1. Hadoop component dependencies
2. Hadoop log format:
Two kinds of logs, ending with out and log respectively:
1 Logs ending with log : Logs recorded in the log4j logging format, using the daily rolling file suffix strategy to name the log files, and the content is relatively complete.
2 Logs ending with out: Logs that record standard output and mark errors, with less content. By default, the system keeps the latest 5 log files.
This can be configured in /etc/hadoop/hadoop-env.sh:
33 #Where log files are stored. $HADOOP_HOME/logs by default. 34 #export HADOOP_LOG_DIR = ${HADOOP_HOME}/logs
Explanation of log names:
Second, hadoop start and stop:
The first way:
start up:
start-dfs.sh
start-mapred.sh（hadoop 2.x为 start-yarn.sh）
stop:
stop-dfs.sh
stop-mapred.sh（Hadoop 2.x为 stop-yarn.sh）
All start:
start-all.sh
Startup sequence : NameNode --> DataNode --> Secondary NameNode --> JobTracker --> TaskTracker
stop all:
stop-all.sh
Stop sequence : JobTracker --> TaskTracker --> NameNode --> DataNode --> Secondary NameNode
The second way (the daemons are started and shut down one by one):
start up:
1 hadoop-daemon.sh start namenode 2 hadoop-daemon.sh start datanoe 3 hadoop-daemon.sh start secondarynamenode 4 hadoop-daemon.sh start jobtracker 5 hadoop-daemon.sh start tasktracker
The startup sequence is the same as above : NameNode --> DataNode --> Secondary NameNode --> JobTracker --> TaskTracker
stop:
1 1 hadoop-daemon.sh stop jobtracker 2 2 hadoop-daemon.sh stop tasktracker 3 3 hadoop-daemon.sh stop namenode 4 4 hadoop-daemon.sh stop datanoe 5 5 hadoop-daemon.sh stop secondarynamenode