[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [cobalt-developers] Replacing sendmail
- Subject: Re: [cobalt-developers] Replacing sendmail
- From: Jeff Lasman <jblists@xxxxxxxxxxxxx>
- Date: Tue Aug 1 07:34:36 2000
- Organization: nobaloney.net
"Robert G. Fisher" wrote:
> > but the question is how can I estimate how many emails per hour equate
> > to 12 or more daemons running. And another question is how high can I
> > safely raise this?
>
> Depends on how many people you have connecting to the mail server.
> For my case when I was debugging this, I actually have mine set to
> 120 for our mail server -- but this is a machine that literally
> hundreds of users access consistently as their SMTP relay as well
> as accepting SMTP connections for a few hundred domains. This
> high a number can be a problem on some machines that don't handle
> the higher number of file handles properly -- latest kernels shouldn't
> have a problem, older ones could run into some poll() glitches.
There's the rub... "shouldn't" have the problem. This is the sort of
answer I'd like from Cobalt.
> > I think it's perfectly fair for Cobalt to give me some numbers here.
> > Can you? I've already got a tech support call in for second level, and
> > it's been a week. Can someone give me an answer here?
>
> Honestly Jeff, it's not really fair to expect anyone to be able to
> give you an exact number --
Not an exact number, just what could be considered safe, in light of
your previous paragraph, for example.
> it's a matter of trial and error and
> judgement calls made by the SA in cases like this. There are a
> lot of other variables to be pulled into the equation -- However,
> if you've found that throttling SMTP connections back to 500
> per hour solves the problem, perhaps that's good enough for now.
Unfortunately it hasn't worked.
> I know we've covered a series of questions on the list already;
> realistically, those are some of the questions that need to be
> answered before your current answer can be found.
I've already answered all questions privately, and forwarded out
megabytes of logfiles to those who wanted to see them. What it all
boils down to is that we're getting a very high number of daemons
running per hour and we need to either allow 'em or expect sendmail to
shut down.
That's why I've asked those few questions above.
Yes, I'm willing to experiment, but are 40,000 people really willing to
get the same email over and over again so I can? I doubt it. I'm also
willing to pay Cobalt their regular hourly rate for some ballpark
answers and some help, and yes, I've told them that.
I think you're confusing me with someone who has no clue when it comes
to email, and that's not the case; I've been running list-servers for
some time. But they're fine-tuned as list-servers, and this is a Cobalt
RaQ3; fine tuned for (as several respondents have pointed out) something
completely different.
I have a meeting with the owner of the RaQ3 in question scheduled for
Thursday. We may very well move email from his RaQ to one of my
list-servers, or to a dedicated Linux computer. But that's not what he
wants; he really wants to use his RaQ to serve everything he needs for
the few (less than ten) domains he's got running on it.
Thanks for your help and for your input.
Jeff
--
Jeff Lasman <jblists@xxxxxxxxxxxxx>
nobaloney.net
P. O. Box 52672
Riverside, CA 92517
voice: (909) 787-8589 * fax: (909) 782-0205