2019 Release Notes
Security Fix for discovered CVE
A security vulnerability has been found and corrected in 3.15.0. For details, please see CVE-2019-7238.
Sonatype recommends that administrators running NXRM3 versions up to and including 3.14.0 upgrade immediately.
Repository Manager 3.18.1
Sonatype is pleased to announce the immediate availability of Nexus Repository 3.18.1. The issues fixed in this release can be found below.
Fixed an exception occurring when a user manually logs out of the user interface.
Propagates the quarantine status code when NXRM proxies another NXRM with Firewall enabled.
Repository Manager 3.18.0
Sonatype is pleased to announce the immediate availability of Nexus Repository 3.18.0. A summary of the highlights in this release is shown below. See the complete release notes for all resolved issues.
Note: the permission model for browsing has been changed to behave similarly to NXRM2, see NEXUS-20453 for details.
Fixes two cross-site scripting (XSS) vulnerabilities.
Improves browse performance particularly when the user has a large number of content selectors.
Database Backup & Upgrade Speed
Optimized compression level and buffer size defaults of Orient database backups, resulting in faster upgrade speeds.
Default Memory Configuration
Increased the default maximum heap and direct memory sizes for Nexus Repository Manager 3. Users will now receive increased sizing profiles and enhanced performance without having to manually change default configurations with new version upgrades.
Repository Manager 3.17.0
Sonatype is pleased to announce the immediate availability of Nexus Repository 3.17.0. A summary of the highlights in this release is shown below.
See the complete release notes for all resolved issues.
Default admin password
You will find that the default password for the admin user is now randomly generated on startup of a new NXRM instance; see the Post Install Checklist for more details.
HA-C Upgrade Changes
The contents of the
$install-dir/etc/fabric/hazelcast-network-default.xml file have changed - you must update your own
$data-dir/etc/fabric/hazelcast-network.xml to include the new section.
Also, if you're using HA-C TCP/IP discovery an additional configuration property is needed in
$data-dir/etc/nexus.properties. See the HA TCP/IP Discovery documentation for details.
New and Noteworthy
For users with strict security controls we've added a feature we've called Routing Rules which can be used to prevent requests from being sent to external proxy repositories. This can for example allow users to block requests containing codenames from leaking to external repositories.
Apt Format Support
As part of our widespread format support we've begun to move one of the plugins from our great community members which supports the apt format into the core product. (Limited to single nodes in this release.)
Go Format Support
Supporting the burgeoning Go community we've added native support for go repositories. (Limited to single nodes in this release.)
Beta Provisioning REST API
We've continued to add to our REST API with new endpoints to support operations on users, and user tokens. In this release they are flagged as beta, please take a look at our API documentation in the administration part of the user interface and submit feedback.
New instances as well as instances which have not changed anonymous configuration will prompt administrators upon logging in to confirm that anonymous users should be allowed to access the system. Similarly if the admin user is using the default password then upon logging in they will be required to change the password.
Removed Recent Connections Accesslog Database
The "accesslog" database which tracks recent unique connections has been removed, along with the Recent Connections section of the administration interface. This removal in no way impacts the request.log text file.
Reduced Logging for Missing Task Listeners
Unnecessary logging of missing task listeners has been reduced.
Repository Manager 3.16.2
Sonatype is pleased to announce the immediate availability of Nexus Repository 3.16.2.
The only fix in this release is for users of high availability.
Log entries for errors of "Task Already Running"
Users of high availability may encounter logged errors that a task failed to start as it was already running.
Repository Manager 3.16.1
Sonatype is pleased to announce the immediate availability of Nexus Repository 3.16.1 . The resolved issues are shown below.
After installing license (switching from OSS to PRO only, will not affect installing license in existing PRO instance) application doesn't start
Upgrading to 3.16.0 PRO can fail when usertoken database contains entries with duplicate (case-insensitive) usernames
Repository Manager 3.16.0
A bug which affects installing a license and moving from Nexus Repository OSS to Nexus Repository Pro has been discovered in version 3.16.0. This issue does not affect existing Nexus Repository Pro installations which are being upgraded to version 3.16.0 and has been fixed in 3.16.1.
