Fleet Management
Fixed
- Set domain name for the Kubernetes APIs server address when logging in to CAPI provider workload clusters.
Fixed
- Fixed common name in certificates generated for workload clusters by stripping https:// prefix from cluster base path
Fixed
- Fix nil pointer panic in
template nodepool command.
Changed
- Make all
kubectl config flags (e.g --context and --kubeconfig global and unify kubeconfig management throughout commands. - Remove CAPA templating from
aws provider. - Add new provider
capa for templating a cluster. - Remove fetching ssh sso ca pub key for capa from management cluster.
- Add test for CAPA provider cluster templating.
Changed
- Upgrade vertical-pod-autoscaler to 0.11.0
Potentially breaking change:
- Added validation - CPU values will be accepted only with resolution of 1 mCPU, memory with resolution of 1 b
Other changes:
- Switch to go 1.16
- Admission controller now logs when it fails to start
- Increase resolution of admission_latency_seconds metric
- Reduce verbosity of some logs
Added
- Add
cilium.giantswarm.io/pod-cidr annotation.
Changed
- Change default webhook timeout to 5 seconds
- Defaulting deployment strategy to
Recreate for the controller
Fixed
Fixed
- Added the possibility to override the deployment strategy for the controller
Fixed
- Added
projected to the list of permitted volumes in the psp-csi-controller.
