Fusion Server 4.2.1 Release NotesSubscribe

Release date: 5 April 2019

Component versions:

Solr 7.5

ZooKeeper 3.4.13

Spark 2.3.2

Jetty 9.4.12.v20180830

Ignite 2.6.0

New features

No new features were introduced in this release.

Improvements

  • The Users page (System > Access Control > Users) now displays the security realm to which each user is assigned.

  • SharePoint and SharePoint Online connector improvements:

    • The SharePoint and SharePoint Online connectors now delete indexed documents after they are deleted from the SharePoint source.

    • The SharePoint connector now retries document fetches after network errors.

  • Addressed an issue with fetching and indexing of Sharepoint list item attachments.

  • Addressed an issue with running the Sharepoint connector without an associated parser configuration.

  • Addressed an incremental crawl scenario for deleted sites.

    • Confluence connector improvements:

  • Fixed an issue with the handling of Confluence 5.6 archive API endpoint.

  • Added security trimming support for the Confluence connector "confluence-administrators" group.

  • Fixed an issue with incremental ACL updates for changes with _lw_confluence_anonymous permission setting.

  • The HTML parser now parses documents whose Mime-Type value does not exactly match text/html, such as text/html; charset=utf-8 and so on. This allows it to parse .aspx pages from SharePoint datasources, for example.

  • The Windows Share connector now retrieves document owner metadata.

  • For connectors that use the Crawl database type/crawlDBType parameter, the default value is now on-disk instead of in-memory in order to support larger crawls by default.

  • Addressed an issue with persisting proxy settings during Javascript evaluation in the Web connector.

  • Addressed an incremental update issue due to unnecessary serialization of the ID field in the MongoDB connector.

  • Updated the Zendesk connector to use RequestEntityProcessing.BUFFERED instead of .CHUNKED to avoid a documented Jersey issue.

  • DevOps Center improvements:

    • In the Log Viewer, exported data is now constrained by all selected filters in addition to the time range selector.

    • The time range selector in the Log Viewer now behaves more consistently.

    • Fixed some issues with filtering in the Hosts tab and the Log Viewer.

    • In the Log Viewer, log messages whose level is DEBUG are now distinguished by an icon for consistency with other log levels.

  • The size of the system_logs collection was significantly reduced by changing the log level for marking user.sessions meter to DEBUG.

  • You can now use the query pipeline to set the status code for Fusion’s response.

  • SQL service improvements:

    • Fusion’s SQL service now supports non-phrase predicates.

    • Results can now be sorted by the score of the underlying Solr query.

    • Random sampling is now supported.

    • The service now provides more consistent performant results from unlimited selects.

    • Fixed an issue in 4.2.0 which cause unlimited "group by" aggregations to use a bucket size limit of 1.

  • Fusion now captures the same metrics about the connectors-rpc services as it does about the connectors-classic service.

  • A variety of minor UI issues were fixed.

Other changes

  • FS (V2) connector saves Item metadata in Crawldb - recrawls work as expected (the new/delete/update items will be picked up) - On migrations, it is suggested to clear the datasource and start with a new crawl, but it is not required.

  • Now when you upload a .war file to the App Studio interface, the View Published UI button remains available, resolving a known issue in 4.2.0.

  • Logging out using Chrome version 73.0.3683.75 no longer causes an infinite loop; this resolves a known issue in 4.2.0.

  • You can now re-install a connector using the same ID but a different filename than the original.

  • Several bug fixes resolve issues with ACL metadata using the Confluence connector.

  • Better permissions checks when importing jobs and blobs.

  • Exporting an app with deep=false now exports only the app definition without the linked objects.

  • Fixed an issue which sometimes prevented "chained" jobs from firing, that is, jobs that are triggered by the success or failure of another job.

  • Fixed an issue that prevented users from disabling the Asynchronous Execution Config/asyncConfig option in some pipeline stages.

