Application set v0.3.0 log regression

[enys]

When running:

docker run --rm -it quay.io/argoproj/argocd-applicationset:v0.3.0 applicationset-controller

It errors out, producing 0 logs
Compared to

docker run --rm -it quay.io/argoproj/argocd-applicationset:v0.2.0 applicationset-controller

This makes debugging CI failures, such as argoproj/argo-helm#1070 tricky.

[mkilchhofer]

Also analysed argoproj/argo-helm#1070. When switching to json logformat, it seems to work again. Maybe due to PR #373

--logformat=text:

$ kubectl logs -f argocd-applicationset-556688f545-sxk96
time="2022-01-05T21:15:15Z" level=info msg="Starting configmap/secret informers"
time="2022-01-05T21:15:15Z" level=info msg="Configmap/secret informer synced"
<<<<< crash with Exit 1

--logformat=json:

$ kubectl logs -f argocd-applicationset-74cb685876-29jbr
{"level":"info","ts":1641416299.9679651,"logger":"setup","msg":"ApplicationSet controller v0.3.0 using namespace 'default'","namespace":"default","COMMIT_ID":"fb33d2c26918be40e1d63d50eeab4f4ea2d16f01"}
{"level":"info","ts":1641416300.5976462,"logger":"controller-runtime.metrics","msg":"metrics server is starting to listen","addr":":8080"}
{"level":"info","msg":"Starting configmap/secret informers","time":"2022-01-05T20:58:20Z"}
{"level":"info","msg":"Configmap/secret informer synced","time":"2022-01-05T20:58:20Z"}
{"level":"error","ts":1641416300.7069428,"logger":"setup","msg":"failed to create webhook handler","error":"Failed to get argocd settings: configmap \"argocd-cm\" not found","stacktrace":"main.main\n\t/workspace/main.go:158\nruntime.main\n\t/usr/local/go/src/runtime/proc.go:225"}

Hmmm... os.Exit(1). Really a hard fail, interesting...
https://github.com/argoproj-labs/applicationset/pull/341/files#diff-2873f79a86c0d8b3335cd7731b0ecf7dd4301eb19a82ef7a1cba7589b5252261R210-R221

[enys]

Oh thanks, @mkilchhofer do you think it would make sense to switch to JSON format for the tests on the helm chart for now, making this a separate concern ?

[mkilchhofer]

Hmmm I don't feel confident to workaround a regression introduced in 0.3.0 for the helm chart build/testing. We should fix it here in the upstream project and wait.
Up to v0.2.0 we could start the Appset pod without deploying ArgoCD.