The qpopper list archive ending on 28 Jun 2000


Topics covered in this issue include:

  1. Compiled fine but one more prob
       Super-User <root at server-icaweb-org.icaweb dot org>
       Wed, 21 Jun 2000 11:39:28 -0500 (CDT)
  2. Re: Compiled fine but one more prob
       Admin Mailing Lists <mlist at intergrafix dot net>
       Wed, 21 Jun 2000 12:51:17 -0400 (EDT)
  3. Re: Compiled fine but one more prob
       "Mitch Vincent" <mitch at venux dot net>
       Wed, 21 Jun 2000 12:59:43 -0400
  4. MTA error: relaying denied
       "Kenneth Porter" <shiva at well dot com>
       Wed, 21 Jun 2000 11:38:53 -0700
  5. Re: qpopper compile on Solaris 8
       Stephen Samuel <samuel at bcgreen dot com>
       Wed, 21 Jun 2000 12:44:35 -0700
  6. Too load !
       Rodrigo Luiz Anami <rodrigoa at bestway.com dot br>
       Wed, 21 Jun 2000 18:39:51 -0300
  7. Re: Too load !
       "Joseph W. Breu" <breu at cfu dot net>
       Wed, 21 Jun 2000 16:59:23 -0500 (CDT)
  8. Re: Too load !
       "Mitch Vincent" <mitch at venux dot net>
       Wed, 21 Jun 2000 18:04:19 -0400
  9. Re: Too load !
       Fergal Daly <fergal at esatclear dot ie>
       Wed, 21 Jun 2000 23:16:38 +0100
 10. Re: Too load ! (fwd)
       Vinay Bharel <vinay at bharel dot com>
       Wed, 21 Jun 2000 22:33:05 +0000 (GMT)
 11. POP3 ERR
       "Dave Linden" <davel at west-teq dot net>
       Wed, 21 Jun 2000 16:41:15 -0600
 12. Re: Too load !
       Jorge Kobeh <jkobeh at spin.com dot mx>
       Wed, 21 Jun 2000 18:04:22 -0500 (CDT)
 13. Re: Too load ! (fwd)
       Peter Evans <peter at gol dot com>
       Thu, 22 Jun 2000 08:23:31 +0900
 14. Re: Too load ! (fwd)
       Vinay Bharel <vinay at bharel dot com>
       Wed, 21 Jun 2000 23:49:59 +0000 (GMT)
 15. Qpopper 3.1b3 available
       Qpopper Support <qpopper at qualcomm dot com>
       Wed, 21 Jun 2000 18:29:30 -0700
 16. Re: Too load !
       Alan Brown <alan at manawatu.gen dot nz>
       Thu, 22 Jun 2000 21:37:56 +1200 (NZST)
 17. Re: How can I make qpopper create dot-lock file in different directory
       Jason Englander <jason at interl dot net>
       Thu, 22 Jun 2000 06:39:46 -0500 (CDT)
 18. Re: enable-hash-spool=22
       Jason Englander <jason at interl dot net>
       Thu, 22 Jun 2000 07:08:50 -0500 (CDT)
 19. Re: Too load !
       "Keith Warno" <keith at HaggleWare dot com>
       Thu, 22 Jun 2000 14:21:17 -0400
 20. ANN: New Red Hat Linux kernel addresses security vulnerability
       "Kenneth Porter" <shiva at well dot com>
       Thu, 22 Jun 2000 11:30:17 -0700
 21. 3.1b3 buglet?
       Jason Englander <jason at interl dot net>
       Thu, 22 Jun 2000 13:55:34 -0500 (CDT)
 22. Re: 3.1b3 buglet?
       Steven Champeon <schampeo at hesketh dot com>
       Thu, 22 Jun 2000 15:21:31 -0400
 23. Re: Too load !
       Alan Brown <alan at manawatu.gen dot nz>
       Fri, 23 Jun 2000 07:37:43 +1200 (NZST)
 24. Re: Too load ! (fwd)
       Qpopper Support <qpopper at qualcomm dot com>
       Sun, 25 Jun 2000 14:26:42 -0700
 25. Re: 3.1b3 buglet?
       Jason Englander <jason at interl dot net>
       Sun, 25 Jun 2000 18:39:18 -0500 (CDT)
 26. messages preceeded by ! any danger?
       "ganizani phiri" <ganizani at malawi dot net>
       Mon, 26 Jun 2000 08:28:37 +0300
 27. Procmail
       Rodrigo Luiz Anami <rodrigoa at bestway.com dot br>
       Mon, 26 Jun 2000 10:28:47 -0300
 28. Re: Procmail
       Eric Luyten <Eric.Luyten at vub.ac dot be>
       Mon, 26 Jun 2000 15:43:56 +0200 (MET DST)
 29. Re: Procmail
       Rodrigo Luiz Anami <rodrigoa at bestway.com dot br>
       Mon, 26 Jun 2000 10:58:25 -0300
 30. compile error: undefined symbol __inet_ntoa
       David Schlussel <david at bestoffer dot com>
       Mon, 26 Jun 2000 17:10:28 -0700 (PDT)
 31. Re: compile error: undefined symbol __inet_ntoa
       "Kenneth Porter" <shiva at well dot com>
       Tue, 27 Jun 2000 10:06:12 -0700
 32. Re: compile error: undefined symbol __inet_ntoa
       David Schlussel <david at bestoffer dot com>
       Tue, 27 Jun 2000 10:11:12 -0700 (PDT)
 33. Re: compile error: undefined symbol __inet_ntoa
       "Kenneth Porter" <shiva at well dot com>
       Tue, 27 Jun 2000 10:38:09 -0700
 34. Re: compile error: undefined symbol __inet_ntoa
       David Schlussel <david at bestoffer dot com>
       Tue, 27 Jun 2000 11:30:12 -0700 (PDT)
 35. Re: compile error: undefined symbol __inet_ntoa
       "Kenneth Porter" <shiva at well dot com>
       Tue, 27 Jun 2000 12:01:12 -0700
 36. Re: compile error: undefined symbol __inet_ntoa
       David Schlussel <david at bestoffer dot com>
       Tue, 27 Jun 2000 11:59:56 -0700 (PDT)
 37. How to upgrade properly
       Linda Pelleu Antil <linda.pelleu at nist dot gov>
       Tue, 27 Jun 2000 16:07:03 -0400
 38. Re: How to upgrade properly 
       Ken Hornstein <kenh at cmf.nrl.navy dot mil>
       Tue, 27 Jun 2000 16:11:03 -0400
 39. Re: How to upgrade properly 
       Linda Pelleu Antil <linda.pelleu at nist dot gov>
       Tue, 27 Jun 2000 16:28:23 -0400
 40. Re: How to upgrade properly 
       Ken Hornstein <kenh at cmf.nrl.navy dot mil>
       Tue, 27 Jun 2000 16:29:33 -0400
 41. hello all!
       rek2 <rek2 at linuxbusca dot com>
       Wed, 28 Jun 2000 00:50:03 -0400
 42. 
       "jmcintyr" <jmcintyr at i-plus dot net>
       Wed, 28 Jun 2000 09:08:15 -0400
 43. qpopper 3.0.2 does compile on Solaris 2.3?
       "Jack Barnett" <jbarnett at axil.netmate dot com>
       Wed, 28 Jun 2000 09:15:49 -0500
 44. trouble with latest beta
       "J.D. Bronson" <jdb at xpec dot com>
       Wed, 28 Jun 2000 09:29:16 -0500
 45. AUTHFILE in 3.0.2 on Solaris2.6
       "Yvonne.Hodson" <yhodso01 at bcuc.ac dot uk>
       Wed, 28 Jun 2000 15:54:38 +0100 (BST)
 46. -1 hidden messages.
       Ayan George <ayan at kiwi.ayan dot net>
       Wed, 28 Jun 2000 11:00:15 -0400 (EDT)
 47. dot-lock file not deleted
       Alexis Albinet <albinet at fluxus dot net>
       Wed, 28 Jun 2000 16:54:28 +0200 (CEST)
 48. connection refused 
       Andrew Brent <dbrent at linkexchange dot com>
       Wed, 28 Jun 2000 11:38:09 -0400
 49. RE: -1 hidden messages.
       Joseph Peterson <jay at qtm dot net>
       Wed, 28 Jun 2000 11:49:26 -0400
 50. Re: connection refused
       Ayan George <ayan at kiwi.ayan dot net>
       Wed, 28 Jun 2000 12:17:54 -0400 (EDT)

