Re: [cobalt-users] Pop3 Server needs to be reset each morning

[Larry Smith <lesmith@xxxxxxxxx>]

On Tuesday 18 November 2003 08:27, Jeremy Towne wrote:
> Does anyone know the original/default nowait is?
>
> Jeremy Towne
>
>
> ---------- Original Message -----------
> From: Larry Smith <lesmith@xxxxxxxxx>
> To: cobalt-users@xxxxxxxxxxxxxxx
> Sent: Tue, 18 Nov 2003 07:59:09 -0600
> Subject: Re: [cobalt-users] Pop3 Server needs to be reset each morning
>
> > On Tuesday 18 November 2003 07:04, Jason Gottschalk wrote:
> > > Hello cobalt-users,
> > >
> > >   Every morning, just when I guess many of our customers start showing
> > >   up to work, I have to stop and start the e-mail services on my RAQ4i
> > >   because nobody can get their mail.
> > >
> > >   If I watch the maillog file with tail -f, I can see a steady stream
> > >   of sendmail activity but no pop3 activity.
> > >
> > >   As soon as I stop and restart the email services from the control
> > >   panel, I start seeing all of the pop3 logins and everything starts
> > >   working fine.
> > >
> > >   Do I have a configuration problem? or too many users?  There are
> > >   about 200 users getting e-mail from this server.
> > >
> > >
> > >   Thanks........
> >
> > Sounds like you have "too much" activity for that service in a short
> > period of time.  This has been discussed before.  All Linux kernels
> > are designed to "shutdown" processes that are showing "abnormal"
> > activity (just in case there is a problem).  the answer is to modify
> > the /etc/inetd.conf line for your mail (pop3) to allow more processes:
> >
> > Original:
> > pop-3   stream  tcp     nowait  root    /usr/sbin/tcpd  in.qpopper -R
> >
> > Modified:
> > pop-3   stream  tcp     nowait.500  root    /usr/sbin/tcpd
> >  in.qpopper -R
> >
> > The dot (period) 500 says allow 500 processes per minute before
> > shutting down the service.  Adjust this number with CAUTION since it
> > can affect your server if you allow any one process to have too much
> > time/space.
> >

Don't remember exactly but think it is around 200 for the "default".  You 
might want to start with 300, then if the problem still exists, up to 400 and 
so forth but watch your process stats....

Also NOTE, you must "restart" the inetd process for this change (any change to 
inetd.conf) to take affect (or reboot)...

/etc/rc.d/init.d/inet restart  (I believe)


-- 
Larry Smith
SysAd ECSIS.NET
sysad@xxxxxxxxx