Error "http: TLS handshake error" from F5 IP in Openshift Container Platform 3

Issue

When the Nginx talks to master servers via F5, there are intermittently timeouts and TLS certificate errors at atomic-openshift-master-api, where 10.0.0.0 is the F5 IP.

atomic-openshift-master-api[90708]: rest.go:324] Starting watch for /oapi/v1/netnamespaces, rv=325366149 labels= fields= timeout=8m38s
atomic-openshift-master-api[90708]: logs.go:41] http: TLS handshake error from 10.0.0.0:35163: EOF
atomic-openshift-master-api[90708]: logs.go:41] http: TLS handshake error from 10.0.0.1:49844: EOF
atomic-openshift-master-api[90708]: logs.go:41] http: TLS handshake error from 10.0.0.0:36886: EOF
atomic-openshift-master-api[90708]: logs.go:41] http: TLS handshake error from 10.0.0.1:51570: EOF
atomic-openshift-master-api[90708]: logs.go:41] http: TLS handshake error from 10.0.0.0:38607: EOF
atomic-openshift-master-api[90708]: logs.go:41] http: TLS handshake error from 10.0.0.1:53296: EOF
atomic-openshift-master-api[90708]: trace.go:76] Trace[499470393]: "List /oapi/v1/builds" (started: 2018-06-10 20:20:03.982300412 -0700 MST) (total time: 1.275543598s):
atomic-openshift-master-api[90708]: Trace[499470393]: [312.182118ms] [312.166927ms] Listing from storage done
atomic-openshift-master-api[90708]: Trace[499470393]: [1.275540498s] [959.465371ms] Writing http response done (1934 items)