  • In the Solr Query pipeline stage, configuring one or more requestHandlers now blocks usage of the select request handler. To configure usage of multiple request handlers including select, add select to your requestHandlers.

  • Fixed an issue which prevented the parsing of paths containing backslashes in fusion.properties on Windows.

  • Fixed an issue when using the default init/systemd/install.sh systemd service which cause all Fusion services to fail when one service failed. To avoid this issue, use /opt/fusion/4.2.1/init/systemd/install-services.sh instead; see the instructions at fusion/4.2.1/packaging/init/systemd/README.

  • Fixed an issue which caused intermittent Zendesk connector failures with a ConnectionClosedException: Premature end of chunk coded message body: closing chunk expected message.

Known issues

  • When under load, the Fusion proxy service can occasionally become stuck, causing user authentication to fail. This is the result of the proxy InputStream failing to close properly.

    An upgrade to Fusion 4.2.4 is required to fix this issue. See Upgrade Fusion.

  • The Zendesk connector will fail occasionally when fetching content, due to a known HTTP client library issue.

  • While indexing a SharePoint load test environment, some documents may be skipped due to an immense term error caused by a lengthy value in the signature_s field.

  • The Spark Driver may run on restricted API nodes despite its access being forbidden by setting the DrunSparkDriver field to False.

  • Jobs in the Query Workbench’s Recommender do not include associated Recommendation collections.

  • The Query Workbench’s Text Tagger is not working with the spell correction collection as expected.

  • Attempting to import the SharePoint connector object results in a validation error.

  • When using the Tika parser with "Add original document content (raw bytes)" enabled, the resulting Base64 string contains too many bytes. This causes distortion of the data.

  • The v2 connector reduces a multivalued field to a single value.

  • The Google Drive Connector is not permitting users with the appropriate permissions to access data.

  • While using the Google Drive Connector, duplicate document records are being produced for each user that has access to the folder and its items.

  • User access to group rules are not being appropriately restricted by the values in the rules-<group name> field.

  • When searching the DevOps center for messages by a specific logger class, a partial class name does not return results. The full class name must be used.

    For example, searching for "JmxServiceInstanceMetricReporter" will not return results. Users must search the full class name, "apollo.metrics.JmxServiceInstanceMetricReporter".

  • Basic select SQL queries on {collection}_signals, {collection}_query_rewrite, {collection}_query_rewrite_staging fail unless an ORDER BY clause is included.

  • Spark jobs are not appearing in Job History until after the job is completed.

  • The error.log and output.log files are being written to the var/api/work path instead of var/log/api.

  • When building a new UI in Fusion App Studio, the Last Published Date field contains a date and time value, despite the UI being unpublished. This value does not change when the new UI is published.

    This is only known to affect users of Windows 10.

  • If a connector is deleted while the connectors-rpc service is stopped, the connectors-rpc service will fail to start.

  • Setting the Maximum Output Count value does not stop the crawl process from continuing.

  • The Job History tab in the Scheduler will occasionally list the start and end time as identical values despite having a duration greater than 0 seconds.

  • Recently created jobs are not immediately available in the Scheduler.

  • When Minimum Matching (mm) is applied to a synonym expansion, both parts of the synonym pair are required to produce query results.

  • The status indicator for services listed in the Cluster dashboard may not update properly.

    This is only known to affect users of Windows 10 and Internet Explorer 11.

  • The number of collections displayed in the Cluster dashboard may not reflect the actual number of collections.

    This is only known to affect users of Windows 10 and Internet Explorer 11.

  • In some cases, recently updated items in the Service Logs table will not appear.

  • For Windows 10 users, the Access Logs section of the DevOps Center does not display all request types, including GET, PUT, etc. Additionally, values may be missing from some columns in the log table.

  • Non-administrator users are being granted the ability to delete index and query pipelines, despite not having the appropriate permissions.

  • While creating a new rule in Fusion UI, some options may no appear due to the width restrictions of the page.