Services sometimes not ready for connection

[pauldoomgov]

Sometime a job with a service will fail. It appears that the service app is not quite ready when the main app attempts to access it.
An example from a job that curls against the "echo" service:

curl: (7) Failed to connect to glrw-rXXXXXXXXX-pXXXXXXXXX-c0-jXXXXXXXXX-svc-echo.apps.internal port 8080 after 2 ms: Connection refused

Rerunning the job generally results in success. A hacky fix would be to add some sleep time, but having a way to block for service readiness would be preferred.