CloudBees CI on traditional platforms 2.426.1.2

The transport used by the Remote Directory API was changed from Remoting to HTTP to provide compatibility with the High Availability (HA) controllers. This can be turned off using -Dcom.cloudbees.opscenter.client.plugin.RemoteDirectoryServiceImpl.useHttp=false in case of a compatibility issue.

A newly scheduled build is now, in most cases, directed toward the replica with the least load (according to a simple metric that counts builds already running, queue items already scheduled, and online agents). This applies to interactive triggers (Build Now), scheduled builds, branch indexing, webhooks (multibranch events), REST API triggers, and event triggered builds. The downstream builds scheduled by the build or triggerRemoteJob steps continue to run on the same replica as the upstream build.

The High Availability (HA) configuration screen now has an option to set a maximum aggregate load for the controller (number of builds, agents, and queue items across all replicas). If the maximum aggregate load is exceeded, the new builds are rejected to avoid overloading and destabilizing the system.

Asynchronous messaging relies on the remoting channel that exists between the operations center and the controllers. When you run a High Availability (HA) controller, only one replica has a live remoting channel, so asynchronous messaging did not work between other replicas and the operations center or other controllers.

After a Pipeline build that uses the regular durability mode is completed, its metadata storage is now compacted for faster access from tools, such as the CloudBees Pipeline Explorer. This also reduces the disk space and inode usage.

The Move/Copy/Promote operations were not working in the High Availability (HA) controller. This issue was resolved and you can now Move/Copy/Promote the Pipelines to and from the HA controller.

This fix provides compatibility with the publishEvent step when executed from within a High Availability (HA) controller.

High Availability (HA) controllers now have the triggerRemoteJob step. This allows you to use the Pipeline build step to trigger a build on another replica.

Certain API endpoints that return a list of builds, such as /job/name/api/json?tree=builds[number], give incomplete information in a High Availability (HA) controller. Information about builds that run on a replica other than the one processing the request are now automatically inserted into the response.

With the removal of prototype.js from Jenkins core, usages of this library were removed from the following CloudBees Assurance Program (CAP) plugins:

We can now define different sources for bundles with the same repository name from different organizations. In addition, the whole URL is used to perform the checkout, allowing the same repository and organization in a different host.

The bundle update log now displays the bundle version on the Load Configuration as Code bundles screen for the Operations center.

Timestamps now display on the bundle status page and bundle update log of the CloudBees Configuration as Code bundles page.

The following feature enhancements were added to this release:

The Maven plugin has been removed from CAP. The existing administrative monitor that informs users about this deprecation has been modified and now informs users about the removal and provides a link to the documentation that describes how to migrate jobs based on the Maven plugin.

The following issues have been resolved:

Due to the deprecation of the SHA1 algorithm in RHEL 9, we have moved to a more secure algorithm for our RPM signing process.

When several checkout requests arrived before the checkout request in progress finished, the queue was blocked and timed out.

Now, whenever an update is made to Bundle Update Timing configuration, change is synchronized across replicas.

The SSH connections to SCM were broken in the last CasC Bundle Retriever update. This issue is resolved.

The order of JVM arguments was corrected to allow -Djava.io.tmpdir to be overwritten with JAVA_ARGS in /etc/default/cloudbees-core-cm. Previously, the user-specified argument was overwritten by the product default value.

If a Pipeline build that runs the triggerRemoteJob step waits for a downstream build event (normally completion) when the upstream controller restarts, and that event occurred when the upstream controller was restarting, the step can hang indefinitely (or fails with a timeout error) when it should have continued.

Accessing the job REST API, such as https://jenkins/job/my%20project/api/json?tree=builds, can return an error when the job name contains spaces or other special characters.

When many queue items were scheduled in a High Availability (HA) controller, the load balancing can result in momentary anomalous statuses from REST APIs. Quiet periods were also not consistently honored.

Every use of an API token to make a REST request produced a warning in the system log of a High Availability (HA) controller. This warning is now suppressed.

A High Availability (HA) controller automatically will proxy the REST API requests to look up (or cancel) queue items for the replica that owns the item. If the registered replica is unresponsive, for example, because it just terminated, the originating replica now automatically retries the registration lookup and will proxy every second up to ten times.

Under certain race conditions, when a build that runs on a High Availability (HA) controller was adopted because the owning replica terminated, an inbound agent in use by that build can fail to reconnect (giving an “Unknown client name” error) ultimately hitting a 5m timeout. Some background processes are now suspended after a replica begins shutdown that may have been responsible for the cluster losing track of the proper owner of the agent.

SnakeYAML API has been upgraded from 1.33 to 2.2.

Upgraded org.json:json from 20230227 to 20231013.

Removed the call to the hetero-list YUI button that was removed from Jenkins core.

Synchronizing the next build number of jobs can be a bottleneck in HA controllers on slow network filesystems. This process is now improved and the issue has been resolved.

Previously, High Availability (HA) controller replicas could partially load build records even when the builds were in progress on other controllers, leading to errors. Now, a replica only loads a build record when the build is complete or when it is adopting a build in progress.

If you schedule a build using a High Availability (HA) controller, but it does not start before the replica shuts down, the corresponding adopted queue item in another replica now uses the same ID. This allows external tools to track the build when it does start (or the item is cancelled).

If a Pipeline Template Catalog is configured in the CasC jenkins.yaml file and the id property is not defined, the catalog is duplicated on each instance restart and in the exported CasC configuration.

When the name of an existing cloud node is updated, the user receives a 404 error after selecting save. This is because the cloud page uses the cloud name as part of its URL. When the user saves the name, Jenkins sends the user to the URL with the old cloud name. Please note that all node changes are saved successfully.

The following plugins have been added to the Operations center CloudBees Assurance Program since 2.414.3.7: