-hadoop simple monitor
周梦想 2013-01-05, 09:29
I write a bash script to monitor all nodes of Hadoop system. It can monitor
memory, disk usage and hadoop,thirift,hbase process exit.
Simple Monitor for Hadoop Processes, Memory and Diskspace usage. If memory
and disk usage reaches to a threshold, or some hadoop system process exit,
the program will send email to make warnings.
1. Simple Monitor Hadoop Processes, Memory and Diskspace usage of every
hosts that configured.
2. Deploy is simple. Just copy or uncompress the code to one station,
and it can monitor each remote hosts.( Do not need to deploy to any nodes
to be monitored.)
3. Write by Bash script, It's easy to modify.
4. It can monitor java processes(Such as HBase,Thrift,Hadoop processes)
or other processes(need a little change)
Just copy or uncompress the code to a directory, configured to monitor all
remote hosts, the remote hosts needn't any configure or deploy.
1. modify config.sh to configure parameters
2. check loopcheck.sh to modify paths, because crontab has a limited env
3. check email sending is ok or not, you may need configure MTA such as
4. configure crontab, every 5 minutes check the hosts
0-59/5 * * * * $HOME/smr/loopcheck.sh
May it could help some body.