all-in-one server can't start if an aggregated apiserver isnt responding

[jwforres]

1. oc cluster up with --host-config-dir and --host-data-dir set
2. Set up an aggregated apiserver running on the cluster
3. oc cluster down / up using the existing data

The origin container cant start successfully because the aggregated apiserver isn't reachable.

Seems to be an issue with all-in-one openshift and how it interacts with aggregation.

Logs:

   Starting OpenShift using container 'origin'
FAIL
   Error: could not start OpenShift container "origin"
   Details:
     Last 10 lines of "origin" container log:
     I0606 19:16:12.396882   24515 start_master.go:601] Controllers starting (*)
     I0606 19:16:12.403671   24515 start_master.go:642] Started serviceaccount-tokens controller
     I0606 19:16:12.410771   24515 logs.go:41] warning: ignoring ServerName for user-provided CA for backwards compatibility is deprecated
     W0606 19:16:12.423329   24515 node_config.go:114] Using "localhost" as node name will not resolve from all locations
     I0606 19:16:12.426837   24515 docker.go:364] Connecting to docker on unix:///var/run/docker.sock
     I0606 19:16:12.426886   24515 docker.go:384] Start docker client with request timeout=2m0s
     W0606 19:16:12.427989   24515 cni.go:157] Unable to update cni config: No networks found in /etc/cni/net.d
     I0606 19:16:12.517345   24515 start_node.go:345] Starting node localhost (v3.6.0-alpha.1+5cfbdc3-1045)
     I0606 19:16:12.522380   24515 start_node.go:354] Connecting to API server https://127.0.0.1:8443
     F0606 19:16:12.537619   24515 start_master.go:454] failed to get supported resources from server: unable to retrieve the complete list of server APIs: servicecatalog.k8s.io/v1alpha1: an error on the server ("Error: 'dial tcp 172.30.8.37:443: getsockopt: connection refused'\nTrying to reach: 'https://172.30.8.37/apis/servicecatalog.k8s.io/v1alpha1'") has prevented the request from succeeding

[liggitt]

• aggregator includes an unavailable APIService in discovery (handled by APIService status upstream, not picked to origin 3.6)
• controller manager exits if it can't successfully discover resources, which would put it in a crashloop when running standalone, but takes down the apiserver as well when running the all-in-one

[liggitt]

@deads2k did the merged picks fix this?

[deads2k]

yeah, it should finally be fixed.