AIX

Hi All
We have a WEB Based application running on the IBM AIX server. There is a EOD Job which runs a UNIX script containing EOD Jobs. Say If any job fails then we have to explicitly comment out the jobs which were successfully executed and then re run the same.

Is there any was by which we can automate it so that it keeps track of the job which is getting executed and say if it failed at some point, when I re run it, it should automatically start from the same point where it got failed?

Kindly help me on this.

Waiting for positive replies

dear prashant,
you can do one thing but its bit tough..
you can add a check condition after each script..
i.e check the return code $?
if its not zero it should write that script name to a new file so that you can re run that new file..

I will elaborate a bit on this. The original script is something like this::::

. ./runEod.sh 6000 $brn_num>> error.txt 2>&1
. ./runEod.sh 6001 $brn_num>> error.txt 2>&1
. ./runEod.sh 6002 $brn_num>> error.txt 2>&1
. ./runEod.sh 1010 $brn_num>> error.txt 2>&1
. ./runEod.sh 1070 $brn_num>> error.txt 2>&1
. ./runEod.sh 1071 $brn_num>> error.txt 2>&1
. ./runEod.sh 6003 $brn_num>> error.txt 2>&1
. ./runEod.sh 6004 $brn_num>> error.txt 2>&1
. ./runEod.sh 6005 $brn_num>> error.txt 2>&1

Now say while running this script job 6003 failed. So at present I have to comment out all the jobs which have run till now and then execute the same. For eg in the above case the script will be as follows:::

# . ./runEod.sh 6000 $brn_num>> error.txt 2>&1
# . ./runEod.sh 6001 $brn_num>> error.txt 2>&1
# . ./runEod.sh 6002 $brn_num>> error.txt 2>&1
# . ./runEod.sh 1010 $brn_num>> error.txt 2>&1
# . ./runEod.sh 1070 $brn_num>> error.txt 2>&1
# . ./runEod.sh 1071 $brn_num>> error.txt 2>&1
. ./runEod.sh 6003 $brn_num>> error.txt 2>&1
. ./runEod.sh 6004 $brn_num>> error.txt 2>&1
. ./runEod.sh 6005 $brn_num>> error.txt 2>&1


Now I don't want this manual intervention of commenting out the previously executed Jobs. The script should run and if it fails it should start from the same point (i.e. 6003 in this case).

then can't you grep the failed jobs from your error.txt??
i mean RC..

What if you create a file including the list of numbers:
batch_num.lst
6000
6001
6002
1010
...

Then your execution script could do something like the following (in sort of code - as I have only had one cup of coffee this morning!)

Code:

while read bn
  do
  ./runEod.sh $bn $brn_num>> error.txt 2>&1
  if [ $# -eq 0 ]   #ok execution
     then
     sed 's/$bn/#$bn/' batch_num.lst > batch_num.rev
     cp batch_num.rev batch_num.lst
  fi
done < batch_num.lst

You could perhaps do the same kind of logic with the whole command lines as you provided, but still need to do error-checking after each cycle to determine if you should comment-out the successful execution.

its looks fine....
But if you can explain the script a bit more?
I am a bit new to UNIX ...:-)

Depends on shell type you can redefine prompt also, e.g. in ksh
typeset PS4='scriptName [$LINENO]'
set -x
..
someCode
...