Apache Hadoop Changelog

Release 2.6.4 - 2016-02-11

INCOMPATIBLE CHANGES:

JIRA Summary Priority Component Reporter Contributor
YARN-3154 Should not upload partial logs for MR jobs or other "short-running’ applications Blocker nodemanager, resourcemanager Xuan Gong Xuan Gong

IMPORTANT ISSUES:

JIRA Summary Priority Component Reporter Contributor

NEW FEATURES:

JIRA Summary Priority Component Reporter Contributor

IMPROVEMENTS:

JIRA Summary Priority Component Reporter Contributor
HDFS-9415 Document dfs.cluster.administrators and dfs.permissions.superusergroup Major documentation Arpit Agarwal Xiaobing Zhou
HDFS-9314 Improve BlockPlacementPolicyDefault’s picking of excess replicas Major . Ming Ma Xiao Chen
HDFS-8722 Optimize datanode writes for small writes and flushes Critical . Kihwal Lee Kihwal Lee
HDFS-8647 Abstract BlockManager’s rack policy into BlockPlacementPolicy Major . Ming Ma Brahma Reddy Battula
HDFS-7694 FSDataInputStream should support “unbuffer” Major . Colin Patrick McCabe Colin Patrick McCabe
MAPREDUCE-6436 JobHistory cache issue Blocker . Ryu Kobayashi Kai Sasaki

BUG FIXES:

JIRA Summary Priority Component Reporter Contributor
HADOOP-12706 TestLocalFsFCStatistics#testStatisticsThreadLocalDataCleanUp times out occasionally Major test Jason Lowe Sangjin Lee
HADOOP-12107 long running apps may have a huge number of StatisticsData instances under FileSystem Critical fs Sangjin Lee Sangjin Lee
HADOOP-11252 RPC client does not time out by default Critical ipc Wilfred Spiegelenburg Masatake Iwasaki
HDFS-9600 do not check replication if the block is under construction Critical . Phil Yang Phil Yang
HDFS-9574 Reduce client failures during datanode restart Major . Kihwal Lee Kihwal Lee
HDFS-9445 Datanode may deadlock while handling a bad volume Blocker . Kihwal Lee Walter Su
HDFS-9313 Possible NullPointerException in BlockManager if no excess replica can be chosen Major . Ming Ma Ming Ma
HDFS-9294 DFSClient deadlock when close file and failed to renew lease Blocker hdfs-client DENG FEI Brahma Reddy Battula
HDFS-9220 Reading small file (< 512 bytes) that is open for append fails due to incorrect checksum Blocker . Bogdan Raducanu Jing Zhao
HDFS-9178 Slow datanode I/O can cause a wrong node to be marked bad Critical . Kihwal Lee Kihwal Lee
HDFS-8767 RawLocalFileSystem.listStatus() returns null for UNIX pipefile Critical . Haohui Mai Kanaka Kumar Avvaru
HDFS-6945 BlockManager should remove a block from excessReplicateMap and decrement ExcessBlocks metric when the block is removed Critical namenode Akira AJISAKA Akira AJISAKA
HDFS-4660 Block corruption can happen during pipeline recovery Blocker datanode Peng Zhang Kihwal Lee
MAPREDUCE-6621 Memory Leak in JobClient#submitJobInternal() Major . Xuan Gong Xuan Gong
MAPREDUCE-6619 HADOOP_CLASSPATH is overwritten in MR container Major mrv2 shanyu zhao Junping Du
MAPREDUCE-6618 YarnClientProtocolProvider leaking the YarnClient thread. Major . Xuan Gong Xuan Gong
MAPREDUCE-6577 MR AM unable to load native library without MR_AM_ADMIN_USER_ENV set Critical mr-am Sangjin Lee Sangjin Lee
MAPREDUCE-6554 MRAppMaster servicestart failing with NPE in MRAppMaster#parsePreviousJobHistory Critical . Bibin A Chundatt Bibin A Chundatt
MAPREDUCE-6492 AsyncDispatcher exit with NPE on TaskAttemptImpl#sendJHStartEventForAssignedFailTask Critical . Bibin A Chundatt Bibin A Chundatt
MAPREDUCE-6363 [NNBench] Lease mismatch error when running with multiple mappers Critical benchmarks Brahma Reddy Battula Bibin A Chundatt
MAPREDUCE-5982 Task attempts that fail from the ASSIGNED state can disappear Major mr-am Jason Lowe Chang Li
YARN-4598 Invalid event: RESOURCE_FAILED at CONTAINER_CLEANEDUP_AFTER_KILL Major nodemanager tangshangwen tangshangwen
YARN-4581 AHS writer thread leak makes RM crash while RM is recovering Major resourcemanager sandflee sandflee
YARN-4546 ResourceManager crash due to scheduling opportunity overflow Critical resourcemanager Jason Lowe Jason Lowe
YARN-4452 NPE when submit Unmanaged application Critical . Naganarasimha G R Naganarasimha G R
YARN-4414 Nodemanager connection errors are retried at multiple levels Major nodemanager Jason Lowe Chang Li
YARN-4380 TestResourceLocalizationService.testDownloadingResourcesOnContainerKill fails intermittently Major test Tsuyoshi Ozawa Varun Saxena
YARN-4354 Public resource localization fails with NPE Blocker nodemanager Jason Lowe Jason Lowe
YARN-4180 AMLauncher does not retry on failures when talking to NM Critical resourcemanager Anubhav Dhoot Anubhav Dhoot
YARN-3857 Memory leak in ResourceManager with SIMPLE mode Critical resourcemanager mujunchao mujunchao
YARN-3849 Too much of preemption activity causing continuos killing of containers across queues Critical capacityscheduler Sunil G Sunil G
YARN-3842 NMProxy should retry on NMNotYetReadyException Critical . Karthik Kambatla Robert Kanter
YARN-3697 FairScheduler: ContinuousSchedulingThread can fail to shutdown Critical fairscheduler zhihai xu zhihai xu
YARN-3695 ServerProxy (NMProxy, etc.) shouldn’t retry forever for non network exception. Major . Junping Du Raju Bairishetti
YARN-3535 Scheduler must re-request container resources when RMContainer transitions from ALLOCATED to KILLED Critical capacityscheduler, fairscheduler, resourcemanager Peng Zhang Peng Zhang
YARN-2975 FSLeafQueue app lists are accessed without required locks Blocker . Karthik Kambatla Karthik Kambatla

TESTS:

JIRA Summary Priority Component Reporter Contributor
HADOOP-12736 TestTimedOutTestsListener#testThreadDumpAndDeadlocks sometimes times out Major . Xiao Chen Xiao Chen

SUB-TASKS:

JIRA Summary Priority Component Reporter Contributor
YARN-3893 Both RM in active state when Admin#transitionToActive failure from refeshAll() Critical resourcemanager Bibin A Chundatt Bibin A Chundatt
YARN-2902 Killing a container that is localizing can orphan resources in the DOWNLOADING state Major nodemanager Jason Lowe Varun Saxena

OTHER:

JIRA Summary Priority Component Reporter Contributor