Date: Wed, 21 Jun 2000 11:39:28 -0500 (CDT)
From: Super-User <root at server-icaweb-org.icaweb dot org>
Subject: Compiled fine but one more prob

Thanks for all of the suggestions to help me get up and running.  I appreciate 
it.  However, I have one more issue.

I can retrieve mail sent to the mailbox on my server BUT

I can't send under my server.  What I mean to say is whenever I try to send an 
email from Outlook Express it says

A message in your outbox could not be sent.  Server did not recognise the 
recipients.  Explanation: <user at foo dot com>...relaying denied Error:5550

Any ideas what I did wrong here?

TIA,
Greg Strange


Date: Wed, 21 Jun 2000 12:51:17 -0400 (EDT)
From: Admin Mailing Lists <mlist at intergrafix dot net>
Subject: Re: Compiled fine but one more prob

this is an issue with your MTA (sendmail, postfix, qmail, exim)
not qpopper.

most likely you're not giving that client IP access to send mail through
your mail server to non-local addresses.
Make sure all your local ips are defined in your MTA so they can send
through.
If the client IP is not local to you (i.e. a roaming user) you'll
have to either implement SMTP AUTH in your MTA (if it supports it) or
implement a pop-before-smtp solution (such as DRAC).

-Cygnus
.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-.
Anthony J. Biacco                       Network Administrator/Engineer
thelittleprince at asteroid-b612 dot org       Intergrafix Internet Services

    "Dream as if you'll live forever, live as if you'll die today"
http://www.asteroid-b612.org                http://www.intergrafix.net
.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-._.-.

