Controller fails to start if proxy variables are set

[netthier]

Controller Version

v0.15.0

Kubernetes Version

v1.21.6+vmware.1

Bug description

If the controller is deployed with the HTTP_PROXY or HTTPS_PROXY variables set, the controller seemingly fails to start.
No logs are produced, but the status shows the following:

Readiness probe failed: Get "http://192.168.5.116:9440/readyz": dial tcp 192.168.5.116:9440: connect: connection refused
Liveness probe failed: Get "http://192.168.5.116:9440/healthz": dial tcp 192.168.5.116:9440: connect: connection refused

The controller starts up fine if those variables are not set, but then later fails reconciling deployments as it can't access the remote git repositories (due to a lack of proxy).

Steps to reproduce

Specify either HTTP_PROXY or HTTPS_PROXY variables in the deployment.

Relevant log output

No logs were produced

[netthier]

I suspect this may be due to the controller trying to access the Kubernetes API through the proxy, which won't work.
I tried adding .cluster.local to NO_PROXY, but that didn't help. I'll look into this further.

Update: Adding 10.96.0.1 to NO_PROXY fixed this issue.
Since the controller now directly pulls git repos from their sources, configuring a proxy might be something users want to do, so maybe documenting this somewhere would be useful?
Or at least some visible logs that indicate the Kubernetes API server cannot be reached.