Sonatype is pleased to announce the immediate availability of Nexus Repository 3.16.0. A summary of the highlights in this release is shown below.
See the complete release notes for all resolved issues.
New and Noteworthy
REST Node Status Endpoint
If you are running a cluster, or otherwise monitoring node health, you need a simple check to determine if your node should receive requests. To meet that need, we have added a new REST endpoint, /v1/status/writable. This endpoint returns a 200 status code when the node is ready to process requests. It's perfect for integration with load balancers to remove a node from the active set when it becomes unhealthy and return it after the node has been recovered. See the documentation Status API for more.
Support Zip Improvements
Getting all of the necessary support zips can be arduous in a HA-C environment. Administrators no longer have to access each node to generate support zips. Generating support zips will have the option to generate them on all nodes instead, and the pop up window will have a tab per node.
Previously, gathering support zips was a GUI driven process. So administrators could not use scripts to automate gathering support zips. The new REST endpoint, /v1/support/supportzip, gives programmatic access to support zips. The documentation for this API is here: Support API
HA-C topologies have whole new classes of metrics that are beneficial to gain insight into your cluster. A new log file, nexus_cluster.log, has been added to contain them. This log contains information about the cluster's health, such as internode latency and replication queue depth. Support zips will include this log to make introspection on the history of cluster health easier.
Additional System Status Checks for Professional HA Clusters
System Status Checks are a great way for components of NXRM to report vital information to administrators. To that end, the checks have been added to the banner. In normal operation, every check is passing and administrators will see a green checkmark. However, if at least one status check is failing then the header will have a red exclamation point.
Three new status checks were added specifically targeting high availability health.
- Cluster Size is a new status check that ensures that your cluster has 3 nodes. If a cluster is missing a node, it can still respond to requests and save new components, but it is at risk of losing one more node and becoming inoperable.
- Transactions status check monitors an instance for any transaction with excessive retries in the last hour. Excess transactions can indicate things like heavy cluster load, inter-node network issue, or something more malignant.
- Lifecycle Phase is a status check that verifies that the node is running the proper lifecycle phase. Running in the wrong phase can occur due to startup issues, improper use of the lifecycle API, etc.
When a node becomes unhealthy, we now have a REST API to control the lifecycle phase of a NXRM node. A lifecycle phase is a step in the start up process used to group similar components and ensure that their dependencies are started before them. The REST API documentation has more information at Lifecycle API.
The Maintenace API helps NXRM administrators understand their databases' status and recover from some error cases. Each endpoint can optionally take a node Id determining on which node the request is executed. See the REST API documentation for more insights at Maintenance API.
REST Search Improvements
REST search now supports sorting results, as well as filtering prerelease components, allowing the search and download endpoint to return the latest version of an asset. More information in a community post.
Update packaged JRE in installers to OpenJDK
Mac and Windows installers now include the Zulu OpenJDK 8 JRE.
Improve stability of using HA-C with Nexus Firewall
Using Nexus Repository Manager in HA-C with Nexus Firewall could cause the cluster to have numerous concurrent modification exceptions. There were a number of changes made to both reduce the frequency of conflicts and more robustly handle them when they do occur.
Bower Resolver Updated
In addition to 3.16.0, we have also updated the bower resolver dependancies to the latest and most secure versions. We strongly recommend any users using bower update their bower resolver to the latest.
Metrics Timers Identifiers Have Changed
Some customers may be parsing the /service/metrics/data REST resource or utilizing the info/metrics.json file in support zips. This release updates our metrics library, which has changed the names of timers to include the suffix '.timer'. For example:
The "version" field of the metrics file reflects this change, reving the version from 3.0.0 to 4.0.0.
Audit Database Replaced With Log File
The audit database at $data-dir/db/audit is no longer in use and is replaced with audit log files written and archived in $data-dir/log/audit . Administrators may remove the audit database files from disk at their own discretion.
