Minor changes and bug fixes are as listed below.
- HDFS metrics available with Ganglia
- HAD-425: Added configuration variable ‘fs.inputpathprocessor.ignoreIOException’, which when set ‘true’ will ignore FileNotFound Exception thrown when IPP (Input path processor) optimization for S3 paths is turned off. This variable is set to ‘false’ by default.
- HADTWO-261 Tweak autoscaling algorithm and change job latency to two minutes. Yarn would upscale on pending immediately rather than waiting for 60 seconds.
- HIVE-887. Supporting Set type in thrift
- HIVE-880. Fix hang during ORC split computation
- HIVE-831. Fix for cross account roles support in hive13
- UI-1254 - show only unique tags, also make tags case insensitive
- UI-1774 -Scheduler Tag input box masks input
- UI-1763: Scheduled job: 'kill' button caption change to 'stop'
- UI-1777: show only 200 chars of command template command detail in tabular form
- HAD-440 Kill orphan child jobs for composite commands
- Tagging support for scheduled commands
- QBOL-3797: Private key is not required while creating an on-premise db-tap
- SQL must take Spark submit args specified by the user
- QBOL-4070: Avoiding DBExport/DBImport hive queries to use hive2 jars
- UI-1769: Post to Forum, Video Tutorials Integration in Help Widget
- QPIG-17 pig-on-hadoop2 commands supported as top level command.
- HBase clusters termination turned off by default
- Properly encode SQL command line argument
- SPAR-208: Allow avro to be used in spark.
- SPAR-309 Make driver memory configurable in zeppelin
- SPAR-351 Fix the idle timeout calculation in zeppelin