What's Changed
- Bump Kubernetes controller to v0.0.4 by @eliecharra in #50
- Fixed a race condition when deleting a worker. This could cause a worker to show as IDLE, but be unable to pick up new runs.
- Fixed an issue where workers were not being marked as ready after a run failed to start (for example an initialization policy failure, or not being able to create the run Pod). This would prevent them picking up new runs.
- Fixed an issue where workers were not correctly cleaned up when deleted if the MQTT connection was lost (for example if the pool was deleted in Spacelift, or if there was a network error). This could cause infinite reconnection attempts to show up in the logs.
- Fixed an issue where the controller would incorrectly try to register a worker with Spacelift after it had been deleted.
Full Changelog: v0.15.0...v0.16.0