On Wed, 21 Jun 2000, Super-User wrote:

> Thanks for all of the suggestions to help me get up and running.  I appreciate 
> it.  However, I have one more issue.
> 
> I can retrieve mail sent to the mailbox on my server BUT
> 
> I can't send under my server.  What I mean to say is whenever I try to send an 
> email from Outlook Express it says
> 
> A message in your outbox could not be sent.  Server did not recognise the 
> recipients.  Explanation: <user at foo dot com>...relaying denied Error:5550
> 
> Any ideas what I did wrong here?
> 
> TIA,
> Greg Strange
> 
> 


From: "Mitch Vincent" <mitch at venux dot net>
Subject: Re: Compiled fine but one more prob
Date: Wed, 21 Jun 2000 12:59:43 -0400

That's not qpopper, qpopper is only a POP3 server, you need to look at your
SMTP server (sendmail, postfix, exim, qmail etc) configuration and allow
relaying from selected IP addresses/domains/whatever.

-Mitch
----- Original Message -----
From: "Super-User" <root at server-icaweb-org.icaweb dot org>
To: "Subscribers of Qpopper" <qpopper at lists.pensive dot org>
Sent: Wednesday, June 21, 2000 12:39 PM
Subject: Compiled fine but one more prob


> Thanks for all of the suggestions to help me get up and running.  I
appreciate
> it.  However, I have one more issue.
>
> I can retrieve mail sent to the mailbox on my server BUT
>
> I can't send under my server.  What I mean to say is whenever I try to
send an
> email from Outlook Express it says
>
> A message in your outbox could not be sent.  Server did not recognise the
> recipients.  Explanation: <user at foo dot com>...relaying denied Error:5550
>
> Any ideas what I did wrong here?
>
> TIA,
> Greg Strange
>
>


From: "Kenneth Porter" <shiva at well dot com>
Date: Wed, 21 Jun 2000 11:38:53 -0700
Subject: MTA error: relaying denied

On Wed, 21 Jun 2000 11:39:28 -0500 (CDT), Super-User wrote:

>I can't send under my server.  What I mean to say is whenever I try to send an 
>email from Outlook Express it says
>A message in your outbox could not be sent.  Server did not recognise the 
>recipients.  Explanation: <user at foo dot com>...relaying denied Error:5550

http://www.sendmail.org/tips/relaying.html

(I'm assuming you're using sendmail. Further sendmail questions should
go to the newsgroup comp.mail.sendmail. And send any qpopper questions
from there to this list. ;-)

Ken
mailto:shiva at well dot com
http://www.sewingwitch.com/ken/
http://www.harrybrowne2000.org/



Date: Wed, 21 Jun 2000 12:44:35 -0700
From: Stephen Samuel <samuel at bcgreen dot com>
Subject: Re: qpopper compile on Solaris 8

Try adding /usr/ccs/bin and /usr/xpg4/bin to your path:

PATH=$PATH:/usr/ccs/bin:/usr/xpg4/bin  

If you've already got cc/gcc installed, that should help.

You might also try adding /usr/local/bin if that's where you've 
installed gcc.

Super-User wrote:
> 
> Hello everyone.
> 
> I am new to mail administration and UNIX in general.  I am trying to
> compile and run qpopper.  I am Solaris 8 on Sparcstation.
> 
> Whenever I run ./configure, I get
> 
> creating cache ./config.cache
> checking whether make sets ${MAKE}... ./configure: make: not found
> no
> checking for gcc... no
> checking for cc... no
> configure: error: no acceptable cc found in $PATH

-- 
Stephen Samuel +1(604)876-0426                samuel at bcgreen dot com
		   http://www.bcgreen.com/~samuel/
The question, for this world, is not "will I die?". 
It is, rather, "how will I live?".

Date: Wed, 21 Jun 2000 18:39:51 -0300
From: Rodrigo Luiz Anami <rodrigoa at bestway.com dot br>
Subject: Too load !

Hi there,

We are using qpopper 3.0.2 and sendmail 8.10.2 on a Slackware linux 
machine. Our customers totalizes about 6,000 users. We see too much load on=
 
our server all day, about 2 and 3. And the server has been present some 
peaks of 16 ! The server is an IBM Net Finity 5000, with one Pentium III 
600 Mhz and 512 MB RAM. Anyone has an idea why this have so much load ?


Thanx in advance.

[]'s
_________________________________________________________________
Eng° Rodrigo Luiz Anami                   rodrigoa at bestway.com dot br
Administrador de Sistemas    Voz:(19) 736 3721 Cel:(19) 9123 9995
Best Way Internet Provider                (19) 736 3700 (Campinas)
http://www.bestway.com.br                (19) 736 3727 (FAX)
webmaster at bestway.com dot br          0800 112262 (Outras Localidades)


Date: Wed, 21 Jun 2000 16:59:23 -0500 (CDT)
From: "Joseph W. Breu" <breu at cfu dot net>
Subject: Re: Too load !

On Wed, 21 Jun 2000, Rodrigo Luiz Anami wrote:

> We are using qpopper 3.0.2 and sendmail 8.10.2 on a Slackware linux 
> machine. Our customers totalizes about 6,000 users. We see too much load on 
> our server all day, about 2 and 3. And the server has been present some 
> peaks of 16 ! The server is an IBM Net Finity 5000, with one Pentium III 
> 600 Mhz and 512 MB RAM. Anyone has an idea why this have so much load ?

