Can't hit redis using DNS name in multi-container pod on Kubernetes

Question

Can anyone shed some light on why this might not work? I can access redis just fine using the pod's IP address,

$ rdcli -h 10.152.183.65
10.152.183.65:6379> 

but when I try to use the DNS name, I cannot connect.

$ rdcli -h dbbackupd.default.cluster.local
dbbackupd.default.cluster.local:6379> (error) Redis connection to dbbackupd.default.cluster.local:6379 failed - getaddrinfo ENOTFOUND dbbackupd.default.cluster.local dbbackupd.default.cluster.local:6379

Inside the pod that is trying to contact redis, I get the right-looking output when using the host command, so name resolution looks good:

$ host dbbackupd.default.svc.cluster.local

dbbackupd.default.svc.cluster.local has address 10.152.183.65

My theory is that because redis is in a multi-container pod, that dns is not routing it to the container properly. Here is my Service's definition in case that sheds light:

apiVersion: v1
kind: Service
metadata:
  name: dbbackupd
  labels:
    app: dbbackupd
spec:
  selector:
    app: dbbackupd
  type: NodePort
  ports:
    - name: redis-mq
      protocol: TCP
      port: 6379
      targetPort: 6379
    - name: sql-comm
      protocol: TCP
      port: 1433
      targetPort: 1433

Answer 1

You're using two different host names:

$ rdcli -h dbbackupd.default.cluster.local
$ host dbbackupd.default.svc.cluster.local

The one used with rdcli is missing .svc.