Apache Hadoop Changelog

Release 0.19.1 - 2009-02-24

INCOMPATIBLE CHANGES:

JIRASummaryPriorityComponentReporterContributor
HADOOP-5225workaround for tmp file handling on DataNodes in 0.19.1 (HADOOP-4663)Blocker.Nigel DaleyRaghu Angadi
HADOOP-5224Disable appendBlocker.Nigel Daley
HADOOP-4061Large number of decommission freezes the NamenodeMajor.Koji NoguchiTsz Wo Nicholas Sze

IMPORTANT ISSUES:

JIRASummaryPriorityComponentReporterContributor

NEW FEATURES:

JIRASummaryPriorityComponentReporterContributor
HADOOP-5034NameNode should send both replication and deletion requests to DataNode in one reply to a heartbeatMajor.Hairong KuangHairong Kuang

IMPROVEMENTS:

JIRASummaryPriorityComponentReporterContributor
HADOOP-5127FSDirectory should not have public methods.Major.Konstantin ShvachkoJakob Homan
HADOOP-5086Trash URI semantics can be relaxedMinorfsChris DouglasChris Douglas
HADOOP-4739Minor enhancements to some sections of the Map/Reduce tutorialTrivial.Vivek RatanVivek Ratan
HADOOP-3894DFSClient chould log errors better, and provide better diagnosticsTrivial.Steve LoughranSteve Loughran

BUG FIXES:

JIRASummaryPriorityComponentReporterContributor
HADOOP-5665Namenode could not be formatted because the “whoami” program could not be run.Major.Evelyn Sylvia
HADOOP-5268Using MultipleOutputFormat and setting reducers to 0 causes org.apache.hadoop.hdfs.protocol.AlreadyBeingCreatedException and job to failMajor.Thibaut
HADOOP-5193SecondaryNameNode does not rollImage because of incorrect calculation of edits modification time.Major.Konstantin ShvachkoKonstantin Shvachko
HADOOP-5192Block reciever should not remove a finalized block when block replication failsBlocker.Hairong KuangHairong Kuang
HADOOP-5166JobTracker fails to restart if recovery and ACLs are enabledBlocker.Karam SinghAmar Kamat
HADOOP-5161Accepted sockets do not get placed in DataXceiverServer#childSocketsMajor.Hairong KuangHairong Kuang
HADOOP-5156TestHeartbeatHandling uses MiniDFSCluster.getNamesystem() which does not exist in branch 0.20MajortestKonstantin ShvachkoHairong Kuang
HADOOP-5134FSNamesystem#commitBlockSynchronization adds under-construction block locations to blocksMapBlocker.Hairong Kuangdhruba borthakur
HADOOP-5067Failed/Killed attempts column in jobdetails.jsp does not show the number of failed/killed attempts correctlyMajor.Amareshwari SriramadasuAmareshwari Sriramadasu
HADOOP-5009DataNode#shutdown sometimes leaves data block scanner verification log unclosedMajor.Hairong KuangHairong Kuang
HADOOP-5008TestReplication#testPendingReplicationRetry leaves an opened fd unclosedMajortestHairong KuangHairong Kuang
HADOOP-50022 core tests TestFileOutputFormat and TestHarFileSystem are failing in branch 19Blocker.Ravi GummadiAmareshwari Sriramadasu
HADOOP-4992TestCustomOutputCommitter fails on hadoop-0.19Blocker.Amar KamatAmareshwari Sriramadasu
HADOOP-4983Job counters sometimes go down as tasks run without task failuresCritical.Owen O'MalleyAmareshwari Sriramadasu
HADOOP-4982TestFsck does not run in Eclipse.MajortestKonstantin ShvachkoKonstantin Shvachko
HADOOP-4967Inconsistent state in JVM managerMajor.Amareshwari SriramadasuDevaraj Das
HADOOP-4966Setup tasks are not removed from JobTracker's taskIdToTIPMap even after the job completesMajor.Amar KamatAmareshwari Sriramadasu
HADOOP-4965DFSClient should log instead of printing into std err.MajortestKonstantin ShvachkoKonstantin Shvachko
HADOOP-4955Make DBOutputFormat us column names from setOutput(...)Major.Kevin PetersonKevin Peterson
HADOOP-4943fair share scheduler does not utilize all slots if the task trackers are configured heterogeneouslyMajor.Zheng ShaoZheng Shao
HADOOP-4924Race condition in re-init of TaskTrackerBlocker.Devaraj DasDevaraj Das
HADOOP-4918Fix bzip2 work with SequenceFileMajorioZheng ShaoZheng Shao
HADOOP-4906TaskTracker running out of memory after running several tasksBlocker.Arun C MurthySharad Agarwal
HADOOP-4862A spurious IOException log on DataNode is not completely removedBlocker.Raghu AngadiRaghu Angadi
HADOOP-4847OutputCommitter is loaded in the TaskTracker in localizeConfigurationBlocker.Owen O'MalleyAmareshwari Sriramadasu
HADOOP-4836Minor typos in documentation and commentsTrivialdocumentationJordà PoloJordà Polo
HADOOP-4821Usage description in the Quotas guide documentations are incorrectMinordocumentationBoris ShkolnikBoris Shkolnik
HADOOP-4797RPC Server can leave a lot of direct buffersBlockeripcRaghu AngadiRaghu Angadi
HADOOP-4760HDFS streams should not throw exceptions when closed twiceMajorfs, fs/s3Alejandro AbdelnurEnis Soztutar
HADOOP-4759HADOOP-4654 to be fixed for branches >= 0.19Major.Amareshwari SriramadasuAmareshwari Sriramadasu
HADOOP-4731Job is not removed from the waiting jobs queue upon completion.Major.Hemanth YamijalaAmar Kamat
HADOOP-4727Groups do not work for fuse-dfs out of the box on 0.19.0Blocker.Brian BockelmanBrian Bockelman
HADOOP-4720docs/api does not contain the hdfs directory after buildingMajorbuildRamya Sunil
HADOOP-4697KFS::getBlockLocations() fails with files having multiple blocksMajorfsLohit VijayarenuSriram Rao
HADOOP-4635Memory leak ?Blocker.Marc-Olivier FleuryPete Wyckoff
HADOOP-4632TestJobHistoryVersion should not create directory in current dir.Major.Amareshwari SriramadasuAmar Kamat
HADOOP-4616assertion makes fuse-dfs exit when reading incomplete dataBlocker.Marc-Olivier FleuryPete Wyckoff
HADOOP-4508FSDataOutputStream.getPos() == 0when appending to existing file and should be file lengthMajorfsPete Wyckoffdhruba borthakur
HADOOP-4494libhdfs does not call FileSystem.append when O_APPEND passed to hdfsOpenFileMajor.Pete WyckoffPete Wyckoff
HADOOP-4420JobTracker.killJob() doesn't check for the JobID being validMinor.Steve LoughranAaron Kimball

TESTS:

JIRASummaryPriorityComponentReporterContributor

SUB-TASKS:

JIRASummaryPriorityComponentReporterContributor

OTHER:

JIRASummaryPriorityComponentReporterContributor