feat: implement config cache-proxy command

[tisutisu]

This PR implements the konflux-build-cli config cache-proxy --enable <true/false> cmd, to be used during build-definitions init task replacement using golang CLI.

[snyk-io]

✅ Snyk checks have passed. No issues have been found so far.

Status	Scanner	Critical	High	Medium	Low	Total (0)
✅	Open Source Security	0	0	0	0	0 issues
✅	Licenses	0	0	0	0	0 issues
✅	Code Security	0	0	0	0	0 issues

💻 Catch issues earlier using the plugins for VS Code, JetBrains IDEs, Visual Studio, and Eclipse.

[tkdchen]

Should this piece of code in else branch?

[tkdchen]

Otherwise setting default*Proxy has no effect. They are reassigned by reading http-proxy and no-proxy from cm.Data.

[tisutisu]

have felt this same while writing the logic, but kept the logic same as per the current bash implementation here

[tkdchen]

The below if-else is relative to the link you mentioned.

https://github.com/konflux-ci/build-definitions/blob/a91adc4f10bf5dc684fe32b373fd718253945cd4/task/init/0.3/init.yaml#L44-L61 is relative to the commented lines I think. Code from line 44 to 61 has this logic:

• if configmap is present, http_proxy and no_proxy are assigned according to the config data.
• if configmap is not present, use the defaults.

[tisutisu]

Got your point now, thanks for catching it.
Addressed it on the latest commit 787fee7

[tkdchen]

Maybe we can also wrap kubectl (it's included in task-runner) command instead of calling kubenetes APIs, which is similar with other nested commands. Invoking kubectl could also help us having less effort on maintaining dependencies updates.

[tisutisu]

Maybe we can also wrap kubectl (it's included in task-runner) command instead of calling kubenetes APIs, which is similar with other nested commands. Invoking kubectl could also help us having less effort on maintaining dependencies updates.

@tkdchen even if we use cli, for reading/unmarshalling of kubectl cmd outputs, we need to depend on the kubernetes pkgs like k8s.io/api/core/v1, otherwise we have to duplicate code in our own implementation for reading ConfigMaps, not sure if we should go with wrapper around kubectl. WDYT?

[mmorhun]

This might be out of scope of this PR, but it anyway blocks it.
The main problem is that we accessing k8s cluster directly from within CLI making it tightly bound and dependent on k8s environment, making it impossible to run without a cluster, say locally or in another than Tekton CI. (Obviously, there are workarounds, but we need a proper solution).

My suggestion is to create a abstraction layer, say ConfigReader that would detect the environment (easiest is to use env var) and then read k8s resources if run within k8s cluster or access config in different source (say ini or env file) otherwise.

[mmorhun]

I think we need ConfigReader interface with a factory (NewConfigReader) that constructs required implementation (K8sConfigReader or EnvFileConfigReader) based on an env var (if not set, k8s is the default). Then config command(s) can use it.