[Mimedefang] Timeout for filter_tick
Jeff Rife
mimedefang at nabs.net
Fri Apr 13 22:46:27 EDT 2007
In the man page for mimedefang-multiplexor, it says the following:
-b busyTime
The longest a Perl process is allowed to spend scanning an e-mail
before
it is declared hung up and killed. The default is 120 seconds.
Does this timeout also apply to calls to "filter_tick"? It appears to
from this call inside handleSlaveReceivedTick, but I want to be sure:
s->event = EventTcp_ReadBuf(es, s->slaveStdout, MAX_CMD_LEN, '\n',
handleSlaveReceivedAnswerFromTick,
Settings.busyTimeout, 1, s);
If it uses the same timeout as it appears to, I'd like to submit an RFE
to add another flag to make this timeout different. Set it the same if
the new flag isn't there, but allow it to be different.
My reasoning is that you want to limit e-mail scanning time because you
want mail to flow through, but filter_tick is only called on an idle
slave, and there isn't any mail scanning going on. I have a couple of
ideas for things to do there, but they might take longer than I want a
mail scan to take.
What I'm doing right now is optimistically generating 26-bit Hashcash
values for recipients to which we have sent e-mail often and recently.
It usually only takes about 30 seconds, but sometimes takes over 2
minutes, so I've had to set the "busyTime" to a lot higher than I'd
like for regular e-mail scanning.
I know I could do this in a separate process running as a daemon or
cron job, but I'd just end up duplicating a lot of the code, and would
have to change things in too many places to keep business rules the
same.
--
Jeff Rife |
| http://www.nabs.net/Cartoons/Dilbert/NoHelpDesk.jpg
More information about the MIMEDefang
mailing list