RC card is pending while it should be in error state

[bryk]

It has pending pods that fail to schedule. It is pending though.

@floreks opinions?

[floreks]

I believe event reason string doesn't contain failed in this case. Maybe error? Partial checks have to be extended I believe

[floreks]

Could you provide me error message from events that was not recognized so I can track it?

[bryk]

[bryk]

Name:               kube-dns-v9-xb42d
Namespace:          kube-system
Image(s):           gcr.io/google_containers/etcd:2.0.9,gcr.io/google_containers/kube2sky:1.11,gcr.io/google_containers/skydns:2015-10-13-8c72f8c,gcr.io/google_containers/exechealthz:1.0
Node:               gke-guestbook-f35bcada-node-yxnp/
Start Time:         Tue, 22 Dec 2015 00:53:36 +0100
Labels:             k8s-app=kube-dns,kubernetes.io/cluster-service=true,version=v9
Status:             Failed
Reason:             CapacityExceeded
Message:            Pod cannot be started due to exceeded capacity

[floreks]

This is interesting. I can't really reproduce that myself and it will be hard to track exact problem without that. Events table recognized warnings correctly. I can add additional reason partials but i can't even track this CapacityExceeded in kubernetes code.

[floreks]

I've tracked the code that is doing this: https://github.com/kubernetes/kubernetes/pull/5501/files#diff-bf28da68f62a8df6e99e447c4351122dR1676

It is actually removed on newest k8s HEAD :) Still this partial should be recognized as error. I will add that and some other partial strings that may indicate errors. I think this is good base: https://github.com/kubernetes/kubernetes/blob/53f0f9d59860131c2be301a0054adfc86e43945d/pkg/kubelet/container/event.go

[bryk]

Ok, describe in your commit where did you get the list of partials from.

[bryk]

Corner case. Off to post 1.0