Hadoop 2.6.2 Release Notes
These release notes include new developer and user-facing incompatibilities, features, and major improvements.
Changes since Hadoop 2.6.1
Critical sub-task reported by Yesha Vora and fixed by Xuan Gong (yarn)
RM should print alert messages if Zookeeper and Resourcemanager gets connection issue
Major sub-task reported by Xuan Gong and fixed by Xuan Gong (resourcemanager)
RM HA UI redirection needs to be fixed when both RMs are in standby mode
Major bug reported by Jian He and fixed by Jian He
Followup fixes after YARN-2019 regarding RM behavior when state-store error occurs
Major bug reported by Jun Gong and fixed by Jun Gong
Completed container whose app is finished is not removed from NMStateStore
Major bug reported by Jun Gong and fixed by Jun Gong (resourcemanager)
RMNode transitioned from RUNNING to REBOOTED because its response id had not been reset synchronously
Major bug reported by zhihai xu and fixed by zhihai xu (resourcemanager)
Two RMNodes for the same NodeId are used in RM sometimes after NM is reconnected.
Blocker bug reported by Bibin A Chundatt and fixed by Varun Saxena (resourcemanager)
ZKRMStateStore shouldn't create new session without occurrance of SESSIONEXPIED
Minor bug reported by zhihai xu and fixed by zhihai xu (resourcemanager)
Should use equals when compare Resource in RMNodeImpl#ReconnectNodeTransition
Major improvement reported by zhihai xu and fixed by zhihai xu (nodemanager)
For better error recovery, check if the directory exists before using it for localization.
Major bug reported by Jason Lowe and fixed by Naganarasimha G R
Default value for maximum nodemanager connect wait time is too high
Blocker bug reported by Rohith Sharma K S and fixed by Rohith Sharma K S (resourcemanager)
RM should handle NMContainerStatuses sent by NM while registering if NM is Reconnected node
Critical bug reported by Junping Du and fixed by Jian He
Retrospect on decision of making RM crashed if any exception throw in ZKRMStateStore
Major bug reported by Shinichi Yamashita and fixed by Shinichi Yamashita
Fix wrong value of JOB_FINISHED event in JobHistoryEventHandler
Critical bug reported by Junping Du and fixed by Junping Du
MapReduce doesn't set the HADOOP_CLASSPATH for jar lib in distributed cache.
Blocker bug reported by Eric Payne and fixed by Eric Payne
Fetcher#copyMapOutput is leaking usedMemory upon IOException during InMemoryMapOutput shuffle handler