Fusion Server 4.2.4 Release Notes

Release date: 3 September 2019

Component versions:

Solr 7.7.2

ZooKeeper 3.4.13

Spark 2.3.2

Jetty 9.4.19.v20190610

Ignite 2.6.0

New features

SharePoint Optimized and SharePoint Online Optimized Connectors

Fusion 4.2.4 introduces ground-up rewrite of our our SharePoint and SharePoint Online connectors. The new connectors, SharePoint V1 Optimized and SharePoint Online V1 Optimized offer a variety of new features and benefits, including:

  • Performance of the initial crawl and incremental crawls have been significantly improved.

  • Security trimming has been centralized so it can be shared between different connector configurations. This is handled by the Active Directory Connector for ACLs.

    SharePoint Optimized Security Trimming

  • More options are available for limiting documents.

    SharePoint Optimized Limit Docs

  • More options are available for customizing crawl performance.

    SharePoint Optimized Crawl Performance

Users are encouraged to upgrade from the classic connectors to the optimized connectors.

Active Directory Connector for ACLs

The new Active Directory Connector for ACLs indexes Active Control List (ACL) information into a configured collection so that it can be used by the SharePoint Optimized and SharePoint Online Optimized connectors.

Active Directory Connector for ACLs

Forked Apache Tika Parser

Fusion Server 4.2.4 adds a new parser, the Forked Apache Tika parser. This parser is a simplified Apache Tika parser geared towards an Enterprise Search crawl in which all documents are parsed in a forked Java Virtual Machine. If memory issues occur while parsing, such as an Out Of Memory condition, the connector job is unaffected.

Forked Tika Parser

Auto-scaling Policies

Fusion now supports search, analytics, and system auto-scaling policies for application and system collections. When creating a collection in Solr, Fusion enables autoscaling by passing a policy parameter to apply an auto-scaling policy to the collection.

This functionality uses Solr’s auto-scaling policies.

Improvements

  • The Rules UI now supports bulk actions, including Add Tags, Enable, Disable, and Delete.

    Typeahead Feature
  • Fields in the Rules UI now support typeahead.

    Typeahead Feature
  • In Rules, the Field Value filter queries, fq, are no longer required to be a perfect match. For example, the following filter queries return the same result:

    • fq=foo:bar

    • fq=foo:("bar")

    • fq={!tag=foo}foo:bar

  • When deleting a rule from a list of rules, you remain on the current page instead of being sent to page 1.

  • CrawlDB operations now let MapDB store information in multiple files instead of just one. This enables a user to delete an individual table by deleting the corresponding MapDB file.

    This change affects V1 connectors that support bulk import operations. Other connectors are unaffected by this change.

  • The Confluence connector now uses prefetch logic to significantly improve crawl performance.

    Enable prefetch

  • The Confluence connector has a new property, CONFLUENCE API REQUEST PAGE SIZE, which specifies the number of records per response page for REST API calls to Confluence.

    Confluence API Request Page Size

  • Minor improvements are made to the Jobs panel.

  • Minor improvements are made to the Scheduler panel.

Other changes

  • Jersey has been updated to version 2.29.

  • Fusion now saves the facet rule minCount value as expected.

  • The facet fields in Rules > Business Rules now display correctly.

  • Fixed a bug that stopped recommender jobs from detecting associated recommendation collections.

  • Fixed a memory leak in the Security Trimming stage of the Alfresco connector.

  • Updated the Boost List and Bury List rules so they don’t conflict if both are triggered for the same query.

  • Fixed a Head/Tail Analysis Job UI bug that prevented a job from being saved if the KEYWORD BLOB NAME was modified before being cleared.

  • Fixed a UI bug in Rules that left the Publish process active after changes were made to a deactivated rule.

  • Fixed an empty dialogue box which appeared while importing an app after starting over.

  • Fixed a bug that prevents App Insights and App Studio from accessing authentication tokens or session tokens if Fusion is running with Trusted HTTP enabled.

  • Fixed a bug in Rules > Business Rules that prevented users from sorting rules by date if only rules with future dates were present.

  • Fixed a bug that prevented machine learning models from being recreated after being deleted from the blob store.

  • Unnecessary items are now deleted from the CrawlDB database as expected after the crawl is completed. This change affects V2 connectors only.

  • Each connector job now remains dedicated to the node on which it originally ran.

  • Access control items are now indexed in batches instead of one at a time in V2 connectors.

  • The reliability of the Proxy service has been improved.

  • The HTML parser can now parse more media types, including plain text.