Change Log - Version 1.1.0 - Release Date October 8, 2015

This release includes a wide range of improvements and fixes, including significant performance enhancements, additional documentation, and changes for operational support. We recommend that all customers upgrade to this release.

The upgrade process is automated, and will preserve all data, but will require some time offline. You may also need to alter your SSL configuration, if you configured it for 1.0. Please see the *Upgrading* section of the documentation before beginning the upgrade process.

There are also incompatibilities for customers who customized the portal but directly referenced the installed "preemptive" queries or transformations. Please see below for details.

Performance Improvements:

  • Query performance improvements for the portal.
  • Improved the responsiveness of the application filter in the portal.
  • Added a configurable query timeout setting in Portal and increased the default timeout.
  • Performance improvements for message ingestion.
  • Scalability improvements for processing extremely long-running sessions.
  • Scalability improvements for large number of unique feature names within a session.
  • Temp-bin cleanup ("garbage collection") runs more often and uses fewer system resources.


  • Portal user interface:
    • Added a new `Featured Metric` Widget and examples of it on the Overview report.
    • Table Widget now uses progressive scrolling. (This deprecates the `show_count` configuration option.)
    • Table widget sorting has been simplified to improve usability.
    • Chart tooltip improvements and other minor UI tweaks.
    • Exceptions report now shows information about exception types.
    • Improvements to report printing.
  • Data processing & customization:
    • Extensive expansion and clarification of the documentation.
    • Large values (>300 characters) are now supported for feature names.
    • Exception indexer now records the exception type and exposes it via the exception query.
    • Usability improvements to the query API.
    • Added support for data sent by new features of the JavaScript instrumentation API.
    • The new/returning user filter is now configurable to improve accuracy for very large user bases.
    • Significantly increased accuracy of unique users calculation.
    • When applying a Custom Data Filter as a Pattern, key values are propagated to the parent Scope.
    • Added Application Run Scope as a parent of Session Scope.
    • For JSON envelopes, the endpoint will now use the originating IP (rather than the immediate upstream) to determine the user name hash.
    • The Query API streams the results back to the clients as they are collected instead of waiting for the entire result set.
    • Improved unique user identification for JavaScript messages.
  • Administrative:
    • Reliability improvements when interacting with MongoDB and RabbitMQ.
    • Additional WMI counters to enable better monitoring.
    • Error queue is now automatically processed periodically.
    • Improvements to documentation for custom data filters.
    • Improvements to the installation process including upgrade support.
    • Admin tool now supports authentication for the interaction service.


  • Administrative:
    • The Endpoint and Portal (with Query Web Service) are now installed on different ports by default so that one can be re-configured and/or restarted without affecting the other.
    • The required MongoDB version is now 2.6.10+ (up from 2.6.1+) (MongoDB 3.x is not supported.
    • IIS websites no longer go idle after 20 minutes.
    • Batch size for error queue processing is now configurable.
    • Data Hub is now required to be installed before installing Workbench.
  • Data processing & customization:
    • Significant changes have been made to Portal components included with the installation, in the `preemptive` namespace. User-defined components may have relied on `preemptive` components and therefore may be affected by these changes; please update these user-defined components as necessary. Please see the "Upgrading" page in the documentation for details.
    • When a Custom Data Filter is configured as a Pattern, the `CustomCount` field will not be added to other queries, because it was not semantically meaningful for those queries.
    • Standardized and documented the AllowCopyToChildren setting.
  • Portal user interface:
    • Feature table now shows counts for features even if there is no associated session.
    • The default value for the `dontSetIfAlreadyThere` flag has been changed to `true`. Please see the `Advanced Topics` page in the documentation for details.


  • Data processing & customization:
    • Fixed bug where an error while writing to MongoDB could corrupt temp bin documents.
    • Fixed bug where negative counts and session lengths were reported by the KeyStats query due to incomplete sessions.
    • Fixed bug where aggregating by month or week did not return fully aggregated data.
    • Fixed bug where OS and Runtime aggregation returned different results depending on message processing order.
    • Fixed bug where feature ticks were not counted correctly when they were received in a batch followed by end of session.
    • Fixed bug where the transpose transform did not set metadata of new columns correctly.
    • Fixed bug with handling null character literals.
    • Fixed bug related to custom data filters when configured as patterns.
    • Fixed bug related to counting of sessions associated with features.
    • Fixed bug where "Data" fields were not getting copied down to children scopes.
    • Fixed bug related to the uninstall.
  • Portal user interface:
    • Table widget sorting in the portal is no longer case-sensitive.
    • Fixed portal rendering issues on iOS.
    • Fixed bug where a column is shown twice in the same spot in a bar chart.
  • Administrative:
    • Fixed bug where removing a custom pattern resulted in errors.
    • Fixed bug with "Replace" merge option where the old data was being preserved instead of being replaced by new data.
    • Fixed bug where permanent store values were getting overwritten instead of being properly merged when new data for the same document arrived after 60 seconds.
    • Installer no longer installs the unneeded 32-bit MSVC++ runtime as a prerequisite.