I would say directory lookups in /var/spool/mail.  Look at hashing the
mail spool and home dirs.

-- 
	Thanks,
	-Joseph W. Breu

-----------------------------------------------------------------------
  Joseph W. Breu          Systems Administrator / Cedar Falls Utilities
  phone: (319) 268-5228        Utility Parkway, Cedar Falls, Iowa 50613 
  pager: (319) 235-4209  NIC: jwb96   breu at cfu.net   breu dot pager at cfu dot net
---- Al Gore: "...I took the initiative in creating the Internet. -----


From: "Mitch Vincent" <mitch at venux dot net>
Subject: Re: Too load !
Date: Wed, 21 Jun 2000 18:04:19 -0400

Look at the output of 'ps' and 'top' and figure out what processes are
taking up all the resources.. It may not have anything to do with qpopper at
all..

-Mitch

----- Original Message -----
From: "Rodrigo Luiz Anami" <rodrigoa at bestway.com dot br>
To: "Subscribers of Qpopper" <qpopper at lists.pensive dot org>
Sent: Wednesday, June 21, 2000 5:39 PM
Subject: Too load !


Hi there,

We are using qpopper 3.0.2 and sendmail 8.10.2 on a Slackware linux
machine. Our customers totalizes about 6,000 users. We see too much load on
our server all day, about 2 and 3. And the server has been present some
peaks of 16 ! The server is an IBM Net Finity 5000, with one Pentium III
600 Mhz and 512 MB RAM. Anyone has an idea why this have so much load ?


Thanx in advance.

[]'s
_________________________________________________________________
Engƒ Rodrigo Luiz Anami                   rodrigoa at bestway.com dot br
Administrador de Sistemas    Voz:(19) 736 3721 Cel:(19) 9123 9995
Best Way Internet Provider                (19) 736 3700 (Campinas)
http://www.bestway.com.br                (19) 736 3727 (FAX)
webmaster at bestway.com dot br          0800 112262 (Outras Localidades)




Date: Wed, 21 Jun 2000 23:16:38 +0100
From: Fergal Daly <fergal at esatclear dot ie>
Subject: Re: Too load !

At 22:59 21/06/00, Joseph W. Breu wrote:
>On Wed, 21 Jun 2000, Rodrigo Luiz Anami wrote:
>
> > We are using qpopper 3.0.2 and sendmail 8.10.2 on a Slackware linux
> > machine. Our customers totalizes about 6,000 users. We see too much 
> load on
> > our server all day, about 2 and 3. And the server has been present some
> > peaks of 16 ! The server is an IBM Net Finity 5000, with one Pentium III
> > 600 Mhz and 512 MB RAM. Anyone has an idea why this have so much load ?
>
>I would say directory lookups in /var/spool/mail.  Look at hashing the
>mail spool and home dirs.

We got away with 30k users in one directory on a lower spec for years. We 
were using server mode, which would make a much bigger difference in the 
right circumstances. 512M means that the complete directory structure could 
easily be held in memory,

Fergal



Date: Wed, 21 Jun 2000 22:33:05 +0000 (GMT)
From: Vinay Bharel <vinay at bharel dot com>
Subject: Re: Too load ! (fwd)

---------- Forwarded message ----------
Date: Wed, 21 Jun 2000 22:30:51 +0000 (GMT)
From: Vinay Bharel <vinay at bharel dot com>
To: Rodrigo Luiz Anami <rodrigoa at bestway.com dot br>
Subject: Re: Too load !

Yes.  Look in your logs.  Do you see any dquot, VFS error messages like
these:

May 11 21:37:34 beast kernel: get_empty_dquot: pruning dcache and          =
                                                   
icache                                                                     =
                                                   
May 11 21:37:34 beast kernel: VFS: No free dquots, contact                 =
                                                   
mvw at planets.elm dot net                                                        =
                                                   
May 11 21:37:35 beast kernel: get_empty_dquot: pruning dcache and          =
                                                   
icache                                                                     =
                                                   
May 11 21:37:35 beast kernel: VFS: No free dquots, contact                 =
                                                   
mvw at planets.elm dot net                                                        =
                                                   
May 11 21:37:35 beast kernel: get_empty_dquot: pruning dcache and icache   =
                                                   

What Linux Kernel are you running?  There seems to be a problem with Quota
FS in Linux when handling many concurrent users.

On one of our mail servers running Slackware 7 we have the same
problem.  We have approx 11,000 users on that box.  I would see load avg
go up to 600 and 700 ( and no i am not kidding!).

What would happen is that the Quota subsystem would not be able to handle
the multiple simultaneous requests and would not be able to complete the
quota checks requested.

This would cause popper proccesses to just hang.  I was not able to kill
them with kill -9 either.  Also eventually sendmail will die.

The only fix was to turn off quota.  I have not been able to find any
other solutions to this problem.  I am hoping the 2.4 kernel would fix
this bug.

If you find any solutions, do let me know.

