Registry Readiness and liveness probe failures for http2: no cached connection was available

Issue

• Liveness probe failed: Get https://registry1.example.com:5000/healthz: http2: no cached connection was available
• Readiness probe failed: Get https://registry1.example.:5000/healthz: http2: no cached connection was available
• The docker registry pod is being restarted from time to time.
• logging similar to below seen

Jun 16 18:32:48 node1.example.com atomic-openshift-node[1678]: I0616 18:32:48.993207    1678 prober.go:106] Liveness probe for "docker-registry-1-6fm3d_default(12ee05ee-ad9d-11e7-8c22-02010ac97bd9):registry" failed (failure): Get https://127.0.0.1:5000/healthz: http2: no cached connection was available
Jun 16 18:32:58 node1.example.com atomic-openshift-node[1678]: I0616 18:32:58.989816    1678 prober.go:106] Liveness probe for "docker-registry-1-6fm3d_default(12ee05ee-ad9d-11e7-8c22-02010ac97bd9):registry" failed (failure): Get https://127.0.0.1:5000/healthz: http2: no cached connection was available
Jun 16 18:33:08 node1.example.com atomic-openshift-node[1678]: I0616 18:33:08.993434    1678 prober.go:106] Liveness probe for "docker-registry-1-6fm3d_default(12ee05ee-ad9d-11e7-8c22-02010ac97bd9):registry" failed (failure): Get https://127.0.0.1:5000/healthz: http2: no cached connection was available
Jun 16 18:33:08 node1.example.com atomic-openshift-node[1678]: I0616 18:33:08.993486    1678 kubelet.go:1883] SyncLoop (container unhealthy): "docker-registry-1-6fm3d_default(12ee05ee-ad9d-11e7-8c22-02010ac97bd9)"
Jun 16 18:33:09 node1.example.com atomic-openshift-node[1678]: I0616 18:33:09.298053    1678 kuberuntime_manager.go:501] pod "docker-registry-1-6fm3d_default(12ee05ee-ad9d-11e7-8c22-02010ac97bd9)" container "registry" is unhealthy, it will be killed and re-created.