New features

Expanded detailed “pod” view

To improve the detailed view of pods, the Kubernetes agent view screen now displays the entire contents of the pod resource configuration YAML file.

Controller provisioning settings in team controller creation recipes

Allows a user to specify the CPU, RAM Memory, JVM Heap ratio (runtime data area) and disk space allocated when creating a team controller. The user assigns these parameters to a particular team controller creation recipe (defined using the CLI tool) and selects the recipe when using the team creation wizard.

Provides faster and less error-prone onboarding of new teams by replacing several manual configurations with a simplified team controller configuration selection. This feature enables a team to get started with CI/CD faster.

(Manageability) CLI command to perform backups

Automates backups of controller or controllers with the CLI, rather than triggering the backup with CloudBees Jenkins Operations Center GUI (Cluster Operations).

Provides an additional option for backing up controllers. The new option uses the CLI for the CloudBees Backup Plugin; the existing method of triggering the backup through the CloudBees Core Operations Center GUI is preserved.

Resolved issues

Extraneous “=” in cje.yml causing UI problems

Removed an extraneous “=” which was hiding in a regular expression.

Kubernetes agents not connecting to controller after many attempts

Fix for agents unable to connect to controller.

Pod retention defaults to 'on error', causing scheduling issues

Change default Kubernetes pod retention so that error state pods are removed.

Default pod retention is 'on error' causing lots of errored pods to remain and cause scheduling problems

CloudBees Core is incompatible with Configuration as Code plugin

In the Jenkins Configuration as Code plugin, ensure that the plugin follows plugin compatibility requirements.

Stopping a controller can fail with errors about PATCH invalid field

In the Kubernetes plugin, made PatchUtils thread-safe.

Jenkins processes being killed by OOMkiller

In Operations Center, changed the memory configuration in the Kubernetes YAML configuration file to prevent memory overusage.

UnsupportedOperationException errors in Kubernetes pods

Upon creating lots of agents in parallel (Cloud provisioning containers), the act of moving temporary files to node/config.xml could cause multiple UnsupportedOperationException errors.

Addressed by modifying Pod agent error serialization.

Controller provisioning fails due to an invalid spec.selector

Upgrade the Master Provisioning Plugin to version 2.2.4. For background on this issue, see the article controller provisioning fails due to an invalid spec.selector.

Long team names are truncated in UI

The “team switcher” was limiting the number of characters displayed. The team switcher now displays the full name of the team (up to 40 characters, the supported limit) when hovered over.

Known issues

None

Revisions

Revision 3 (2020-04-14)

Plugin updates

Revision 2 (2018-11-22)

Plugin updates