Unless you get a bare linefeed. At which point you need to find the
offending smtp connection and kill it.
I average about one "broken?" MTA or two a week. Causes logfiles to swell
and general performance problems.
Paul Farber
Farber Technology
[EMAIL PROTECTED]
Ph 570-628-5303
Fax 570-628-5545
On Tue, 16 Nov 1999, Eric Dahnke wrote:
>
> Hi,
>
> I manage a qmail server which serves about 5000 pop users. We do it on a
> PII, with about 16G of disk space. A PII, is plenty of machine, but
> would have added more disk space if I had to do it over again. That
> server does about 20,000 msgs per day, and load average rarely breaks 1.
>
> I think Exchange is probably as fast as qmail (if you throw a few more
> resources at it), but it needs constant attention. Qmail needs no
> attention. That's the thing.
>
>
> - Eric
>
>
>
> Peter Green escribi�:
> >
> > On Tue, Nov 16, 1999 at 12:56:12PM +0100, Marthe Nes�en Gangfl�t wrote:
> > > Hi people,
> > >
> > > I need benchmarks and such to show my employer that Linux can rock mail
> > > better than NT Exchange (dohh). We use RedHat Linux and qmail for this,
> > > because that's where we have knowledge. This is somehing we might sell to
> > > customers of ours, and it's about 2000-5000 pop-users.. Can qmail manage
> > > that much on ONE single server?
> >
> > I can't speak for Exchange; nor can I speak for pop-users specifically.
> > However, I have no qualms about qmail scaling that high. On our network, we
> > have two qmail machines: a main relay server and an ezmlm ml server. Neither
> > machine delivers anything local, so all deliveries are "remote" (even if on
> > the same network).
> >
> > Here is the output from `zcat qmail-19991115.gz|localtai|matchup|zoverall`:
> >
> > [ relay server ]
> > Completed messages: 50939
> > Recipients for completed messages: 191315
> > Total delivery attempts for completed messages: 200494
> > Average delivery attempts per completed message: 3.93596
> > Bytes in completed messages: 339800621
> > Bytes weighted by success: 917491749
> > Average message qtime (s): 180.223
> >
> > Total delivery attempts: 224784
> > success: 197167
> > failure: 7485
> > deferral: 20132
> > Total ddelay (s): 17580934.661599
> > Average ddelay per success (s): 89.167734
> > Total xdelay (s): 1606541.947916
> > Average xdelay per delivery attempt (s): 7.147048
> > Time span (days): 0.993594
> > Average concurrency: 18.7141
> >
> > [ ezmlm server ]
> > Completed messages: 4786
> > Recipients for completed messages: 5319
> > Total delivery attempts for completed messages: 5415
> > Average delivery attempts per completed message: 1.13142
> > Bytes in completed messages: 26064596
> > Bytes weighted by success: 27721323
> > Average message qtime (s): 112.261
> >
> > Total delivery attempts: 74849
> > success: 62120
> > failure: 1960
> > deferral: 10769
> > Total ddelay (s): 19233264.798056
> > Average ddelay per success (s): 309.614694
> > Total xdelay (s): 661400.438022
> > Average xdelay per delivery attempt (s): 8.836463
> > Time span (days): 0.983027
> > Average concurrency: 7.78727
> >
> > The hardware/software for each:
> >
> > [ relay ]
> > AMD K6-2 333
> > 384MB RAM
> > UW SCSI disk on Buslogic controller
> > True tulip 100bTX NIC on 3com 10/100 switch, full-dup
> > RH6 w/ 2.2.13 kernel
> > qmail-1.03 + jbuce.diff + newlines.patch
> >
> > [ ezmlm ]
> > Dual P90 EISA
> > 128MB RAM
> > FW SCSI disk on Adaptec 2940
> > 3com 3c509 10bT on 3com 10/100 switch
> > RH6 w/ 2.2.10 SMP kernel
> > qmail-1.03 + jbuce.diff + newlines.patch
> > ezmlm+idx-0.322
> >
> > The relay machine is also our primary name server; the ezmlm machine (made
> > entirely out of "junk" parts) also serves a little news with INN. I'd say a
> > couple of thousand POP accounts are not totally unheard of... :)
> >
> > /pg
> > --
> > Peter Green
> > Gospel Communications Network, SysAdmin
> > [EMAIL PROTECTED]
>
