========================================================================================= Oracle Big Data Discovery GENERAL (1.4.0) - Software Release Date: October 2016 - The release notes and related product documentation may have been updated since the release date. The latest release notes and documentation are available through the Oracle Big Data Discovery documentation page on the Oracle Help Center at https://docs.oracle.com/cd/E78171_01/index.htm or by contacting Oracle Technical Support for Big Data Discovery. INSTALLATION AND UPGRADE (1.4.0) - Refer to the "Oracle Big Data Discovery Installation Guide" for installation information. - See the "Oracle Big Data Discovery Upgrade Guide" for information on upgrading to this release. BUG FIXES (1.4.0) Upgrade: - BDDSPT-339: Fixed an issue that caused the upgrade script to fail when cleaning up the ZooKeeper data. Data Processing: - EDPROC-4713: Previously, the RECORD_SEARCH_THRESHOLD and VALUE_SEARCH_THRESHOLD configuration properties in the DP code were misspelled. The misspelling meant that Data Processing could not match the properties to those in bdd.conf. This problem has been fixed. Dgraph: - ESERVER-2819: Fixed a problem where the Dgraph did not log warnings about invalid log levels or topics if the "dgraph" topic was assigned an ERROR or INCIDENT_ERROR log level. - ESERVER-3220: Fixed a problem where the Dgraph would allow users to set an invalid configuration for cache and memory limit settings. Now, the Dgraph validates that the cache size (--cmem flag) must be smaller than the memory limit (--memory-limit flag). Studio: - ESTUDIO-12103: Fixed an issue where extracting a time part on time attributes using the HH or hh hour options results in an incorrect result in Transform. Also added two missing options 'Minutes (mm)' and 'Seconds (ss)' to the editor. - ESTUDIO-12320: Fixed an issue where you may have encountered connection problems when importing user groups into Studio from an OpenLDAP system. - ESTUDIO-12338: Fixed an issue where the favorite attributes weren't displayed in the first positions within the dropdown list of Scratchpad. - ESTUDIO-12344: Fixed an issue where hidden attributes in a linked view were still displayed in Data Views page of Control Panel. - ESTUDIO-12478: Fixed an issue in a linked data set used by a Chart where the name of a data set follows the names of the attributes in the attribute list. - ESTUDIO-12601: Fixed an issue where a Negative refinement cannot work if the refine value included single quote. - ESTUDIO-12604: Fixed an issue in Scatterplot Matrix where the marquee selection cannot be cleared. - ESTUDIO-12638: Fixed an issue where there were redundant map refreshes when adding or removing refinements. - ESTUDIO-12679: Fixed an issue when changing the attribute in Y-axis zone of a Bar chart where the sort rule of X-axis was reset to default. - ESTUDIO-12706: Fixed an issue where toggling sorts between "Information potential" and "Null percentage" didn't update the attributes order in Explore. - ESTUDIO-12708: Fixed an issue when saving a Bubble chart from Explore to Discover where the sort option reset to the default value. - ESTUDIO-12735: Fixed an issue where the Chart may not display correctly when resizing the browser. - ESTUDIO-12747: Fixed an issue where an LDAP Test connection would fail for a non-DN principal. - ESTUDIO-12770: Fixed an issue in the Data Set Manager where the Record Identifier information disappeared after refreshing page. - ESTUDIO-12773: Fixed an issue where, when saving a Scatterplot Matrix to Discover, the component didn't render after changing the sub-type. - ESTUDIO-12777: Fixed an issue where reordering transform steps would give an obscure error message instead of showing the problems clearly with script validation in the Transform panel. - ESTUDIO-12827: Fixed an issue in the Tag Cloud component where changing to the list format didn't update the component. - ESTUDIO-13008: Fixed an issue where the attribute group name can't be localized. Now, a group name can be localized in all available locales. - ESTUDIO-13026: Fixed an issue where the Selected Refinement disappears after adding a data set to a project. - ESTUDIO-13263: Fixed an issue with the Custom Visualization Component where it would not update token names after changing the view name in the EQL query. - ESTUDIO-13311: Fixed an issue where a data set cannot be deleted if it is created using Export. - ESTUDIO-13312: Fixed an issue where JDBC import would fail if you changed the SQL query after having already applied filtering. - ESTUDIO-13314: Fixed an issue where you can't create a data set from a database when using a string filter rule. - ESTUDIO-13454: Fixed an issue where a Chart cannot display when an attribute's format