I'm using OKD 4.15 and have accidentally deleted the Argo namespace. After that, I've manually recreated it and redeployed Argo via Helm. The pods cannot start as unable to connect to the API server.
time="2025-01-14T14:46:08Z" level=info msg="ArgoCD ApplicationSet Controller is starting" built="2023-03-23T14:57:27Z" commit=5bcd846fa16e4b19d8f477de7da50ec0aef320e5 namespace=argocd version=v2.6.7+5bcd846
time="2025-01-14T14:46:38Z" level=error msg="Get \https://172.30.0.1:443/api?timeout=32s\: dial tcp 172.30.0.1:443: i/o timeoutunable to start manager"
I've deployed a debug pod with ubi image to test the connectivity in both the Argo namespace and another one. Only the one deployed in the Argo namespace cannot connect to the API server. No network policy is present, the service account used was the same. The issue is surely at network level.
The issue was a missing CIDR block in the EgressFirewall rule.