http://bugzilla.spamassassin.org/show_bug.cgi?id=4310





------- Additional Comments From [EMAIL PROTECTED]  2005-06-06 16:48 -------
fwiw, Daniel had another bug which I don't think was bugzilla'd, but it was
related to spamd hangs with the new prefork code; so I'll comment here since
I've just fixed it, I think.

The symptom is that if you strace the master spamd, it's in a loop of killing a
child process repeatedly.

There was a race condition, in that if the lowest idle child died or exited
while still in the I state, the master spamd would still attempt to assign it
tasks and get stuck in an infinite loop trying to write the 'accept' order to
the dead child.  This is now avoided by recomputing the lowest idle child in the
signal handler.

r185098 has that fix.



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

Reply via email to