Shell Programming and Scripting

Hi guys,

I am new guy for unix. I need help writing the following shell script.
My requirement is:

I have few pl/sql procedures(lets say 10). I want to run some of them (say 5) parallally in the background and once these are completed trigger the next ones one after other.

Could some one give some ideas on how to keep track of the first 5 background processes and how it can be implemented.(some sample script would be appreciated.)

Thanks
Chandu

Welcome, this seems to be a homework question. You should post what you have worked out already. If this is a homework question, please read the Unix.com forum's Rules.

This may be a legitimate post. Create a temp table in your database, have each procedure post back to the table. This is how our project tends to do these sorts of things. Table is small and has columns such as: job ID, current status (in progress, success, failure) and a date/time column of the last update.Drop the table if it is not needed after your run.

This is actually our project requirement. I described here it as a small assignment, but actually we have about 150 procedures and some of them can be run parallally and some of them have dependency. Our objective is to write a shell script to schedule this task. I have no idea about how to start with and request your ideas/suggestions.

Google: could you be more specific. When the first set of procedures triggered in the background how do we monitor for its status and trigger the next dependent job when it is done. I mean do we need to use some kind script which keeps on checking this job status table in specific intervals?

thanks for your suggestions in advance.

We have a couple of tables set up:a job dependency table (holds job id and jod dependency ID among other things) and aother is a job master table (table containing all jobs and job ids) and finally a job status table. Shell scripts control the execution of stored procedures. Before a procedure is executed the shell runs a small sql statement to check the job status table to see if the current job is able to run. If the last status is success, and the dependency is also met everything is good. The controlling shell also updates the job status table based upon the status of the PL/SQL it kicked off (in progress, success, failure). this works well for us but we dont have 150 procedures to run in parallel either

You may not need this but your first post mentions you are new to UNIX, disregard if you already know this. Here is a link that shows how you execute a stored procedure from a shell and capture output variables.

Once you follow google's tips you will have the individual jobs. Now let's say that you want job1, job2, and job3 to run in parallel followed by job4, job5, and job6 running in parallel. You could write a script:
#! /usr/bin/ksh
/path/to/job1 &
/path/to/job2 &
/path/to/job3 &
wait
/path/to/job4 &
/path/to/job5 &
/path/to/job6 &
wait
exit 0

The ampersands tell ksh to background the jobs. The "wait" without any parameters says to wait for all background jobs.