classifier

Fixes:

• Add support for Lua script in Classifier
• Fix event-manager RBAC (permission to create/update/delete secrets)

• Fix sveltos manager core
• Fix event manager handling when oneForEvent field is set to false

🚀 Features

1. event driven add-on deployments: define an event and what add-ons need to deployed when event is detected. Add-ons can be expressed as template. When an event is detected in managed cluster, Sveltos will instantiate (if needed) and deploy add-ons;
2. cluster health check: define an health check passing Sveltos a Lua script. Sveltos will keep watching resources in a managed clusters and keeps evaluating the cluster health. When cluster health changes, Sveltos sends a notification (Webex and/or Slack message and/or a Kubernetes event).

🚀 Features

• Techsupport: sveltos can collect tech support (both logs and resources) from managed clusters
• Add support to send webex notifications

🚀 Features
• Introduce cluster liveness checks. Ability to receive a notifications (either a Kubernetes event of type reason=ClusterHealthCheck or a slack notification) when Kubernetes add-ons are provisioned in a cluster.

🐛 Bugs
• Fix a bug that prevented add-ons to be provisioned if Cluster (with proper labels) was created after ClusterProfile

🚀 Features

• ConfigMaps/Secrets referenced by ClusterProfile and RoleRequest can be set with an empty namespace. When namespace is empty, at the time of deployment, Sveltos will find ConfigMap/Secret in the cluster namespace.

⚠️ Breaking Changes

• Secret referenced by ClusterProfile and RoleRequest must be of type addons.projectsveltos.io/cluster-profile

🚀 Features

• multi-tenancy

🌱 Others

• bump cluster-api pin to v1.3.3

🚀 Features

• register existing, non clusterAPI, clusters;
• configuration drift detection

🌱 Others

• bump cluster-api pin to v1.3.1

🐛Bug fixes

• fix a classifier crash when cluster had no labels;
• fetch Kubernetes version by using k8s.io/client-go/discovery instead of relying on node labels

Values sources for HelmRelease or manifest sources for Kubernetes resources, can be treated as templates. Their values can be fetched from management cluster run-time state. Reading any value from following resources is supported:

1. Cluster
2. InfrastructureProvider: this is the Cluster infrastructure provider
3. KubeadmControlPlane: this is the Cluster Kubeadm controlPlaneRef

1. Introduce cluster classification. Using the Classifier CRD, it is possible to define which set of labels clusters should have based on their runtime state (Kubernetes version and/or deployed resources);
2. Introduce ability to change log severity for each Sveltos pod without restarting the pod. DebuggingConfiguration CRD allows to set that at per sveltos component level;
3. Added knob to select behavior when cluster stops matching a ClusterProfile. Default behavior causes all kubernetes add-ons deployed by the ClusterProfile instance to be withdrawn from cluster. Knob allows changing this behavior and leaving add-ons in the cluster.

• bump cluster-api pin to v1.2.5