scratch disk space full

[peep]

we have scratch of 100 gb and the 3 nodes. The space is not allocated so it can use whole 100 gb.
But jobs are aborting and its not showing any space errors in unix.

but in job logs there are space outage errors.
Are there any settings in administrator client that are restricting the jobs to use certain space?

Do I need to change any values in environment variables ?

[nagarjuna]

Once scratch disk filled up job will abort & cleanup all the temp files created during that run .

[peep]

Nag I tried to send u private message the other day.but I could not .

any ways

while running the jobs the scratch was filling totally and once its done the scratch space is free.

will that b okay to delete temp files?
how can i delete ?

[peep]

what could be the reason and what can be a solution ?

[ArndW]

Scratch is there to be used - how much data are you processing in parallel and are you doing things like multiple sorts or several joins on different keys (which imply sorts)? Are many jobs running in parallel?