最近集群空间有点紧张,总是担心空间不足而崩溃,近期扩容又不太现实,经与集群用户沟通发现:集群上存储了很多无用的历史数据,可以删除,这样就可以通过一个crontab脚本每天生成集群空间使用报告,当使用量超过70%、80%是分别报警,并通知那些用户占用空间较大,预留当机冗余空间,这样就不需要时刻担心集群空间爆满了。
[hdfs@hanagios48 root]$ more /home/hdfs/dfsadmin_report.sh
#!/bin/bash
source ~/.bash_profile
today=$(date +%Y%m%d)
#report=`hadoop dfsadmin -report|head -n 11`
report=/tmp/report
echo "Hi,各位集群用户
当Hadoop集群空间使用量达到70%时,各位将会收到Warning邮件,提示清理空间;
当Hadoop集群空间使用量到达80%时,各位将会收到Critical邮件, 提示清理空间;
为保证集群数据安全,预留down机冗余空间,请各位务必清理,如因数据量确实巨大,集群无法容纳,请及时告知运维进行扩容,谢谢!">$report
echo >>$report
hadoop dfsadmin -report|head -n 11 >>$report
echo ---------------------- >>$report
echo dfs used details: >>$report
hadoop fs -du / >>$report
echo >>$report
hadoop fs -du /user >>$report
dfs_used_percent=`cat $report|grep "DFS Used%"|awk -F: ‘{print $2}‘`
dfs_used=`echo ${dfs_used_percent}|awk -F% ‘{print $1}‘`
# 百分百小数比较
#expr ${dfs_used} \>\= 80
user=laijingli2006@126.com
title=`echo "${today}[${dfs_used_percent}] WBY Hadoop Cluster Hdfs Useage Report: dfs_used ${dfs_used_percent}"`
echo $title
if [ $(expr ${dfs_used} \>\= 80) = 1 ];then
echo dfs_used 80
#cat $report|mail -s "Critical: $title" $user
cat $report|/usr/bin/mutt -s "Critical: $title" $user
elif [ $(expr ${dfs_used} \>\= 70) = 1 ];then
echo 70
#cat $report|mail -s "Warning: $title" $user
cat $report|/usr/bin/mutt -s "Warning: $title" $user
else
echo 60
#cat $report|mail -s "Normal: $title" $user
#cat $report|/usr/bin/mutt -s "Normal: $title" $user
fi
#cat $report|mail -s $today DfsReport: dfs_used ${dfs_used_percent} 362560701@qq.com
crontab执行效果还不错:
[hdfs@hanagios48 root]$ crontab -l
05 8 * * * /home/hdfs/dfsadmin_report.sh
[hdfs@hanagios48 root]$ more /tmp/report
Hi,各位集群用户
当Hadoop集群空间使用量达到70%时,各位将会收到Warning邮件,提示清理空间;
当Hadoop集群空间使用量到达80%时,各位将会收到Critical邮件,提示清理空间;
为保证集群数据安全,预留down机冗余空间,请各位务必清理,如因数据量确实巨大,集群无法容纳,请及时告知运维进行扩容,谢谢!
Configured Capacity: 124854950621184 (113.55 TB)
Present Capacity: 118317151626783 (107.61 TB)
DFS Remaining: 38704545865728 (35.2 TB)
DFS Used: 79612605761055 (72.41 TB)
DFS Used%: 67.29%
Under replicated blocks: 0
Blocks with corrupt replicas: 0
Missing blocks: 0
-------------------------------------------------
Datanodes available: 15 (15 total, 0 dead)
----------------------
dfs used details:
Found 6 items
0 hdfs://hamaster140:9000/benchmarks
125752 hdfs://hamaster140:9000/data0
0 hdfs://hamaster140:9000/system
13721821810608 hdfs://hamaster140:9000/tech
1803375805154 hdfs://hamaster140:9000/tmp
6411197575455 hdfs://hamaster140:9000/user
Found 14 items
33222938 hdfs://hamaster140:9000/user/azk
4072247213805 hdfs://hamaster140:9000/user/cla
40705761240 hdfs://hamaster140:9000/user/din
0 hdfs://hamaster140:9000/user/fea
0 hdfs://hamaster140:9000/user/gao
36454169547 hdfs://hamaster140:9000/user/gmz
1877816487439 hdfs://hamaster140:9000/user/hdf
148965233376 hdfs://hamaster140:9000/user/imp
2416017438 hdfs://hamaster140:9000/user/in
0 hdfs://hamaster140:9000/user/lin
0 hdfs://hamaster140:9000/user/luo
149973222708 hdfs://hamaster140:9000/user/shi
82586246964 hdfs://hamaster140:9000/user/wuy
0 hdfs://hamaster140:9000/user/zho
[hdfs@hanagios48 root]$
原文地址:http://blog.csdn.net/funbox/article/details/25288609