Announcements

Help us improve the Equinix Community by completing a quick survey

Announcements

Cluster-API-Provider-Packet v0.7.0 Release

cprivitere
Equinix Employee

Version 0.7.0 of cluster-api-provider-packet, introduces metro level support vs facility level in accordance with the soon-to-be updated Metal API. Users of previous versions will want to take advantage of this immediately. The basic requirements to upgrade your existing clusters can be found here. Please work with your Equinix support team to determine the best migration path for your architecture. Assistance can also be found in the Community Slack and Community site. Read more at our Metros Quick Reference and see the facility deprecation announcement.  

*Please note that if devices are already in the correct metros you’ve specified, no disruption will happen to clusters or their devices. As with any production change, test your changes before applying them to clusters in production. 

In addition to metro-level support, this release installs the latest cloud-provider-equinix-metal v3.6.1 by default and is built on golang version 1.19 and cluster-api 1.3. The default OS used is Ubuntu 20.04 and kube-vip is updated to v0.5.12 in the kube-vip flavor templates.  

Lastly, the CI workflow has been refactored with caching removed and updated tests considering cluster-api 1.3 dependencies. 

See more at the github release here! 

https://github.com/kubernetes-sigs/cluster-api-provider-packet/releases/tag/v0.7.0

1 Comment
achantarotwong
Equinix Product Manager

CAPP v0.7.2 released - Fixes install and new node issue 

achantarotwong_0-1685724348743.png Release v0.7.2 · kubernetes-sigs/cluster-api-provider-packet  

We've released v0.7.2 to comply with the new kubernetes package repositories new key signing process. This new process prevented cluster-api installs from working at all. Please update to v0.7.2 so you can deploy new nodes again. 

Included is of course v0.7.1  which switched the default repo to be achantarotwong_1-1685724348744.png Quay to avoid docker rate limiting and fix an issue where v0.7.0 used ghcr which wasn't public to begin with.