---
Vinay Bharel
vinay at bharel dot com
http://www.bharel.com

On Wed, 21 Jun 2000, Rodrigo Luiz Anami wrote:

> 
> Yes, just in /var to limit individual spool mail. Any problem with quota =
?
> 
> 
> []'s
> 
> 
> At 22:03 21/06/00 +0000, you wrote:
> >Are you using Quota on any of your partitions?
> >
> >---
> >Vinay Bharel
> >vinay at bharel dot com
> >http://www.bharel.com
> >
> >On Wed, 21 Jun 2000, Rodrigo Luiz Anami wrote:
> >
> > > Hi there,
> > >
> > > We are using qpopper 3.0.2 and sendmail 8.10.2 on a Slackware linux
> > > machine. Our customers totalizes about 6,000 users. We see too much=
 
> > load on
> > > our server all day, about 2 and 3. And the server has been present so=
me
> > > peaks of 16 ! The server is an IBM Net Finity 5000, with one Pentium =
III
> > > 600 Mhz and 512 MB RAM. Anyone has an idea why this have so much load=
 ?
> > >
> > >
> > > Thanx in advance.
> > >
> > > []'s
> > > _________________________________________________________________
> > > Eng° Rodrigo Luiz Anami                   rodrigoa at bestway.com dot br
> > > Administrador de Sistemas    Voz:(19) 736 3721 Cel:(19) 9123 9995
> > > Best Way Internet Provider                (19) 736 3700 (Campinas)
> > > http://www.bestway.com.br                (19) 736 3727 (FAX)
> > > webmaster at bestway.com dot br          0800 112262 (Outras Localidades)
> > >
> 
> _________________________________________________________________
> Eng° Rodrigo Luiz Anami                   rodrigoa at bestway.com dot br
> Administrador de Sistemas    Voz:(19) 736 3721 Cel:(19) 9123 9995
> Best Way Internet Provider                (19) 736 3700 (Campinas)
> http://www.bestway.com.br                (19) 736 3727 (FAX)
> webmaster at bestway.com dot br          0800 112262 (Outras Localidades)
> 



From: "Dave Linden" <davel at west-teq dot net>
Subject: POP3 ERR
Date: Wed, 21 Jun 2000 16:41:15 -0600

Take a look at the messages log file below.  I'm pretty new at this and
suspect there are some trouble spots here.  Could anyone please make a
recommendation.  Running Qpopper 3.0.2 on FREEBSD.  I'm really worried about
that Unknown authentication mechanism: twinkie instead   I have have no idea
what it means.
Thanks in advance for the suggestions.

Dave



Jun 21 10:15:15 cherokee named[2477]: Ready to answer queries
Jun 21 10:33:00 cherokee popper[2632]: dreammaker at dialin44.wlk.oa.net
(207.15
3.22.172): -ERR Unknown command: "xsender".set using SOA minimum instead
Jun 21 11:12:28 cherokee popper[2865]: (null) at dialin11.wlk.oa.net
(207.153.22
.139): -ERR Unknown authentication mechanism: twinkie instead
Jun 21 11:18:08 cherokee su: davel to root on /dev/ttyp0
Jun 21 11:18:18 cherokee passwd: login_getclass: unknown class 'USERS'
cherokee:{19}:/var/log#



Date: Wed, 21 Jun 2000 18:04:22 -0500 (CDT)
From: Jorge Kobeh <jkobeh at spin.com dot mx>
Subject: Re: Too load !

In my own experience, sendmail is the one that causes the problem when our
connection to the Internet is over 90% used. In those cases, sendmail takes
longer to deliver outgoing mail, so more sendmail process stay around takin=
g
resources from the CPU.

-- 
Sincerely, Jorge Kobeh - Servicio SPIN

On Wed, 21 Jun 2000, Rodrigo Luiz Anami wrote:

> Hi there,
> 
> We are using qpopper 3.0.2 and sendmail 8.10.2 on a Slackware linux 
> machine. Our customers totalizes about 6,000 users. We see too much load =
on 
> our server all day, about 2 and 3. And the server has been present some=
 
> peaks of 16 ! The server is an IBM Net Finity 5000, with one Pentium III=
 
> 600 Mhz and 512 MB RAM. Anyone has an idea why this have so much load ?
> 
> 
> Thanx in advance.
> 
> []'s
> _________________________________________________________________
> Eng° Rodrigo Luiz Anami                   rodrigoa at bestway.com dot br
> Administrador de Sistemas    Voz:(19) 736 3721 Cel:(19) 9123 9995
> Best Way Internet Provider                (19) 736 3700 (Campinas)
> http://www.bestway.com.br                (19) 736 3727 (FAX)
> webmaster at bestway.com dot br          0800 112262 (Outras Localidades)
> 
> 


Date: Thu, 22 Jun 2000 08:23:31 +0900
From: Peter Evans <peter at gol dot com>
Subject: Re: Too load ! (fwd)

Vinay Bharel (vinay at bharel dot com) wrote:
> The only fix was to turn off quota.  I have not been able to find any
> other solutions to this problem.  I am hoping the 2.4 kernel would fix
> this bug.
 
	1 - dont use linux.  ^_^;

	not really an option, and I don't really care what OS people use,
	I have enough people wanting to convert our mail system to linux
	as it is.

	2 - dont use quota.

	since mail.local or its delivery equivalent is the main culprit
	for putting things in your mailbox, why not implement quota there,
	its not even 50 lines of code. I can stick the diff's on a web page
	if anyone is interested.

