Hostname is incorrect after deploying to k8s

I am using `Cluster.Strategy.Kubernetes.DNSSRV` for the strategy. Everything works fine on the initial deploy. But after a few deploys I get the following error:

```
[backend-elixir-0] {"context":{"runtime":{"application":"libcluster","file":"lib/logger.ex","function":"error/2","line":17,"module_name":"Cluster.Logger","vm_pid":"<0.3785.0>"},"system":{"hostname":"backend-elixir-0","pid":1}},"dt":"2019-10-06T23:46:44.601190Z","event":null,"level":"error","message":"[libcluster:k8s] 'backend-elixir.backend-elixir-staging.svc.cluster.local.' : lookup against backend-elixir failed: :nxdomain"}
[backend-elixir-0] {"context":{"runtime":{"application":"libcluster","file":"lib/logger.ex","function":"error/2","line":17,"module_name":"Cluster.Logger","vm_pid":"<0.3785.0>"},"system":{"hostname":"backend-elixir-0","pid":1}},"dt":"2019-10-06T23:46:44.604630Z","event":null,"level":"error","message":"[libcluster:k8s] 'backend-elixir.backend-elixir-staging.svc.cluster.local.' : lookup against backend-elixir failed: :nxdomain"}
```

The only way to recover from it is delete the mnesia cache disk for each of the pods and restart everything.

The only thing using mnesia so far is [pow](https://powauth.com).

I am new to Elixir, but I am happy to gather more information if someone points me in the right direction.