http://issues.apache.org/SpamAssassin/show_bug.cgi?id=5313





------- Additional Comments From [EMAIL PROTECTED]  2007-03-21 23:21 -------
(In reply to comment #14)
> This is not a memory resource problem as Daryl C. W. O'Shea keeps on trying to
> suggest.

I was actually looking for evidence of a timing issue.  I'm pretty sure there
are still race conditions in the preforking code.  Of the dozens of times people
have reported problems with running a large number of children they're almost
always using a GB of two of swap so it's sensible to rule out memory issues
first.  One question about how much RAM, and another to confirm if it's enough
(because on systems I've seen with 4GB of RAM and lots of rules and plugins and
user configs it is tight) isn't as if I'm hung up on it being swap thrashing or
OOM problems.


> Memory usage is light, and the problem happens regardless or load.  Even with 
> 60
> concurrent processes, my setup actually happens to reach a disk limitation 
> with
> high amounts of I/O, rather than consuming too much memory and swapping.  This
> *problem* presents itself on any of the FBSD 6.2 machines, while the 4.11
> machine runs without spamassassin needing to be restarted.

That's useful info.  Not surprising at all either.

Now that we've got info about affected versions of FreeBSD Justin or myself can
actually attempt to track this down.



------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

Reply via email to