	Actually, for the brief run of qpopper3 I did on irix, I had
	a remarkable load, all qpoppers. using server mode. I think they
	stuck something in there.

	It might be worth doing some profiling to see where it is wasting
	its time.


	P

-- 
Remember The 5 K's.
The Justified Agents of Munya-munya-muuuu ...

Date: Wed, 21 Jun 2000 23:49:59 +0000 (GMT)
From: Vinay Bharel <vinay at bharel dot com>
Subject: Re: Too load ! (fwd)

> 	2 - dont use quota.
> 
> 	since mail.local or its delivery equivalent is the main culprit
> 	for putting things in your mailbox, why not implement quota there,
> 	its not even 50 lines of code. I can stick the diff's on a web page
> 	if anyone is interested.
> 

That would be much appreciated!  Has anyone done something similar with
Procmail?


Date: Wed, 21 Jun 2000 18:29:30 -0700
From: Qpopper Support <qpopper at qualcomm dot com>
Subject: Qpopper 3.1b3 available

Qpopper 3.1b3 is available at 
<ftp://ftp.qualcomm.com/eudora/servers/unix/popper/>.

A somewhat abbreviated change list is on the Qpopper web site at 
<http://www.qpopper.org/changelist.html>

The full list of changes from one release to the next is on the FTP 
site, at 
<ftp://ftp.qualcomm.com/eudora/servers/unix/popper/beta/Changes>.

The 3.1 release notes are at: 
<ftp://ftp.qualcomm.com/eudora/servers/unix/popper/beta/Release.Notes>.

Changes from 3.1b2 to 3.1b3
---------------------------

  1.  popauth passed-in password no longer prompts to confirm.
  2.  Added -help switch to popauth.
  3.  Fixed return value of Qtouchlock; added error checking.
  4.  Fixed stray control character in PAM section of pop_pass.c
  5.  No longer reporting "-1 hidden" msgs for users with bull errors.
  6.  Fixed crash un Qtouchlock() on NeXT.
  7.  Added '--disable-old-spool-loc' to not check for old .user.pop
      files in old locations when HASH_SPOOL or HOMEDIRMAIL used.
  8.  No longer checking for old .user.pop file in old locations unless
      HASH_SPOOL or HOMEDIRMAIL used.
  9.  Added '--disable-check-hash-dir' to not check for or create hash
      spool directories.  Use this if you pre-create the directories.
10.  Added '--enable-server-mode-group-include=group' and
      '--enable-server-mode-group-exclude=group' to set server mode on
      or off by default for users in the specified group.
11.  Using passed-in passwd structure in pop_bull instead of calling
      getpwnam yet again.
12.  Removed redundant call on genpath from apop, rpop, and scram code.
13.  Deleted unused variables.
14.  Fixed typo in popauth causing failures.
15.  Applied patch from Kenneth Porter to fix many compiler warnings.
16.  Fixed lots and lots of compiler warnings.
17.  Fixed references to value.dptr for even pickier platforms.
18.  When SECURENISPLUS defined, now getting password as user and then
      as root, to handle case where 'compat' is used in 'nsswitch.conf'.
      Based on patch by Neil W. Rickert.
19.  Added '--enable-secure-nis-plus' to set SECURENISPLUS.
20.  Fixed check for BSD DBM files to avoid false positive with pedantic
      compilers.
21.  Deleted '--with-warnings' configure flag.
22.  '--enable-warnings' flag now sets correct CFLAGS on HP systems.



Date: Thu, 22 Jun 2000 21:37:56 +1200 (NZST)
From: Alan Brown <alan at manawatu.gen dot nz>
Subject: Re: Too load !

On Wed, 21 Jun 2000, Fergal Daly wrote:

> We got away with 30k users in one directory on a lower spec for years. We 
> were using server mode, which would make a much bigger difference in the 
> right circumstances. 512M means that the complete directory structure could 
> easily be held in memory,

Size of individual mailboxes make a big difference to performance.

Several users with a few thousand messages left on the server can cause
a signifcant load if they're constantly checking mail.

AB


Date: Thu, 22 Jun 2000 06:39:46 -0500 (CDT)
From: Jason Englander <jason at interl dot net>
Subject: Re: How can I make qpopper create dot-lock file in different directory

On Mon, 12 Jun 2000, Randall Gellens wrote:

> You can't.  The .lock file must be in the same directory as the file 
> being locked, or there is no point.
> 
> The .lock file prevents the delivery agent (or other processes) from 
> accessing the spool while Qpopper is processing it.

Forgive me if this is a dumb question (been up all night so it probably
is), but is it possible for qpopper create the .lock file as (owned by) a
different user (one that doesn't have a quota)?  ie. root, nobody, bin,
daemon.

Before the process becomes owned by the user?  Does it go back to being
owned by root when it's done before it dies off? (so it could remove the
.lock file)

...boy would that be nice.  If a user that's over quota could go in
(a webmail, telnet to port 110) and delete their mail without having to
retr 20 meg of junk over a dialup connection.

  Jason 

-- 
Jason Englander <jason at interl dot net>
Systems Administrator - InterLink L.C.



