Druid 0.8.1 - Stable
Updating from 0.8.0
There should be no update concerns and standard updating procedures can be followed for rolling updates
New Features
- #1259 Experimental Query Time Lookups (QTL) -– Ability to do limited joins at query time.
Simple example use case is Country Code to Country Name. - #1374 Experimental Hadoop batch re-indexing and Delta ingestion.
Re-Indexing allows you to ingest existing druid segments using a new schema with certain columns removed, changed granularity etc. "Delta" Ingestion allows appending data to existing interval in a datasource. See the new dataSource inputSpec and multi inputSpec for more information.
Improvements
- #1465 Read Hadoop configuration file from HDFS
- #1472 Support using combiner for Hadoop ingestion
- #1506 Better support for null input rows during ingestion
- #1518 More support added for Azure deep store
- #1550 Add configuration option to print all HTTP requests to log
- #1563 #1602 Improved merging performance on Broker
- #1567 #1568 Improved error logging for segment activities
- #1596 Improved coordinator console, now a separate maven dependency instead of giant code dump
- #1601 Reduced lock contention during segment scan
- #1603 Improved performance of Lexicographic TopNs
- #1643 helpful cause explaining why SegmentDescriptorInfo did not exist
Improved test coverage for indexing service, ingestion, and coordinator endpoints
Bug Fixes
- #1406 Fix groupBy breaking when exceeding max intermediate rows
- #1441 Fix flush errors being suppressed when closing output streams
- #1469 Fix inconsistent property names for druid.metadata.* properties
- #1484 JobHelper.ensurePaths will set properties from config properly
- #1499 Fix groupBy caching with renamed aggregators
- #1503 Fix leaking indexing service status nodes in ZK
- #1534 Fix caching for approximate histograms
- #1616 Fix dependency error in local index task
- #1627 Fix realtime tasks getting stuck on shutdown even after status being shown as SUCCESS
- #1634 Allow IrcFirehoseFactory to shutdown cleanly
- #1640 Package extensions in release tarball + script to run druid servers
- #1653 Fix success flag emitted in router query metrics
- #1659 on kill segment, don't leave version, interval and dataSource dir behind on HDFS
- #1681 Fix overlapping segments not working for ingest segment firehose
Documentation
- New documentation for firehoses, evaluating Druid, and plenty of fixes.
- Improved documentation for working with CDH
- Added instructions for PostgreSQL metadata store
- More documentation on how to use ApproximateHistograms
The full list of changes can be found here
Thanks
Special thanks to everyone that contributed (code, docs, etc.) to this release!
@drcrallen
@davideanastasia
@guobingkun
@himanshug
@michaelschiff
@fjy
@krismolendyke
@nishantmonu51
@rasahner
@xvrl
@gianm
@pjain1
@samjhecht
@solimant
@sherry-q
@ubercow
@zhaown
@mvfast
@mistercrunch
@pdeva
@KurtYoung
@onlychoice
@b-slim
@cheddar
@MarConSchneid