernetes.
	* **min_memory_limit**
	hard memory minimum what we consider to be required to properly run Postgres
	clusters with Patroni on Kubernetes.

But it's not currently possible to set them to empty string, due to the regular expressions in charts/postgres-operator/crds/operatorconfigurations.yaml.

Also, the enforceMinResourceLimits function currently supports only the empty string case, which is impossible to use due to the aforementioned regexes.

This PR aims to fix these two issues 🙂

[FxKu]

@dmotte thanks! Could you provide also unit tests that cover this "new" behavior so I could also try them on the main branch and see what doesn't work? This would be a great improvement to this PR.

[dmotte]

@FxKu done 🙂 I hope it's ok

[FxKu]

Seems to me that only the schema validation needs to change. It's still good though to add a "0" example to the unit test.

[dmotte]

@FxKu yeah, I think I got your point and you are right :)

Just to double check, correct me if I'm wrong: util.IsSmallerQuantity(somePositiveQuantity, "0") returns false, so isSmaller ends up being false and the limit is not enforced.

Thank you! Fixed in 24df591

[FxKu]

Thanks. Can you also update these files with the new validation pattern:

• CRD Manifest example
• CRD go representation

[dmotte]

@FxKu I think you meant manifests/operatorconfiguration.crd.yaml instead of manifests/postgresql.crd.yaml, so I updated that one instead. Done in 107fe9b 😉 let me know if it's OK or you have other requests

[FxKu]

👍