On Wed, Sep 20, 2000 at 11:02:28AM +0200, Christoffer Hall-Frederiksen wrote:
> I am currently running with a concurrencylocal of 40. But if problems
> arise with a mailbox (eg. no more quota) peoples procmail processes
> hang for a long time and ties up all the local processes. I can soulve
It's actually a tricky problem as a user can hog all of the concurrencylocal
by having a .qmail file with "| sleep 100000" as a delivery instruction. Your
procmail problem is exhibiting the same effect.
> this partially with /etc/procmailrc but thats not really a solution.
> Will qmail handle a pr. user process-limit (set by the OS) of eg 5 or
> 10 gracefully or are there any pitfalls or bette ways to limit the
> problem?
That's about the best solution if it's one particular user who's causing
the problem. qmail will simply attempt to re-deliver using it's usual
delivery schedule.
You have correctly pointed out though that solving the problem in procmail
is a better long-term solution.
Does procmail normally just wait and retry if there is no quota?
If you wanted to, you can probably get procmail to generate a bounce message
saying "over quota", by judicious use of the exit codes recognized in
qmail-local processing.
Regards.