Upgrade Notes
- Operations center CloudBees Assurance Program plugin changes since 2.462.2.2
-
The following plugins have been added to the Operations center CloudBees Assurance Program since 2.462.2.2:
-
Apache HttpComponents Client 5.x API Plugin (
apache-httpcomponents-client-5-api
)
-
- Controller CloudBees Assurance Program plugin changes since 2.462.2.2
-
The following plugins have been added to the Controller CloudBees Assurance Program since 2.462.2.2:
-
Apache HttpComponents Client 5.x API Plugin (
apache-httpcomponents-client-5-api
)
-
New Features
- Kubernetes 1.29 and Red Hat OpenShift 4.16 support
-
CloudBees CI on modern cloud platforms now supports Kubernetes 1.29 on Azure Kubernetes Service, Amazon Elastic Kubernetes Service, Google Kubernetes Engine and Red Hat OpenShift 4.16. For more information refer to, Supported platforms for CloudBees CI on modern cloud platforms.
Feature Enhancements
- Removed support for incremental releases
-
CloudBees CI used to deliver incremental releases when fixes impacted only a subset of plugins.
CloudBees CI now only delivers full releases. This allows for a more consistent experience across the various ways of consuming the CloudBees CI product and being better aligned with best practices for infrastructure as code and Configuration as Code.
This removes support for incremental releases from our product that solves several issues we observed while upgrading our product.
- Support requests made configurable
-
The support request link is now configurable. The new field is an "override" only. If it is set, it is used unconditionally. If the new field is not set, then the following occurs:
-
If the user is an administrator, use
https://cloudbees.com/r/support/request/new/ci
. -
If the user is not an administrator, nothing is shown.
-
The configuration only controls the override. If it is deleted or left empty, it resets to using https://cloudbees.com/r/support/request/new/ci
for administrators.
- Administrative monitor warning displays when agents run outdated software
-
An administrative monitor warning displays if the following occurs:
-
A permanent agent is currently connected
-
A cloud agent has recently connected using an old version of
agent.jar
-
A cloud agent has recently connected using a version of Java that is older than the controller version
This also works for the High Availability (HA) controllers.
-
- UI was adapted to use new Jenkins standards
-
Several parts of the UI have been adapted to use the new Jenkins UI guidelines that are based on Bootstrap.
Resolved Issues
- CloudBees SCM Reporting leaked memory of Pipeline builds and associated metadata
-
Previously, CloudBees SCM Reporting inadvertently retained references to all Pipeline builds that were processed by the plugin, leaking heap memory unnecessarily.
- CyberArk Global Configuration failures when saving
-
The CyberArk Global Configuration failed to save due to a non-serializable
cache
field. Consequently, changes made to the CyberArk global configuration did not persist and were lost on restart.
- CloudBees Pipeline Explorer incorrectly cached localized status descriptions
-
The build status descriptions, such as
success
,failed
,unstable
, in the CloudBees Pipeline Explorer UI previously supported localization. However, their values were controlled by the first user to access the CloudBees Pipeline Explorer after CloudBees CI started and they did not stop caching. This was confusing if the first user to access the CloudBees Pipeline Explorer had a language preference that did not match other CloudBees CI users. These descriptions are no longer localized.
- Controlled agents were not supported in High Availability (HA) controllers
-
In an High Availability (HA) controller, the list of security tokens for a controlled agent was not synchronized to other running replicas, and the agent would not behave correctly until after a rolling restart. The list of security tokens also was not copied to clones when using the multi-executor option. Now, both issues are resolved and the controlled agents can be used normally in High Availability (HA) controllers, with or without the multi-executor option.
- Pipeline was not able to resume builds in some cases
-
In some cases, the Pipeline resumption will time out after one hour to prevent builds that are indefinitely stuck.
- Make Pipeline cleanup more robust in some edge cases
-
Prevent all
NullPointerException
inCpsFlowExecution.cleanUpGlobalClassValue
to make the Pipeline cleanup more robust.
- Notify listeners when the last RBAC group is deleted from a folder, job, or node to avoid caching outdated permission checks
-
Notify listeners when the last RBAC group is deleted from a folder, job, or node. This prevents incorrect permissions from being cached in some cases and ensures that the removal is replicated correctly when using High Availability (HA).
- Mismatch issues between local and remote executors
-
There are some display mismatches that appeared on the High Availability (HA) controller when viewing executors from both the current replica and another replica. The display is now more consistent.
- Administrative monitors could be inconsistent between replicas of a High Availability (HA) controller
-
Some administrative monitors applicable to a High Availability (HA) controller might have been accidentally displayed if certain configuration changes were made using browser sessions on other replicas.
- Prepare the High Availability (HA) controller for future Hazelcast version upgrades
-
Typically, the High Availability (HA) controllers support rolling upgrades between the CloudBees CI versions. This includes adding Hazelcast 5.3.x patch releases that do not change the wire format. To prepare for an upgrade to Hazelcast 5.5 in a future release that does not permit rolling upgrades, the controller now tracks the Hazelcast (minor) version being run in the cluster and automatically switches to a non-rolling upgrade (similar to a restart of a non-High Availability (HA) controller) if it detects that an upgrade is occurring to a new Hazelcast release.
- Mute noisy log
-
Muted some logs that were previously categorized as warnings, such as: "No remote context available. Falling back to default value." when running a standalone controller. These logs should have been considered detailed and should not have been displayed to the system administrator.
- The controller name was not anonymized in Kubernetes Provisioning support components
-
The Support options
Managed Controllers on Kubernetes descriptors
andManaged Controllers on Kubernetes logs
generated files that had anonymized contents but the name was not, and they revealed controller items names. The file names are now anonymized.
- Cryptic error messages in the managed controller log
-
When a managed controller was configured as code without a Kubernetes endpoint or a container image, cryptic error messages could be printed in the controller log. These are now suppressed earlier, and any other similar errors will include a full stack trace for easier diagnosis.
Known Issues
- Authentication to operations center Configuration as Code Retriever API fails
-
After upgrading to version 2.462.1.3 or later, authentication to the operations center Configuration as Code Retriever endpoint API at
${OC_URL}/casc-retriever/*
fails. This is due to the removal of the authentication method enabled by the attribute.OperationsCenter.CasC.Retriever.secrets.adminPassword
.
- Duplicate Plugins in Operations center Plugin Manager UI
-
When searching for a specific plugin under 'Available' tab in operations center Plugin Manager, the search results shows duplicate entries of the searched plugin.