[BIGTOP-223] install/ initscripts fixes

• Type: Bug
• Status: Open
• Priority: Critical
• Resolution: Unresolved
• Affects Version/s: 0.2.0
• Assignee: Peter Linnell
• Reporter: Peter Linnell
• Created: 01/Nov/11 18:53
• Updated: 26/Mar/13 18:14

Description

master is not running failed
node is not running failed

should be prefixed by flume

I will fix this as part of my initscript audit/clean-up

Another case:

hbase:
hadoop-hbase-master: unrecognized service
hadoop-hbase-regionserver: unrecognized service
hadoop-hbase-thrift: unrecognized service

hadoop namenode:
/etc/init.d/hadoopnamenode: line 23: /etc/default/hadoo-: No such file or directory
/usr/lib/hadoop-/bin/hadoop-daemon.sh: line 42: /usr/lib/hadoop/bin/hadoop-config.sh: No such file or directory

sqoop-metastore:
sh: /usr/lib/sqoop/bin/stop-metastore.sh: No such file or directory

zookeeper-server:
JMX enabled by default
Using config: /etc/zookeeper/zoo.cfg
Stopping zookeeper ...

Issue Links

• incorporates: BIGTOP-225 Exit consistently in init scripts for stop on stopped service
• is related to: BIGTOP-135 make jsvc use detection more robust on SLES
• is related to: BIGTOP-172 Hadoop init.d scripts are flaky on stop

Sub-Tasks

• Flume's init script output is confusing when restarting: Open
• Oozie init scripts are way too verbose: Open
• HBase init scripts are shut down in the wrong order: Closed
• hbase initscripts should be more explicit: Open
• hadoop jobtracker and tasktracker status command do not detect running jobtracker/tasktracker: Open
• Initscripts have misleading messages: Open