Originally taken from https://2i2c.slack.com/archives/C07SJJWVCAD/p1773321694805809

I've mentioned this in passing in a couple of github comments, but I wanted to surface it in a high-priority place.

When investigating ways to perform a zero disruption migration with the new ingress, I was using 2i2c-aws-us. After a lot of experimentation, Helm is complaining when I migrate onto nginx-ingress if I don't disable leader election. For some reason, Helm things that it should be patching an existing lease rather than creating it from scratch. This lease doesn't exist in the k8s resource list, and I can therefore only assume Helm is referring to a previous release's manifest.

However, I tried to clear this by making a successful nginx-ingress deployment without leader election, reverting it with a new deploy, and then deploying a final time with leader electrion ... and it still throws the error. So this warrants a bit more investigation.

I don't think anyone will run into this — redeploys of support on 2i2c-aws-us are successful. But it might be a landmine in future unless we can resolve it, so I'm raising it now. I'll try to figure out what's wrong soon, but it's not a high priority for me today! (edited)