README_clustjob

keven clust job explain, using md5sum as an example
the md5sum can be replaced by any other tools, but the tools have to follow the interface

clustjob.sh 
1. this is a control script, receive command SET|START|STATUS|STOP|RESULT|REMOVE
2. command explain
   command SET:
    create environment, split filelist and distribute filelist to node list
    Example: ./clustjob.sh -j ./handlemd5.sh -l ./nodelist -f ./filelist -d /root/clustjob/ -c set
    a tokan string  will be generate and print out

   command START:
    start clust job, actually is running nodejob.sh in every node
    Example: ./clustjob.sh -s 0630 -t 0830 -c start -t 2314-23412 (tokan string get from set command)
    0630 means running at 06:30 am and the same as 0830

   command STATUS:
    show the status of clust job
    Example: ./clustjob.sh -c status -t 2314-23412

   command STOP:
    stop clust job, kill process by pid
    Example: ./clustjob.sh -c stop -t 2314-23412

   command RESULT:
    collect and compress the result file, which is cmplist.HOSTNAME, reslist.HOSTNAME, faillist.HOSTNAME
    Example: ./clustjob.sh -c result -t 2314-23412

   command REMOVE:
    remove the work dirs and files
    Example: ./clustjob.sh -c remove -t 2314-23412

nodejob.sh
1. this script will be callled by clustjob.sh and running in every node
2. read path of file from filelist and call handlemd5.sh to execute md5sum
3. the handlemd5.sh will running from starttime to endtime (for this example is 06:30 to 08:30)
4. Example: ./nodejob.sh -s 0630 -e 0830 -j ./handlemd5.sh -t 2314-23412 -d /root/clustjob/2314-23412

handlemd5.sh
1. this script will recieve 2 parameters, filename and outputfile
2. Example: ./handlemd5.sh -f /path/to/kevenfile  -o outputfile