Le vendredi 22 avril 2005 à 06:37, Randal L. Schwartz écrivait: > >>>>> "Eric" == Eric Wilhelm <[EMAIL PROTECTED]> writes: > > Eric> Just spent way too much time trying to find a bug when it turns out > that > Eric> I just had a full disk. > > I check for errors on close about as often as I check for errors on > print.
I have at least one shell script that uses a Perl one-liner which DOES check for errors in print (which happen when the disk is full) to die early. Since the one-liner basically reads a file and dipatches the lines in different files (based on a date field), there is a small period of time where we have the whole (big) file content duplicated on disk (before the rest of the shell script removes the old file and goes on with its business). The shell script dies when the perl one-liner does, leaving a nice message to be be emailed by the cron daemon. The machine administrator can then make some room and run the shell script again by hand. -- Philippe "BooK" Bruhat Some fair deals are fairer than others. (Moral from Groo The Wanderer #101 (Epic))