http://issues.apache.org/SpamAssassin/show_bug.cgi?id=4852
[EMAIL PROTECTED] changed:
What |Removed |Added
----------------------------------------------------------------------------
Attachment #3451 is|0 |1
obsolete| |
------- Additional Comments From [EMAIL PROTECTED] 2006-04-05 21:34 -------
Created an attachment (id=3452)
--> (http://issues.apache.org/SpamAssassin/attachment.cgi?id=3452&action=view)
patch
Check to see if $sock is defined too... as much fun as "floating point numbers
with Matt" on the users@ list this morning was, insomnia sucks.
I can't reproduce the bug... it only sometimes happens under high load. I'm
going by what I see in Dan's log extract.
As far as I can tell, the bug is purely cosmetic. The problem is that under
load, Perl takes it's sweat ass time to process the child's SIGCHLD handler
after the child kills itself.
It appears that the socket file handle is immediately closed, but
main_server_poll() can end up trying a ping before the SIGCHLD handler removes
the child's PID info from the hash.
So, it should fix the problem (crap in the logs). It probably doesn't even
need to try and kill the child again, but doing so didn't hurt Dan's spamd
process, so I'm inclined to leave it in just incase $sock gets screwed up for
some other reason.
------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.
