workers are deployed with a healthcheck

[bacongobbler]

For example, deploying example-dockerfile-procfile-http:

><> deis scale worker=1
Scaling processes... but first, coffee!
done in 128s
=== www Processes
--- worker:
www-v2-worker-sy6c2 starting (v2)
--- web:
www-v2-web-dp2h0 up (v2)

From checking the pod:

readinessProbe:
      failureThreshold: 1
      initialDelaySeconds: 5
      periodSeconds: 5
      successThreshold: 1
      tcpSocket:
        port: 80
      timeoutSeconds: 5

The worker process does not run the API server, therefore it will always fail the health check. Non-web processes should not have a healthcheck.

[bacongobbler]

Notice that this is causing worker deployments to be really slowwwwwww (128 seconds, to be exact)

[helgi]

Before we only allowed routable applications through: af76733#diff-3b6e1f09121075ad67f216aa2bbfc9a6L949

Quick fix is to do that again

Adding tests to ensure only routable apps can get healthchecks would be nice. Note that the old code didn't raise anything, just ignored non-routable types.