Cloud CMS Release 3.2.92

The following are the new features, enhancements and fixed issues for Cloud CMS 3.2.92 release.

Upgrade Considerations

Please review the upgrade considerations provided here:
https://gitana.io/documentation/gitana/3.2/guide/guide/docker/upgrades/3-2-92.html

With this upgrade:

• MongoDB 4.4 or above is required.
• You will need to perform a cold restart of your API cluster.

The following are the new features, enhancements and fixed issues for Cloud CMS 3.2.92 release.

Enhancements

Binary Storage Providers

• Binary Storage Providers now support Presigned URLs so as to enable faster downloads.
• Archive Controller now utilizes Presigned URLs for downloads of archives when utilizing S3 as a binary storage provider.
• Binary Storage Providers now have their own log4j2 logging configuration, letting you capture logs for every operation against a provider (for auditing and data governance assertion). Enabled using the org.gitana.platform.services.binary.storage.provider.log=true setting - see documentation for more details.

Cache

• The default cache sizes for many of the Hazelcast distributed caches have been increased substantially so as to be optimal for larger content repositories and tenants out-of-the-box. These are adjustable via properties files but the defaults are now more sensible for larger deployments.
• Faster cache predicates for cache invalidation that utilize an optimal Trie data structure.

Copy Project

• When a project is copied and its datastore titles were initially derived from the original project title, the copied datastores will now similarly have their titles updatd to reflect the new project title.

Dependencies

• Updated to the latest versions of dependencies to address all critical and most high, medium and low severity issues identified in CVE and vulnerability scans.

Deployments

• Deployment calls now allow for a transient deployment strategy configuration object to be passed in directly. This allows for a simpler deployment case where a separate, persisted deployment strategy is not needed.

Find

• Find operations now use ES search_after instead of skip for improved performance during iterated lookups.

Job Dispatcher

• Various improvements to make job dispatcher faster and leaner, using less memory and distributing job execution around the cluster in a more even manner.

Logging

• Improved logging for job dispatchers, node policies, cluster membership, transformations, job sweeper and locking services.

Maintenance Thread

• Added a Cluster Maintenance thread that auto-manages the cluster and performs tasks in the background (such as cleaning up defective projects).

Policies

• When generating preview images, the “after” policies are suspended so as to avoid unintended triggers of rules or other downstream behaviors.

Releases

• The user interface now prevents a release from being submitted a second time if a release job was already scheduled on a previous click (includes UI elements so that editors/publishers are aware that the release was submitted and is in-flight).
• The API side now checks if a release already has a release job in queue at time of submit and throws an exception if so.

Transfer

• Faster transfer import that now uses concurrent threads to write binaries to attachable objects.

User Interface

• SSO now supports a param option which allows for a JWT token to be found in the request param.
• SSO now performs an exhaustive and forgiving lookup of JWT token across headers, cookies and request parameters.
• Added a “Deploy To” action which allows for simpler deployment to a target (not requiring a strategy)

WebHooks

• Added a bulk web hook job which now collects a large swath of web hook jobs (up to 500 at a time) and bulk executes them for faster web hook job throughput.

Bug Fixes

Binary Storage Providers

• Fix so that binaries import works properly for long nested keys for all binary storage providers.

Copy Project

• Fix so that if forceURIEncoding isn’t specified on source project, the importer now assumes this to be false to ensure backward compatibility with older projects.

Create Project

• Fix so that a failed create project operation cleans up properly (removing any data stores, stacks or other elements so that nothing is left partially created).

Jobs

• Fix so that the stopped flag is only applied on RUNNING and ERROR state job transitions the first time.
• Fix so that the stopped and stopped_timestamp job attributes are cleared from teh job when the RUNNING phase starts (preventing the awaiting listener from errantly killing the job if it deems it to be in a stalled state).
• Fix so that parent and child dependencies are more accurately tracked for WAITING and AWAITING jobs.

Job Sweeper

• Fix to the Job Sweeper thread so that it isn’t overly aggressive in terms of cleaning up long-running jobs.
• Fix for several bugs whereby the Job Sweeper could incorrectly determine the available server keys and deem a job to be no longer running on a cluster member.

Merge

• Fix so that branch merges fire the correct number of policy invocations on a per-node basis. This avoids policies being triggered multiple times. The result for customers was frequently in seeing rules execute multiple times (such as web hooks).

Scheduled Work

• Fix so two or more scheduled work threads cannot attempt to schedule a common job at the same time. This covers for an edge case where multiple servers may host live scheduled work threads that dispatch the same set of jobs.

Startup

• Fix so that concurrent branch initializations which are run in parallel correctly synchronize on an inner map so as to avoid concurrent modifications during iteration.

User Interface

• Fix so that page and component refreshes work more consistently.
• Fix so that modals for adding a user to a team and adding a user to a project are consistent.
• Fix to Content Type Builder so that children being added are scoped to the correct draggable.
• Fix so that reset password link detects incoming request protocol correctly.
• Fix so that a blocking modal is no longer present when adding/removing a favorite.

Web Hooks

• Fix so that platformId is correctly assigned to all web hook jobs (even when processing via an action).