Customize namespaces of Webhooks

[jacksgt]

Hello,

is it possible to change the location where the Webhooks are supposed to be deployed?

When installing the operator with the release YAML, I can change the location where the Webhook deployment and service are deployed.
However, the expected location (as indicated in the ValidatingWebhookConfiguration and MutatingWebhookConfiguration) is still hard-coded to tekton-operator / openshift-operators:

operator/cmd/openshift/webhook/kodata/validating-defaulting-webhook/500_webhooks.yaml

Line 29 in e4ca6e1

namespace: openshift-operators

operator/cmd/kubernetes/webhook/kodata/validating-defaulting-webhook/500-webhooks.yaml

Line 28 in e4ca6e1

namespace: tekton-operator

[vdemeester]

@jacksgt this is only for the operator webhooks themselves right ?

[jacksgt]

Yes, here I'm only referring to the Webhooks deployed directly by the operator.

[jacksgt]

In principal this should not be difficult to adjust, because the operator already has the SYSTEM_NAMESPACE environment variable.
I'm just not familiar with the codebase and don't know where the correct place to inject this modification is, otherwise I'm happy to contribute a PR :-)

[Miles-Garnsey]

I am also having some problems with this. When I apply the manifests (e.g. here and they are deployed in a custom namespace (e.g. operators) I see namespace-related webhook errors when I try to apply a TektonConfig:

for: "STDIN": Internal error occurred: failed calling webhook "webhook.operator.tekton.dev": Post "https://tekton-operator-webhook.tekton-operator.svc:443/defaulting?timeout=10s": service "tekton-operator-webhook" not found

This should not be happening, because the manifests (which I linked above) actually do set SYSTEM_NAMESPACE for all deployments, and (because it has valueFrom metadata.namespace) this should be set to operators.

So the fact that the webhook service is expected to be at tekton-operator-webhook.tekton-operator.svc (from the error message) indicates that the SYSTEM_NAMESPACE is not being picked up correctly somewhere within the operator itself.

[vdemeester]

/cc @sm43 @nikhil-thomas
I think this is fixed by #637 ?

[nikhil-thomas]

yes, #637 shoud fix this

[vdemeester]

Let's close it then 👼🏼
/close

[tekton-robot]

@vdemeester: Closing this issue.

Details

In response to this:

Let's close it then 👼🏼
/close

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.