oc doesn't time out watches on target server disappearing

[jimmidyson]

oc watches don't handle target server disappearing ungracefully (e.g. power off).

Version

$openshift version                         

openshift v1.3.0-alpha.2+10133fa
kubernetes v1.3.0+57fb9ac
etcd 2.3.0+git

Server https://192.168.42.116:8443
OpenShift v1.3.0
Kubernetes v1.3.0+52492b4

Steps To Reproduce

1. Start OpenShift server & log in
2. oc get pods -w
3. Power off OpenShift host

Current Result

oc client does not close

Notice that netstat -anp|grep 8443 shows connection still established.

Expected Result

oc client should close connections & report error (or potentially reconnect transparently).

Additional Information

I assume this is because WS read timeouts are set to infinity. Would be better to implement WS ping (I've confirmed API server handles this properly) & set a reasonable read timeout.

[liggitt]

oc doesn't use websockets, it uses chunked streaming connections. Not sure what the client side timeout on the connection is

[jimmidyson]

Ah I didn't know that. Exhibits same behaviour though from what I can see so assume read timeout is being disabled - that would be valid of course, receiving no data for a long time would be fine if connection was still actually valid.

[juanvallejo]

Adding a client request timeout here: #11104

[openshift-bot]

Issues go stale after 90d of inactivity.

Mark the issue as fresh by commenting /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.
Exclude this issue from closing by commenting /lifecycle frozen.

If this issue is safe to close now please do so with /close.

/lifecycle stale

[juanvallejo]

@jimmidyson a client-side request timeout has been added.
I am closing this issue, but please re-open if it is still not fixed.