- [NEXUS-18905] Cleanup tasks fail with "No search context found for id" error
- [NEXUS-18509] Make JEXL and CSEL behave the same for expressions without a leading slash
Content Selectors, Scripting, Upgrade
- [NEXUS-17850] API does not validate contents of content selectors. Invalid content selectors can lead to failed upgrade.
Database, Scheduled Tasks
- [NEXUS-18983] If NXRM is read-only or lacks quorum, then run now triggers make startup fail.
- [NEXUS-18816] Slow delete performance when using REST API
- [NEXUS-19125] Docker pull from mcr.microsoft.com results in 403
- [NEXUS-19229] LastDownloadedHandler conflicts in HA can cause long-running retries/rollbacks
LDAP, User Token, Upgrade
- [NEXUS-13639] User tokens not migrated if LDAP user ID case does not match login case
LDAP, Tree View, Logging
- [NEXUS-17616] On browse w/ LDAP, if no perms, a bunch of warns are fired
- [NEXUS-18731] specifics about what is deleted by cleanup policies is not logged
Proxy Repository, Logging
- [NEXUS-17502] Content validation message does not log which repository the invalid content is coming from
- [NEXUS-16853] Enhance content validation for maven-metadata.xml files
Maven, Scheduled Tasks
- [NEXUS-19034] NPE on "Purge unused snapshots" task
- [NEXUS-17908] Tag association may intermittently fail for new artifact
- [NEXUS-12222] NXRM2 repository view privileges are not migrated to NXRM3 browse privileges during upgrade
- [NEXUS-18774] allow scoped NPM package name parts that start with '.' or '_'
- [NEXUS-17896] concurrent requests for large npm metadata can lead to OutOfMemory during serialization performance
- [NEXUS-19121] Delete of component or asset from PyPi proxy repository fails
- [NEXUS-19118] Clicking on the links in a repository for component/asset browse gives a 404
Repository Health Check
- [NEXUS-18950] "Download trend' disabled text misleading
- [NEXUS-14407] REST Search & Download by 'Latest'
- [NEXUS-19085] staging promotion move of more than 500 components may fail with IllegalStateException Unable to find component by id performance
- [NEXUS-13057] "Uncaught TypeError" UI errors
- [NEXUS-19051] Edge and Internet Explorer errors in the user interface
- [NEXUS-12253] Order by version incorrect
- [NEXUS-18277] UI upload creates temporary files in java.io.tmpdir
- [NEXUS-16057] Add UI upload for Yum
- [NEXUS-17884] upload of source rpm fails in yum hosted
- [NEXUS-17920] Deleting an rpm via DELETE to /repository does not update metadata
Repository Manager 3.15.2
Sonatype is pleased to announce the immediate availability of Nexus Repository 3.15.2. A summary of the highlights in this release is shown below.
Fix search filtering using repository query parameter, when group repository is entered (REST and UI).
Our official Docker image switches to OpenJDK as our recomended Java Runtime Environment. It's worth checking whether the difference has impact on your deployment (e.g., contains a different system trust store that may affect your SSL connections).
Repository Manager 3.15.1
Sonatype is pleased to announce the immediate availability of Nexus Repository 3.15.1. A summary of the highlights in this release is shown below.
Fix anti-cross-site request forgery token mismatch blocking valid npm client publish and login.
Repository Manager 3.15.0
Sonatype is pleased to announce the immediate availability of Nexus Repository 3.15.0. A summary of the highlights in this release is shown below.
See the complete release notes for all resolved issues.
We have found an issue in this release that affects NPM users, as well as using the REST search filtering by repository group names. We recommend NPM and/or REST search users upgrade immediately to 3.15.2-01 (see Download Archives for access) or newer.
New and Noteworthy
Beta REST API Endpoints Removed
The Nexus Repo REST API moved from beta to release status in the 3.13.10 release. The old /service/rest/beta endpoints were kept for compatibility in the 3.13.0 and 3.14.0 releases, but they have now been removed (NEXUS-18562). Update any scripts that use /service/rest/beta endpoints to use the new /service/rest/v1 endpoints.
