CDH 5.9.3 Release Notes
The following lists all Spark Netlib Jiras included in CDH 5.9.3
that are not included in the Spark Netlib base version 1.1. The
spark-netlib-1.1-cdh5.9.3.CHANGES.txt
file lists all changes included in CDH 5.9.3. The patch for each
change can be found in the cloudera/patches directory in the release tarball.
Changes Not In Spark Netlib 1.1
Bigtop
Bug
- [BIGTOP-1050] - Permissions on YARN LCE should be 4754
- [BIGTOP-1194] - redhat-lsb-core is sufficient in spec files
- [BIGTOP-1589] - Prioritization is broken in bigtop-detect-javahome
- [BIGTOP-1356] - Generate hive.install on the fly
- [BIGTOP-1438] - asciidoc is no longer needed by Sqoop
- [BIGTOP-1398] - hadoop-hdfs-fuse can't be installed on Ubuntu 14.04
- [BIGTOP-1371] - Spark-master service fails to start due to missing /usr/lib/spark/work
- [BIGTOP-1308] - Installing HBase removes chrony if it is configured
- [BIGTOP-1306] - RPM licensing metadata is inconsistent and wrong
- [BIGTOP-1266] - Zookeeper init script doesn't source /etc/default/zookeeper
- [BIGTOP-1215] - Refactor Tomcat deployment
- [BIGTOP-1193] - Update Spark packaging for various conventions
- [BIGTOP-1189] - spark-python dependency needs to be updated to spark-core
- [BIGTOP-1187] - Spark service packages have extraneous core in the name
- [BIGTOP-1182] - Location of HCat binary default file not overridable
- [BIGTOP-1170] - Add an easier tool for uploading / creating the sharelib
- [BIGTOP-1027] - 'bigtop-detect-classpath' helper script missing from bigtop-utils
- [BIGTOP-1150] - update Spark debian packaging to the latest format
- [BIGTOP-1125] - Return value does not reflect status checks
- [BIGTOP-1114] - Hue metastore app is missing from core registration and Debian packages
- [BIGTOP-1112] - sqoop-client should require bigtop-utils
- [BIGTOP-1044] - hive and hive-jdbc versions should be kept in sync
- [BIGTOP-1045] - Be consistent with shells, bigtop-detect-javahome, and bigtop-utils versions
- [BIGTOP-1037] - Provide a mechanism to control the sourcing of defaults files
- [BIGTOP-1005] - Create versionless symlinks in hadoop client directory
- [BIGTOP-1017] - Revert a commit that introduces some unneeded dependencies
- [BIGTOP-988] - HCatalog does not launch daemon process in a known directory
- [BIGTOP-975] - HBase regionserver init script does not implement "condrestart" command
- [BIGTOP-984] - bigtop-utils doesn't detect JAVA_HOME for Oracle JDK 7 when package by Debian's tool
- [BIGTOP-983] - Oozie-client package ships its docs twice
- [BIGTOP-982] - hcatalog needs to be given HCAT_HOME and HIVE_HOME
Improvement
- [BIGTOP-1085] - spark packages needs to be split
- [BIGTOP-1139] - bigtop-tomcat and bigtop-jsvc don't depend on bigtop-utils
- [BIGTOP-1138] - hadoop-conf-pseudo package uses the deprecated fs.default.name parameter
- [BIGTOP-1137] - provide a way to request a major version of JDK
New Feature
- [BIGTOP-811] - Add /var/lib/bigtop as a location to install SQL connectors and other plug-ins
- [BIGTOP-39] - Include Apache Avro in Bigtop
Task
MapReduce
Bug
- [MAPREDUCE-4649] - mr-jobhistory-daemon.sh needs to be updated post YARN-1
Oozie
Improvement
- [OOZIE-1560] - Log messages should have a way of identifying which server they came from when using HA
CDH
Bug
- [DISTRO-533] - Solr server watchdog makes heartbeat request using server short hostname, not FQDN
Sentry
Bug
- [SENTRY-24] - NOTICE.txt missing from generated tarball
Impala
Bug
- [IMPALA-3529] - Package the minidump collection script along with Impala.
Improvement
- [IMPALA-2889] - Generate one combined binary for impalad, statestored, catalogd
- [IMPALA-2632] - Link LLVM bytecode into impalad binary
Task