leap: k8s 1.25, node sharing setup, no more pre-pulling, no maintenance notice #2337
Conversation
This was an optimization for a workshop where I assume nodes were pre-started and made sense for that, but would slow down startup on new nodes if not this specific image was requested on a non-pre-warmed node after the workshop. Since the workshop has ended, we can safely remove this now. Related tickets: - Workshop: https://2i2c.freshdesk.com/a/tickets/349 - Slow startup discussed: https://2i2c.freshdesk.com/a/tickets/435
|
Merging this PR will trigger the following deployment actions. Support and Staging deployments
Production deployments
|
consideRatio
force-pushed
the
pr/leap-maintenance-node-sharing
branch
from
2bb8c7b
to
1e88bec
Compare
There was a problem hiding this comment.
I don't see an explanation in here or #2209 about removing the pre-puller can you just explain why that's part of this change (or link me to the appropriate bug). Is there any impact to community startup time with that removed?
Everything else looks fine and I'll approve once you've responded
|
Ah I forgot to surface this in the PR description, there is this git commit message about it:
|
|
🎉🎉🎉🎉 Monitor the deployment of the hubs here 👉 https://github.com/2i2c-org/infrastructure/actions/runs/4405036988 |
Already deployed
This is already deployed and the terraform changes are applied together with the profileList changes that are tightly coupled with them. Due to that, it is important that this is merged before another PR makes a redeploy of the leap hub by accident with an old profileList.