Repository navigation
A standby manager that takes the lease no longer exits when a project was deleted while it waited.
What's new
- The manager stays up when a standby takes the lease. A standby keeps a pending watch for each project. When a project was deleted before the standby led, multicluster-runtime v0.25.2 started that watch on a stopped cache at the takeover, and the manager exited with code 1. While no manager was ready, DNSRecordSet writes were refused. dns now requires the
release-0.25commit of multicluster-runtime that skips such a watch (kubernetes-sigs/multicluster-runtime#205), and moves to v0.25.3 when it is released. (#536) - CI:
datum-cloud/actionsmoved to v1.28.0. (#535)
Note
No API, image name or bundle name changes since v0.16.0. An upgrade from v0.15.0 still needs the new names that v0.16.0 introduced.
Full changelog: v0.16.0...v0.16.1