Error: 'tls: oversized record received with length 20527' about dcos-kubernetes-quickstart HOT 3 CLOSED

For anyone else experiencing this.. the heapster issue is not tied to the TLS. After running:

kubectl proxy
On top of the ssh tunnel they describe in the DC/OS Connecting Clients page, the url:

localhost:8001/ui
Does load up the dashboard correctly, however heapster graphs are still broken.

Note that the original url works fine as well and kubectl proxy isn't really needed on top of the SSH tunnel.

Looking at the kubernetes-dashboard logs, there are issues connecting to heapster (likely timing of the scheduler install).

This illustrates the issue:

Creating in-cluster Heapster client
Could not enable metric client: Health check failed: an error on the server ("[-]healthz failed: could not get the latest data batch\nhealthz check failed") has prevented the request from succeeding (get services heapster). Continuing.

I'm new to k8s, and it appears there is no "restart" proper... but that is what is needed... so after digging through a few threads, it seemed killing it was the solution, and k8s would auto-restart it. So...

kubectl delete pod --namespace=kube-system kubernetes-dashboard-blahblah-xyz
(where blahblah is the uuid for your pod)

Checking the logs when it comes back, it connects to heapster cleanly:

Creating in-cluster Heapster client
Successful initial request to heapster

And the graphs are now in the dashboard.

Probably as good as it is going to get for this beta ;)

from dcos-kubernetes-quickstart.

This will be simplified as soon as we remove the need for the SSH tunnel - which will require DC/OS 1.11

from dcos-kubernetes-quickstart.

Thank you for reporting @seanhig. Our next release will replace Heapster with metrics-server and you should stop seeing this issue.

from dcos-kubernetes-quickstart.