I migrated to a new server last weekend. PHP and MySQL versions are the same, the only differences are the OS (old=RH9, new=FC3) and the /tmp directory is "hardened". There are six WP 2.0.1 installations on the server, two of them are mine.
Since then, each access every "page" in every WP installation generates 36 (count 'em!) zero-length files owned by apache:apache in the /tmp directory. They add up fast, eventually filling the /tmp directory causing anything that needs to create a file there to throw up all sorts of nasty errors.
There are dozens of other PHP/MySQL apps running on the server, and none of them behave this way. Neither did WP on the old server.
All is well after cleaning-out the /tmp directory, a laborious job since there were "too many arguments" for a simple rm tmp* command. Had to go through rm tmpa*, rm tmpb*, etc. for upper and lower case letters and all nine numbers.
Anyway, I've temporarily added a cron job, running every six minutes to clear them out. Not a pretty solution, and who really wants rm running in a cron job? Yikes!
What could be the root cause of this? And what is a potential cure? Can I buy a vowel?