scale is "log" and the selected range includes 0. - ESTUDIO-13662: Fixed a problem in the Transform whitelist editor where you cannot delete whitelist items. - ESTUDIO-13679: Fixed an issue with LDAP import that would duplicate user groups if their name contained an ampersand. - ESTUDIO-13752: Fixed an issue where the Summarization Bar wasn't updated to use the linked view after creating a data set link. - ESTUDIO-13784: Fixed an issue where the start time of notification information was displayed in UTC time. Now, the start time is displayed in user's local time and correct format. - ESTUDIO-13837: Fixed an issue where you were able to delete a data set that was currently being used in a project if that data set had been reloaded. - ESTUDIO-13852: Fixed an issue in Thematic map where you couldn't refine when using a linked view. - ESTUDIO-13967: Fixed an issue where the "Accept Data Updates" commit did not delete the project's dataset. - ESTUDIO-14032: Fixed an issue where histograms would appear blank in value search. Documentation: - EDOC-2477: Fixed an issue in the Administrator's Guide where the full path to bdd.conf was not clear. This created some confusion between the post-install location and the installation source location of the file: in some cases, it wasn't clear which version of the file to modify. KNOWN ISSUES (1.4.0) Data Processing: - EDPROC-4625: The $BDD_HOME/dataprocessing/edp_cli/config directory contains these unused configuration files: cluster_jars.txt, dateFormats.txt, and sparkContext.properties. The versions of these files that are used by the Data Processing component are in the $BDD_HOME/workflowmanager/dp/config directory. - EDPROC-4885: DP CLI Refresh updates (using the --refreshData flag) do not ignore the maxRecordsForNewDataSet property in the edp-cli.properties configuration file. As a result, sampled data sets are produced, instead of full data sets. Two workarounds are to adjust the maxRecordsForNewDataSet property to be higher than the number of records in your Hive table or to use the "Load Full Data Set" option in Studio. Dgraph: - ESERVER-1623 and ESERVER-2069: Very rarely, if ZooKeeper is down, after trying to reconnect for a period, Dgraph may stop working or cease all reconnection attempts. - ESERVER-1802 and ESERVER-3069: In rare conditions, Dgraph shutdown may be delayed by up to 40 seconds if ZooKeeper is turned off while the Dgraph is shutting down. - ESERVER-2761, ESERVER-2762, and ESERVER-2763: Shutting down the Dgraph process with the "kill" command, or performing a hard shutdown on the Dgraph during database creation or deletion may result in artifacts left on ZooKeeper that may not be removable by future Dgraph instances. - ESERVER-2900: Dgraph may be temporarily unable to promote to a leader of a database if it is performing a long query that began prior to the latest data update. - ESERVER-3415: Text search does not work correctly against data with multi-byte characters, such as Japanese. - ESERVER-3552: Invalid value of the --zookeeper_index flag will cause Dgraph be a zombie process. That is, the Dgraph remains alive but unable to process any data-related requests. - ESERVER-3608: A snapshot operation will fail if run against a version that is not the absolute latest. - ESERVER-3648: During an ingest, you may see "Database 'name' does not exist" messages in the Dgraph and Dgraph Gateway logs. These messages are benign and do not affect Dgraph operations. The message is logged when a Dgraph databae does not exist before the ingest, but it is created immediately afterwards. Studio: - ESTUDIO-14341: The Notifications panel does not correctly indicate the status of transformations that have been stopped using the Yarn ResourceManager. In this case, the status information remains stuck at the last percentage complete. To work around this issue, manually delete the notification entry for the stopped process from the Studio database. For example, if the Studio database is mySQL, the steps are as follows: 1. Find the log of the YARN job you killed, search for "ecid", and copy the value of ecid (for example, "1.0000LWRqOOD3NAk5Gzl3if1O5s9N00000l;kZjE") 2. Log in to mysql shell with the proper privileges. 3. Delete the related notification entry by typing: "delete from .ENDECA_MESSAGES where ecid=''" where "" is the Studio database name and "" is the ecid value found in the YARN job log. Also, Oracle recommends using the DP CLI to manage jobs rather than using the Yarn ResourceManager. This way, notifications are handled correctly by BDD. For example, using the DP CLI, you can run get-active-jobs to identify the that you want to stop and then run cancel-job to cleanly stop that job. =========================================================================================