LTS 2023.8 / LTS 2023-HF08

What's New in LTS 2023.8 / LTS 2023-HF08

Increase garbageCollectOrphanBlobs Action Bucket Size

Tune blob full GC to improve S3 scroll throughput

Add Query Limit Parameter to Full GC Management Rest API Endpoint

You can now limit the number of blobs to be garbage collected on the Full GC Rest API

Add the queryLimit=10000 parameter to the Rest API call to invoke the Full GC on the first 10000 blobs only:

curl -X DELETE -u Administrator:Administrator http://localhost:8080/nuxeo/api/v1/management/blobs/orphaned?dryRun=true&queryLimit=10000

Full Blob GC in dryRun Should Trace Samples of Blobs to Remove

A new "nuxeo.bulk.action.garbageCollectOrphanBlobs.sample.modulo" property to trace samples of blobs to remove

While running a Full Blob GC in dryRun mode, with the following property:

nuxeo.bulk.action.garbageCollectOrphanBlobs.sample.modulo=1000

every 1000th blob deletion will be logged at WARN level with such message:

dryRun sample: GC would have deleted blob: b967e77cce9e0582af118bfb467fbab9 of size 12 bytes.

You can assert the blob can effectively be deleted in MongoDB with such a query:

db.default.count({ecm:blobKeys: b967e77cce9e0582af118bfb467fbab9})

should return 0. Assuming, the blob was referenced by the file:content document property, such a MongoDB query

db.default.count({content.data: b967e77cce9e0582af118bfb467fbab9})

should return 0.

Assuming the the file:content is used as default document content, if you want to assert that the ecm:blobKeys migration went through without mistake, the MongoDB query:

db.default.count({ecm:blobKeys: {: false}, content.data: {: true}})

should return 0.

Replace Poi-Ooxml-Schemas by Poi-Ooxml-Lite in the Main Pom Dependency Management

Maven dependency management now handles poi-ooxml-lite instead of poi-ooxm-schemas.

Avoid Publishing Empty Timer Metrics to Datadog

You can reduce the number of Datadog metrics by discarding empty timers

Release and Publish DB Packages

Non-default DB packages are now released.

Use Better Default for Bulk bucketSize to Reduce Record Processing Duration

Tune bulk actions to avoid long record processing

Fix Garbage Collection When Default Blob Provider Blob Keys Can Be Both Un/Prefixed

Document Blob Garbage Collection always check for prefixed and unprefixed default provider blob keys.

Update Aws Java SDK to 1.12.650

AWS java sdk upgraded to v1.12.650, ion-java-1.0.2.jar is no longer installed by amazon-s3-online-storage marketplace

FileNotFoundException in Command Line Conversions

Tmp files are now deleted when appropriate. Tmp containers will remain empty.

Provide Different Levels of Metrics Exposition

You can select a minimal filter to reduce the number of metrics

Enable Large Work to Be Serialized by Default

Large works are now serialized by default with warnings at serialization.

Fix SAML Authentication Plugin When Several SAML Authentication Plugins Are Configured in Authentication Chain

SAML auth providers can now be chained.

Provide a Nuxeo Stream Log4j Appender

There is now a log4j appender for Nuxeo Stream

Implement Azure Optimized Copy

Azure Blob provider now suports the optimized copy

To add the optimized copy between 2 Azure blob providers, the Azure Java SDK client has been bumped from v8 to v12 which is a significant change in terms of Azure APIs. The Nuxeo Azure Blob Provider remains the same.

Handle Exceptions in AbstractSession#removeChildren

When a child document can not be removed, the reason is now log-warned.

Throw an Error if a String Is Used to Query a Long/Integer Field

The "nuxeo.primitive.type.strict.validation" nuxeo.conf property allows to throw an error when decoding a string input as a number (integer, long, float, double).

Avoid Retries on Scroll That Has Already Downstream Records

Scrolls with downstream records are not retried anymore.

Avoid NPE on ACL Without Name

Nuxeo doesn't throw NPE on ACL without name anymore.

Learn More

More information about released changes and fixed bugs is available in our bug tracking tool.