etcd service start timeout about puppetlabs-kubernetes HOT 6 CLOSED

Probably it's a Debian specific thing, I had the same on all three controllers hanging at

Notice: /Stage[main]/Kubernetes::Service/Service[etcd]/ensure: ensure changed 'stopped' to 'running'
Info: /Stage[main]/Kubernetes::Service/Service[etcd]: Unscheduling refresh on Service[etcd]

In my case the hostnames are correct.
But etcd had a lot of timeouts in the log (I have to say, that my test cluster is a bit low powered on resources). So I had tls handshake errors due clock difference more than 1 seconds between the peers and also i/o timeouts.

So I think the reason for hanging could differ. But as long as etcd is not fully up, the puppet run hangs.
As soon as I had improved the performance and my clock's where running in a green range, the puppet run continued because etcd started successfully.

systemd 232-25+deb9u4

from puppetlabs-kubernetes.

@khaefeli can you fill in the rest of the issue template please

from puppetlabs-kubernetes.

@scotty-c updated.

from puppetlabs-kubernetes.

@khaefeli I have just tested this with both Ubuntu and RHEL and did not see this issue. The service failed to start instantly due to not being able to connect to the initial cluster. what version of systemd are you running?

from puppetlabs-kubernetes.

we encountered the same issue with local kitchen test setup. only if all controller nodes are provisioned the same time in parallel the run continues. otherwise if one node is not setup it hangs forever

from puppetlabs-kubernetes.

Hello @blodone,
Closing this issue as it has been from a very old version of the module. If the problem persists please feel free to re-open and add more recent information.
Thanks.

from puppetlabs-kubernetes.