Shell Programming and Scripting

Hi ,

I have a scenario, i have a directory where i receive around 14-15 files at a interval of 20-40 min not fixed, i want to write a unix scripts which invoke sqlldr command to load files into oracle automatically as soon as the file hit the directory.

Any help will be appreciated.

Thanks

Rajesh

Code:

find $filepath -maxdepth 1 -type f  | while read file
do
sqlldr $user/$password@dbname control=$controlfile bad=$badfilepath/${file}_bad log=$logfilepath/${file}_log
wait
mv $filepath/$file $archive
done

If you have any specific naming format for data file this code can be trimmed. archive,logfilepath,badfilepath are directories, The reason I am using them is to avoid "find" feeding log/bad files to sqlldr.

Hi

My directory is fixed where the file will come, sqllder is written in a scripts which is in another directory. Now the requirment is that as soon as the file is reached to the directory there should be some means which sense that a file has come and it will trigger the shell scripts containg sqlldr scripts.

Thanks for understanding.

Rajesh

in that case use a cron to run this script every X mins. Assuming name of this script is kept fileuploader.sh

Code:

if ! ps -ef | grep fileuploader.sh | grep -v grep; then
    find $filepath -maxdepth 1 -type f  | while read file
    do
        sh sqlldr_script.sh $file
        wait
    done
fi

File transfers are not instantaneous and data loads are not instantaneous.
This sort of process needs very careful design to avoid loading the same file twice or loading a part-transferred file. Common designs involves transferring files under a temporary name, then renaming them after successful transfer. Similarly loading files, then renaming them after a successful data load. Another design involves using a "interlock" files to mark the common directory as "busy" until the transfers complete and "busy" during the data load.
Do you have such precautions in place?