WillPlatnick

@mgmeyers - If you're willing to prioritize this, I'll donate $100 to the cause.

@alex, do you have any thoughts on this?

Just realized that error is coming from the worker nodes, duh. I can see that I can't run it from kubectl on the nodes themselves either, it has the same...

OK, so the drain is failing because the new version generated new certificates and kubeconfig for workers, but it doesn't put them in place before it tries to do the...

OK, further issue I think is that I removed the password: field from my config file because the new template didn't show that it was needed any longer. I believe...

To get around this, I added the password field back to my config, decremented my /etc/kismatic-version to 1.9.0 on my masters, and ran the upgrade process again.

This would be nice. I like seeing the bigger context of the task.

Readiness probes are still useful even if they're not used in a service. Specifically, when you do a deployment of workers and want to make sure your deployment was successful,...

We ended up ripping celery inspect ping out from our liveness probes because we found that under heavier load, the ping would just hang for minutes at a time even...

@mcyprian We got rid of the liveness probe. My gut is telling me it has something to do with eventlet, but we haven't made it a priority to figure it...