The mailing list for listmasters using Sympa

[Micah Anderson <address@concealed>]

* Patrick von der Hagen <address@concealed> [2008-10-29 10:45-0400]:
> Micah Anderson schrieb:
> [....]
> > I have increased sympa.conf's maxsmtp 200 from 100 (as well as
> > postfix's smtp process limits), but that didn't change anything in the
> > times.
> "distribution_mode fork" might help a little bit, if not already set....
> The major question is: postfix or sympa?

Good question, thanks for pointing that out.

> For example, while the MTA should probably scan all e-mails from the
> outside sent to sympa for spam or viruses, it should not scan messages
> leaving sympa (since they have already been scanned on the way in.

Right, my postfix is not doing this. Only mail coming from the outside
is scanned, as we are using a before-queue SMTP milter
configuration. In any case, I may be misunderstanding, but isn't Sympa
is just handing things off using the value in sympa.conf for 'sendmail'?
Ours is set to this:

sendmail /usr/sbin/sendmail

> I'm no postfix-guy, but my MTA differentiates between in and out....
> You already mentioned postfix-process-limits.... can you verify during
> distribution if those limits are really used of if there is just not
> enough sympa-input?

Good point, I'll begin monitoring this information to see if I can get a
more accurate number for you.

> Postfix includes tools to perform performance-testing (smpt-source,
> smpt-sink), perhaps you can send some hundred test-messages to a
> test-recipient? smtp-source uses smtp and would be less efficient than
> calling /usr/lib/sendmail, but it should give you an idea wheter sympa
> or postfix have to be tuned.

Its hard for me to devise a test case with smtp-source/smtp-sink that
would accurately mimic the situation that I am having with sympa as the
setup of the two are pretty different and I am not sure what options
would best represent the workload that sympa presents, any ideas of the
best way to do this?

> (I don't claim that it is not a sympa-problem. But I could give less
> advise in that case. ;-))

This is a good point, and I appreciate you bringing it up, its always
good to measure the different related pieces that are involved, as it
could very well be not a sympa-related issue at all.

micah