Hi there,
I’ve noticed in all of our Everest postgres clusters, the repo1-host defined in pgbackrest-config is appended with .kubernetes.
e.g. (from the configmap):
[global]
…
repo1-host = dbname-repo-host-0.dbname-pods.production.svc.kubernetes
…
Unfortunately this is not resolvable:
bash-4.4$ host dbname-repo-host-0.dbname-pods.production.svc
dbname-repo-host-0.dbname-pods.production.svc.cluster.local has address 10.42.145.98
bash-4.4$ host dbname-repo-host-0.dbname-pods.production.svc.kubernetes
Host dbname-repo-host-0.dbname-pods.production.svc.kubernetes not found: 3(NXDOMAIN)
While I’m aware I can get around this with a manual change in the CR, and I’ve temporarily gotten around this by appending our Kubedns Corefile with:
rewrite name substring svc.kubernetes svc.cluster.local
I’m wondering if I’m missing something here that sets this as the tld in the first place, or some config within Everest that defines this? The Kubernetes cluster domain is cluster.local and I have no idea where .kubernetes has come from. In fact, I’ve found that this occurs on Everest installs both on fresh unmodified RKE2, Kubeadm and Talos clusters.
Is this a bug, or a global option that I can set? We are planning on rolling out Everest to scale, but this is currently getting in the way! 