Add extra health check for k8s installation #266
Conversation
Currently, contiv k8s installation only check contiv related pod status. This is not sufficient enough as there is a chance that we cannot run any other pod. This ticket is to enhance k8s installation to ensure that we can start a regular pod successfully. Signed-off-by: Kahou Lei <[email protected]>
|
@kahou82: I've looked into this. Unfortunately, we need to add this check after default-net is created. The bootstrap of the cluster never finishes otherwise. The default-net isn't created if we don't finish the setup without error in install.sh. A good place to have this test would be after this block https://github.com/contiv/install/blob/master/scripts/kubeadm_test.sh#L57 Once the pod is fine, we should also clean it up. |
|
|
||
| for i in {0..15}; do | ||
| sleep 2 | ||
| $kubectl get pods | grep test-pod | grep -v "Running" && continue |
There was a problem hiding this comment.
you may want to check all namespaces
|
+1 to @unclejack that we need the default-net created before the pods can run. I dont think we want to check all namespaces as system namespace will have pods running with host=net. If possible, can you add a message saying that installation is complete and you are starting some kind of a self test. That way if things do fail people will know what to do next - may be point them to contiv support FAQ page or something :) |
Currently, contiv k8s installation only check contiv related pod status.
This is not sufficient enough as there is a chance that we cannot
run any other pod.
This ticket is to enhance k8s installation to ensure that we can
start a regular pod successfully.
Signed-off-by: Kahou Lei [email protected]