Description of problem:

 Randomly one of the health checks on the registry fails. 


Version-Release number of selected component (if applicable):
OCP 3.5.5.8

How reproducible:
Happens intermittently 

Actual results:

 atomic-openshift-node[36648][prober.go:106] Readiness probe for "docker-registry-12-lknh8_default(91642388-5c0a-11e7-9c80-005056961e9a):registry" failed (failure): Get https://10.1.4.60:5000/healthz: http2: no cached connection was available

 dockerd-current[1501]: 10.1.4.1 - - [28/Jun/2017:17:17:01 +0000] "GET /healthz HTTP/2.0" 200 0 "" "Go-http-client/2.0"

Expected results:

 dockerd-current[1501]: 10.1.4.1 - - [28/Jun/2017:17:17:11 +0000] "GET /healthz HTTP/2.0" 200 0 "" "Go-http-client/2.0"
 dockerd-current[1501]: 10.1.4.1 - - [28/Jun/2017:17:17:11 +0000] "GET /healthz HTTP/2.0" 200 0 "" "Go-http-client/2.0"


Additional info:

Possible upstream issue: 

  https://github.com/golang/go/issues/16582
  https://github.com/golang/go/commit/7a622740655bb5fcbd160eb96887032314842e6e