[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: Sat Jul 29 19:19:20 2000
- Organization: nobaloney.net
"Robert G. Fisher" wrote:
> On Tue, Jul 25, 2000 at 11:49:05PM -0700, Jeff Lasman wrote:
> > > As for giving you an answer, you really haven't given nearly enough
> > > information about what the current condition is when sendmail goes
> > > to rejecting connections.
> >
> > Actually, I have, but not on this list. I posted top, ps-aux and mqueue
> > info on several other lists, and have yet to get a definitive answer; I
> > really didn't think this was the right list for the discussion; I only
> > mentioned it in passing.
>
> Actually, that's still not enough information. You're only giving
> snapshots, to track the problem -- you need to be a little more active.
> top is only good for tracking what's using alot of memory or CPU
> -- so long as someone's watching the terminal.
>
> What is the full 'command line' option for sendmail when this happens
> (try ps auxw ;)?
This was now some time ago, so I don't have an answer. But presuming
you're talking about the instance of sendmail that should be accepting
connections, it's whatever the RaQ3 starts it with.
> Is there only one instance of sendmail being run when this happens?
There's only the one entry that's accepting connections; that was
started by the RaQ at startup, in other words.
There are plenty of instances sending email; as I've said before in
various places (and no, I'm not blaming you for not reading all those
places, only explaining), I really think that's the problem.
> What entries, if any are being placed in the log file?
None at all.
> If you do a netstat -an, how many connections do you see either
> in the ESTABLISHED or FIN_WAIT state to port 25?
Should it happen again, and should I be notified immediately after it
happens (usually there's a delay of an hour or more <frown>), I'll run
netstat and find out.
> If you do 'nc <box> 25' (or use telnet if netcat isn't installed) what
> response do you get?
Using telnet I get a connection refused; I don't remember the details; I
only did this the first time the problem existed.
> Often following those steps will give you information on where
> it's failing.
If it should happen again, and should I be notified early enough in the
cycle, maybe I can get better information.
I'm beginning to think (having read various replies from various lists)
that sendmail will shut down accepting connections if it's sending too
much email <frown>.
That's why the 500/hour limitation we've installed. So far I've
received no complaints since the client set the throttle.
> Again, how does your mailing list software handle it. If it is
> connecting to SMTP rather than calling sendmail directly it could
> easily cause this on it's own.
I really don't remember which it's doing, but I do know it's switchable;
which should I set it for? Calling sendmail? I think so, but what's
your input? I do want to make sure it's set optimally.
> > what the client is willing to pay for.
>
> Gotcha, no consulting for free egh -- not even to satisfy a geeks
> share of curiosity? :).
Depends how deep I want to go, how deep the customer wants me to go, and
how little else I have to at the moment.
Jeff
--
Jeff Lasman <jblists@xxxxxxxxxxxxx>
nobaloney.net
P. O. Box 52672
Riverside, CA 92517
voice: (909) 787-8589 * fax: (909) 782-0205