Mario Manno
Mario Manno
During an upgrade the fleet controller might hot-reload and act on a config that is from a future release. This might prevent bugs like https://github.com/rancher/fleet/issues/3501 Fleet controller should only reload...
Users want to create one helm values secret on the upstream cluster, then re-use it on downstream clusters. The values in the secret are used to populate the chart's templates....
We use the structured logger instead. Notes: * `logrus.Fatal` -> `log.Fatal` * `setupLog = ctrl.Log.WithName("setup")` in controllers * `logrus.Warn` -> `logger.Error` * `logrus.Debug` -> `logger.V(1).Info` * ...
Update the experimental feature with information from the RFC: - [x] use a new secret type for the ociRegistry secret. - [ ] ociRegistry secret contains URL and connection options,...
A deployment with replica count 1 can take a long time to migrate to another node. We should make the replica count configurable. ### Acceptance Criteria * [x] Fleet controllers...
When using https://fleet.rancher.io/gitrepo-add#use-different-helm-credentials-for-each-path one set of credentials can be given per path. However, users might have multiple (30) bundles in a gitrepo which pull from the same registry, e.g. the...
The introduction of events seems to have removed some log lines. We also need more metrics to troubleshoot large fleet deployments with a huge number of gitrepos. - [ ]...
Agents are in different namespaces * cattle-fleet-local-system * cattle-fleet-system They wait for bundledeployments in different cluster namespaces and use different kubeconfigs. Ideally, their registration resources should not collide and non-namespaced...
Fleet should support PCs and PDBs. They should taken from the existing `AgentDeploymentCustomization` struct. see "RFC: Cattle Cluster Agent Priority Class And Pod Disruption Budgets"