I was able to get the pod ip using "oc describe pod <docker-registry>" and test it with "curl http://<podipaddress>:5000/healthz" It returned nothing and a 0 for an exit code, so I assumed that was working.
But the key was that I was able to see the nodes attempting to reach my pod at 10.1.0.3/healthz in my squid proxy logs.
I found that I was missing the 10.1.0.0/16 entry in the NO_PROXY entries for the master and the node configs
I edited the NO_PROXY entries in /etc/sysconfig/origin-master and /etc/sysconfig/origin-node and restarted both.
I could then see the registry start right up!
Availity | 10752 Deerwood Park Blvd S. Ste 110, Jacksonville FL 32256
From: Paul Weil <pweil redhat com>
Date: Thursday, April 28, 2016 at 1:14 PM
To: Gary Franczyk <gary franczyk availity com>
Subject: Re: Pushing to internal registry
Thanks for the additional info Gary. This definitely looks like it is failing the readiness check and restarting the pod.
The readiness check is performed against the pod ip by the node service. I would take a look at https://docs.openshift.org/latest/install_config/http_proxies.html which has some examples of adding the pod and service CIDRs to the NO_PROXY setting in the master and node configs.
You can also try curling the health check url on the host by setting the proxy settings locally and curling the pod ip:port/healthz. Please note that using a CIDR in the NO_PROXY setting is something supported in code so you'll want to manually add the individual IP for your local test .
Let me know if that doesn't help. Thanks for your patience in sorting this out!
On Thu, Apr 28, 2016 at 10:47 AM, Gary Franczyk <Gary Franczyk availity com> wrote: