Romana agents crashloop on worker nodes about halcyon-kubernetes HOT 13 CLOSED

any chance you could paste/upload the logs?

from halcyon-kubernetes.

Hey pritesh, I've updated the issue with what I'm seeing

from halcyon-kubernetes.

Can you paste the output from kubectl -n kube-system get pods please?
It seems that your romana-services pod hasn't launched, and so the agent containers cannot initialize.

from halcyon-kubernetes.

ubuntu@kube1:~$ kubectl -n kube-system get pods
NAME READY STATUS RESTARTS AGE
dummy-2088944543-bmnod 1/1 Running 0 27m
etcd-172.16.35.11 1/1 Running 0 26m
kube-apiserver-172.16.35.11 1/1 Running 0 27m
kube-controller-manager-172.16.35.11 1/1 Running 0 27m
kube-discovery-1150918428-6g2oo 1/1 Running 0 27m
kube-dns-654381707-aizi8 3/3 Running 0 26m
kube-proxy-3ub3s 1/1 Running 0 26m
kube-proxy-bxk6j 1/1 Running 0 26m
kube-proxy-onr05 1/1 Running 0 26m
kube-scheduler-172.16.35.11 1/1 Running 0 26m
kubernetes-dashboard-3203700628-fvdgs 0/1 ContainerCreating 0 24m
romana-agent-5mjjn 1/1 Running 2 24m
romana-agent-ao7y7 0/1 CrashLoopBackOff 9 24m
romana-agent-dgth5 0/1 CrashLoopBackOff 9 24m
romana-services-icdve 2/2 Running 2 24m

from halcyon-kubernetes.

Great, thanks. So romana-services is running, one agent is working and two failing to start.
I'll see if I can reproduce that.

from halcyon-kubernetes.

@wilkers-steve question...are you working off the PR from earlier, or what is in master? fair amount of activity today. i need to update the submodule for halcyon-vagrant-kubernetes if that's what you're using.

from halcyon-kubernetes.

@v1k0d3n I updated the submodule for my forked halcyon-vagrant-kubernetes to pull from my forked version of halcyon-kubernetes to test the changes I've made. Looking to see if it's something I may have inadvertently caused as well with some of my changes. Currently running again

from halcyon-kubernetes.

@wilkers-steve thanks

from halcyon-kubernetes.

The problem is from removing the public_iface override for Vagrant.
It was important to make sure that traffic between VMs is routed via the host-only interface instead of the default NAT interface.

from halcyon-kubernetes.

Okay. It's definitely part of the change I submitted today. Sorry for the false alarm. Going to close this one out.

from halcyon-kubernetes.

@cgilmour Yeah, thats what I just figured out. That was due to some confusion on my end. Going to fix this now.

from halcyon-kubernetes.

@wilkers-steve i reverted the changes to h-vagrant. do you want to try again, now that the public_interface is more clearly understood? i meant to accept the changes to h-kube for the Centos changes...I didn't realize I was merging changes to the Vagrantfile.

from halcyon-kubernetes.

@v1k0d3n Yep. Working on it now. Just testing the new changes before I submit the next PR

from halcyon-kubernetes.