For our Professional licensed customers, we added a new level of flexibility for administrators when planning and updating their blob stores. Multiple blob stores can be combined into a Blob Store Group. These groups allow an individual repository to use storage across multiple locations and devices. Member blob stores can be added or removed from a group to ease the burden of migrations. Also added Fill Policies which give administrators more control and insight into how components are being stored.
For all our users (OSS and PRO) we have added Soft Quota which allow you to receive a warning when your blobstore has reached a configured metric.
For more detail, see our documentation Storage Guide.
OpenJDK 8 Runtime
Users have been reporting OpenJDK has worked fine as the JVM for repository manager 3. As of 3.15.0, we now run our full test suite using a compliant Java SE standard distribution of OpenJDK 8. In light of the Oracle licensing changes to their Oracle JDK distributions, you can now be sure repository manager is fully tested against the standard.
UI Upload left files in temporary directory after upload was complete
This bug would cause temp files to be left behind when the UI Upload feature was used. This could cause eventual disk space issues that would not be cleaned up by the standard NXRM cleanup tasks. This ticket fixes the issue going forward but does not assume to delete anything from the file system. If you utilize Upload UI frequently, we recommend checking the temp directory to see if you can get some disk space back.
Use HTTPS for outreach Base URL
In order to promote security, we changed our Welcome Outreach capability to use HTTPS instead of HTTP.
Analytics plugin deprecated
Due to complications, the Analytics plugin and respective pages within the application have been removed from the system. If it is restored, we will have a new feature release note.
- [NEXUS-18252] Repository manager will not start on blobstore problems or errors
- [NEXUS-15095] MissingBlobException can occur when publishing Maven index
- [NEXUS-18196] ArrayIndexOutOfBoundsException when uploading large POM
- [NEXUS-12684] HEAD request to /v2/<name>/manifests/<reference> results in 404 error
- [NEXUS-18263] Docker proxy repositories configured with a remote URL including extra path info will not proxy correctly
- [NEXUS-18353] Can't proxy older gcr.io Docker images
- [NEXUS-18100] LDAP UI does not reflect configuration
- [NEXUS-13626] Made Privileges box wider so Privileges fit
- [NEXUS-7996] Changing dropdown does not show users list until refresh
- [NEXUS-18564] Delete orphaned API keys task run before any other HTTP activity can stop some LDAP operations
- [NEXUS-16964] Upgrade from 2.14.8 to 3.10.0 may prevent download of rubygems hosted repository gems
- [NEXUS-17501] Caching of NuGet metadata causes thread serialization, query slowdowns under load performance
- [NEXUS-14465] PyPI hosted repository does not send etag header
- [NEXUS-17903] PyCharm does not work with PyPI repository
- [NEXUS-18187] PyPI proxy of does not work
- [NEXUS-16401] PyPi hosted repository packages can only be searched by pep-0503 normalized name
- [NEXUS-18345] Running a "/search/assets" REST API call with just a "repository" query parameter does not give full results
- [NEXUS-12828] Submitting more than 20 tasks at once causes ERROR for some tasks
Content Selectors, Tree View
- [NEXUS-15085] Tree View is slow when there are large numbers of content selector privileges
- [NEXUS-16384] Maven SNAPSHOT timestamp versioned files are not direct children to the base snapshot version in tree or html view
- [NEXUS-14682] Support deleting all assets under entire selected tree nodes
- [NEXUS-15179] Deleting last pom leaves folder shell
- [NEXUS-18119] New role not available for use until role page revisited
- [NEXUS-12100] Delete button on component and asset is active even without permission to do so
- [NEXUS-18276] UI upload leaves files in temporary directory after upload is complete
- [NEXUS-18494] UI upload fails if it takes more than 60 seconds
- [NEXUS-18299] JsonSyntaxException attempting to create Webhook capability
- [NEXUS-18261] Hosted yum metadata not rebuilding due to parsing issues in the path being queried
Search, Tree View
- [NEXUS-18617] Disable asset download count feature in all new/upgraded installations
- [NEXUS-17803] Allow search by g:a:v string
[NEXUS-18631] Allow multipart copy for AWS S3 blob storage