github.com/senthilrch/kube-fledged

Features and enhancements:

• Feature: Migrate kube-fledged ci to github actions #92
• Feature: end to end test framework for kube-fledged #93
• Feature: Add option to leave image pull/delete Jobs running #101
• Feature: priorityClass support #120
• Permission issue when purging images from cache on OpenShift #134
• Add README.md to kube-fledged helm chart #136
• Add "Blogs and Presentations" section to README #138
• Add Shout-outs section to README #139
• Update Contributing guidelines document #144
• Push images to dockerhub when commit is pushed to develop #152
• Don't run pull_request job when PR is accepted #154
• Update shout-outs section with new links #162
• Bump up versions of alpine, golang & operatorsdk versions #163
• Added jobPriorityClassName arg #173
• use dockerhub busybox image #181
• Package controller and webhook server together #186

Bug fixes

• Fix failing go tests #135
• Image pull jobs failing: "standard_init_linux.go:228: exec user process caused: no such file or directory" #176
• Error deleting image when defining containerd.sock path #188

Features and enhancements:

• Feature: Allow ImageCache CR to be created in a different namespace; not restricted to the one to which kube-fledged was deployed #86
• Feature: cacheSpec with nodeSelector matching no nodes #91
• Feature: Add support for custom serviceAccount in Jobs to support k8s environments with PSP restrictions #100
• Feature: kubefledged-webhook-server as optional component #126
• Enhancement: Improve error reporting when image pull job expires #98
• Enhancement: failed calling webhook #115
• Enhancement: Webhook is not accessible on EKS with Calico #123
• Enhancement: Remove helm hook from validatingwebhookconfiguration #127

Bug fixes

• Bug: x509 error re-installing ImageCache #113
• Bug: Issue with installing via Helm #114
• Bug: Controller occasionally crashes with index out of range #124

Note: The previous version v0.8.2 helm chart would create a validatingwebhookconfiguration resource in the namespace. This would get created via helm pre-install hook. When you execute helm delete or perform helm upgrade to v0.9.0, helm won't remove the validatingwebhookconfiguration resource since it was created via pre-install hook. Before installing v0.9.0 helm chart, delete the validatingwebhookconfiguration resource in the namespace.

Features and Enhancements:-

• Issue #94 Enhancement: update design proposal document
• Issue #99 Document parameters exposed by helm chart in values.yaml
• Issue #109 Expose helm parameters in operator's custom resource
• Issue #112 make deploy-using-operator: deploy operator, controller and webhook-server to same namespace

Bug Fixes:-

• Issue #95 Helm installation ignores namespace
• Issue #102 Operator stops working in interval after error
• Issue #105 Misleading log entry during auto refresh of ImageCache
• Issue #111 Controller gets stuck due to "Error from updatePendingImageWorkResults(): no pods matched job

Features and Enhancements:-

• Issue #79 Feature Request - Have a yaml only (no bash) way to deploy
• Issue #80 Enhancement: Use self-signed certs for webhook server
• Issue #82 Enhancement: Use environment variable for specifying cri-client-image
• Issue #83 Enhancement: Use "kubefledged.io" as domain for annotation keys
• Issue #84 Enhancement: When deploying via helm operator, wait for operator ready status
• Issue #85 Enhancement: Install validatingwebhookconfiguration as helm pre-install hook
• Issue #87 Enhancement request: avoid hardcode busybox init container
• Issue #88 Enhancement: modify helm operator crd group to "charts.helm.kubefledged.io" & update openapiv3schema
• Issue #90 Enhancement: use busybox image from gcr.io to overcome dockerhub ratelimiting

Bug Fixes:-

• Issue #75 Deployment using the Operator 0.8.0
• Issue #76 Deployment via Helm chart fail
• Issue #77 Incompatibility with carvel-kapp: object has been modified
• Issue #89 Bug: Unable to delete image on node with containerd runtime

New Features and Bug fixes:-

• Support for quick installation using helm chart
• Helm charts published to self-hosted helm repo (https://senthilrch.github.io/kubefledged-charts)
• Support for signing and verifying helm charts using PGP
• Imagecache API group modified to kubefledged.io due to kubernetes/enhancements#1111
• Imagecache API version upgraded to v1alpha2 (no support for v1alpha1)
• Helm Operator Kubefledged API version upgraded to v1alpha2
• Wait for webhook-server to be ready before starting controller container
• Create v1/v1beta1 CertificateSigningRequest based on cluster capability (kubectl api-versions)
• Added labels to kube-fledged manifests for easy filtering of resources
• Upgraded client-go to latest/stable (v0.21.1)
• Upgraded operator-sdk to latest/stable (v1.7.2)
• Upgraded versions of go, alpine, docker, cri-tool, go modules to latest/stable
• Updated unit tests due to upgraded client-go package
• CI improvement: build out all four amd64 images during push build.
• Fix for issue #63 Getting x509: certificate signed by unknown authority Error
• Fix for issue #66 CRD Resource API version deprecated
• Fix for issue #70 Controller fails to pick up ImageCache resource on fresh install if webhook server is down

CI Improvements & Bug fixes:-

• Improve CI efficiency:
  • Perform tests during "push" job
  • Build multi-arch images during "pr" job
• Issue #59 Images with latest tag not published for arm & arm64
• Issue #60 Multi-arch image for cri-client
• Issue #61 Coverage error during Travis build restart

New Features and Bug fixes:-

• New Webhook server developed and added as a separate deployment
• Support for mTLS in webhook server using cert/key pairs signed by api-server
• Implemented image cache spec validation during create and update
• YAML Manifests, Helm chart and operator updated for webhook server
• Implemented multi-arch images using buildkit (docker buildx and qemu)
• Support for additional platforms: linux/arm and linux/arm64
• Optimize image manager performance during cache creation, update and refresh
• Prevent creating redundant jobs for re-pulling images already present in the node
• If image name has no or "latest" tag, such images will be always pulled during refresh
• Implemented Conditional Multi-stage builds in Travis CI to reduce time
• Release images will be pushed with both release_version and latest tag
• New make target "deploy-using-operator" to install operator and kube-fledged in separate namespaces
• Two new make targets to uninstall kube-fledged
• Added documentation comments to image cache yaml
• Versions of docker, crictl, golang, alpine and operator-sdk bumped
• Fix for Issue #55 and Issue #56

New Features and Bug fixes:-

• Helm Chart for installing kube-fledged
• Helm Operator for deploying kube-fledged (using operator-sdk)
• Support for deploying Operator and kube-fledged in any namespace
• Docker multi-stage build for fledged container image
• New Makefile targets "make operator-image" and "make release"
• Parameterized golang and alpine image tags in Dockerfiles
• Several improvement to builds and ci
• README updated for quick installation using helm operator
• TOC and logo added
• Versions of docker, golang, alpine bumped
• Issue #52 : Allow kube-fledged to be deployed in any namespace

New Features and Bug fixes:-

• Added support for cri-o runtime
• Added support for on-demand refresh of image cache
• Added new makefile target for removing kube-fledged (make remove)
• Added support for Quick Installation using pre-built images
• Added OpenAPIv3 Schema validation to CRD
• Code refactored to use inclusterconfig from client-go
• Bugfix: Do not auto-refresh purged image caches
• Bugfix: Panic when node not in ready status
• README updated: Support for cri-o runtime
• README updated: Added steps for uninstalling kube-fledged & refreshing image cache
• README updated: Added badges for release & license