Date: Thu, 22 Jun 2000 07:08:50 -0500 (CDT)
From: Jason Englander <jason at interl dot net>
Subject: Re: enable-hash-spool=22

On Tue, 13 Jun 2000, Fergal Daly wrote:

> Not trying to turn this into a perl list but I think there are some
> improvements to be made here

Me either, but I have one that might help someone...

Adding:

umask(0000);

somewhere at the top in there (before the foreach loop) will ensure that
the permissions on the directories are drwxrwxrwx.  If your umask is 022
or something like that, you'll actually get drwxr-xr-x

"man perlfunc"'s umask section and "man umask" explain it better...

  Jason

 
> - using .. for sequences of letters/numbers
> - using the builtins when possible (mkdir, chmod)
> - using foreach when iterating over an array
> - using system instead of `` when you're not actually going to catch
> the output
> 
> not much point writing it in perl if you're going to do all
> the work of a C programmer ;-)
> 
> #!/usr/bin/perl
> 
> @directories = ("a".."z", 0..9, "-", "_");
> 
> $main_dir = ".";        # no trailing slash
> $ownership = "root:mail";       # chown:chgrp values
> 
> foreach $letter1 (@directories)  {
>    mkdir("$main_dir/$letter1", 01777);
> 
>    foreach $letter2 (@directories)  {
>         mkdir("$main_dir/$letter1/$letter2", 01777);
>    }
> 
>    system("chown -R $ownership $main_dir/$letter1");
> }
> 
> On Tue, 13 Jun 2000, you wrote: > Here is
> the script we used to create the directories: >  > =========== Script
> Follows ========================== > 
> > #!/usr/bin/perl
> > @directories = ("a", "b", "c", "d", "e", "f", "g", "h", "i", "j", "k",
> > "l", "m", "n", "o", "p", "q", "r", "s", "t", "u", "v", "w", "x", "y", "z",
> > "0", "1", "2", "3", "4", "5", "6", "7", "8", "9", "-", "_");
> > $main_dir = /var/spool/mail;	# no trailing slash
> > $ownership = "root:mail";	# chown:chgrp values
> > $loop1 = 0;
> > $loop2 = 0;
> > while ($loop1 <= $#directories)	{
> >    `mkdir $main_dir/$directories[$loop1]`;
> >    while ($loop2 <= $#directories)	{
> > 	`mkdir $main_dir/$directories[$loop1]/$directories[$loop2]`;
> > 	`chmod 1777 $main_dir/$directories[$loop1]/$directories[$loop2]`;
> > 	++$loop2;
> >    }
> >    `chown -R $ownership $main_dir/$directories[$loop1]`;
> >    `chmod  1777 $main_dir/$directories[$loop1]`;
> >    ++$loop1;
> >    $loop2 = 0;
> > }
> > 
> > exit;
> > 
> > =========== End Script ================================
> > 
> > Remove any characters necessary from the @directories (we allow #s and
> > -_).  Change the main_dir and ownership if necessary.  Ownership is
> > recursive on all directories created.
> > 
> > -- 
> > ===============================================
> > Kip Turk		    phone: 915.234.5678
> > Systems Administrator   	or 800.695.9016	
> > West Central Net	      fax: 915.656.0071
> > ===============================================
> > 
> > On Tue, 13 Jun 2000, Joseph W. Breu wrote:
> > 
> > > > What does everybody use for a local mail
> > > > delivery agent that will put mail into that
> > > > directory?
> > > 
> > > I use procmail.
> > > 
> > > > I've tried procmail but it seems to have issues
> > > > creating the dirs if they don't already exist.
> > > 
> > > Just "pre-create" the directories using a shell script or whatever you
> > > have handy.
> > > 
> > >
> 

-- 
Jason Englander <jason at interl dot net>
Systems Administrator - InterLink L.C.



From: "Keith Warno" <keith at HaggleWare dot com>
Subject: Re: Too load !
Date: Thu, 22 Jun 2000 14:21:17 -0400

Sendmail?  *choke*  *gasp*

An alternate option: try qmail.  Deliver to ~user/Maildir/ and use
qmail-pop3d.  I don't know about the performance, although someone at
www.qmail.org will (also check out http://cr.yp.to/).  In any case, you'll
get the added bonus of stability AND security.

Flames will be directed to /dev/null ; this was not intended to start an MTA
war.  It's just a possible solution.

kw


----- Original Message -----
From: "Rodrigo Luiz Anami" <rodrigoa at bestway.com dot br>
To: "Subscribers of Qpopper" <qpopper at lists.pensive dot org>
Sent: 21 June 2000, Wednesday 17:39
Subject: Too load !


Hi there,

We are using qpopper 3.0.2 and sendmail 8.10.2 on a Slackware linux
machine. Our customers totalizes about 6,000 users. We see too much load on
our server all day, about 2 and 3. And the server has been present some
peaks of 16 ! The server is an IBM Net Finity 5000, with one Pentium III
600 Mhz and 512 MB RAM. Anyone has an idea why this have so much load ?


Thanx in advance.

[]'s
_________________________________________________________________
Engƒ Rodrigo Luiz Anami                   rodrigoa at bestway.com dot br
Administrador de Sistemas    Voz:(19) 736 3721 Cel:(19) 9123 9995
Best Way Internet Provider                (19) 736 3700 (Campinas)
http://www.bestway.com.br                (19) 736 3727 (FAX)
webmaster at bestway.com dot br          0800 112262 (Outras Localidades)




From: "Kenneth Porter" <shiva at well dot com>
Date: Thu, 22 Jun 2000 11:30:17 -0700
Subject: ANN: New Red Hat Linux kernel addresses security vulnerability

Please forgive the somewhat off-topic posting, but I expect many
qpopper admins also run sendmail on Linux. I reported a couple weeks
ago about a new vulnerability in the kernel affecting sendmail and
other setuid programs that drop priviliges before launching another
program (such as a user-specified filter in a .forward file), allowing
any user to get root. Red Hat just released a new kernel RPM that
addresses the problem. Further details may be found at:

http://www.redhat.com/support/errata/RHSA-2000-037-02.html
http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=11974

Followup discussion should be posted to the newsgroups
redhat.security.general and comp.os.linux.security.

Ken
mailto:shiva at well dot com
http://www.sewingwitch.com/ken/
http://www.harrybrowne2000.org/



Date: Thu, 22 Jun 2000 13:55:34 -0500 (CDT)
From: Jason Englander <jason at interl dot net>
Subject: 3.1b3 buglet?

I'm no C programmer, by any means, but isn't there one too many %s -es
on line 63 of popper/drac.c  ?

            pop_log ( p, POP_PRIORITY, HERE,
-->                   "[drac]: login to %s from host %s (%s) %s",
                      p->user, p->client, p->ipaddr );
            return POP_SUCCESS;

This is producing control characters and high ascii garbage in what's
output to syslog... (so I removed it)

...and BTW, I didn't have a /etc/mail/drac.host file when I first tried it
out, and it didn't go over very well without one :-)  Perhaps a check for
that file would be good for shmoze like me?  Nothing was logged, I didn't
have a clue.  Telnet into 110, accepts my username/pw, boots me out.

I was using Steven Champeon's drac patch for 3.0.2 and never even thought
to look for a file that I'd have to add to get it set up with 3.1b3
Though I think it's a good idea to have the flexibility of it not being
hard coded to localhost.

Thanks,
  Jason

-- 
Jason Englander <jason at interl dot net>
Systems Administrator - InterLink L.C.




Date: Thu, 22 Jun 2000 15:21:31 -0400
From: Steven Champeon <schampeo at hesketh dot com>
Subject: Re: 3.1b3 buglet?

on Thu, Jun 22, 2000 at 01:55:34PM -0500, Jason Englander wrote:
> I'm no C programmer, by any means, but isn't there one too many %s -es
> on line 63 of popper/drac.c  ?
> 
>             pop_log ( p, POP_PRIORITY, HERE,
> -->                   "[drac]: login to %s from host %s (%s) %s",
>                       p->user, p->client, p->ipaddr );
>             return POP_SUCCESS;
> 
> This is producing control characters and high ascii garbage in what's
> output to syslog... (so I removed it)

Yeah, I noticed that, too, back when 3.1b2 came out - but got
distracted before I could say something.
 
> ...and BTW, I didn't have a /etc/mail/drac.host file when I first
> tried it out, and it didn't go over very well without one :-)
> Perhaps a check for that file would be good for shmoze like me?
> Nothing was logged, I didn't have a clue.  Telnet into 110, accepts
> my username/pw, boots me out.
>
> I was using Steven Champeon's drac patch for 3.0.2 and never even
> thought to look for a file that I'd have to add to get it set up
> with 3.1b3

Yeah, some documentation would be a good idea.

> Though I think it's a good idea to have the flexibility of it not being
> hard coded to localhost.

Yeah, me, too - at first.

I'm a bit concerned, though, that popper checks for the existence of, and
reads the file, every time. Is this really the sort of thing that changes
on a regular basis? "Oh, it's Tuesday, better move the host that's serving
as the DRAC host to another IP" ;) I'd rather see it as a command-line arg.

Steve

-- 
http://a.jaundicedeye.com/weblog/
because it just annoys people when I talk to myself...

Date: Fri, 23 Jun 2000 07:37:43 +1200 (NZST)
From: Alan Brown <alan at manawatu.gen dot nz>
Subject: Re: Too load !

On Thu, 22 Jun 2000, Keith Warno wrote:

> Flames will be directed to /dev/null ; this was not intended to start an MTA
> war.  It's just a possible solution.

I'm not going to flame - just point out that qmail breaches RFC 821 in
several areas. It's also fairly unfriendly about bandwidth usage _and_
accepts some relay attempts, then dumps them in the postmaster box,
which can result in local or proxy mailbombings.

Postfix and Exim are two other alternatives to sendmail, both better
behaved.

Qmail _is_ useful if you run a list processor because of its high
throughput abilities, but IMO it's not suitable as a general purpose
mailserver. It was designed for list servers with good bounce management
and it shows.

AB


Date: Sun, 25 Jun 2000 14:26:42 -0700
From: Qpopper Support <qpopper at qualcomm dot com>
Subject: Re: Too load ! (fwd)

At 8:23 AM +0900 6/22/00, Peter Evans wrote:

>              Actually, for the brief run of qpopper3 I did on irix, I had
>  	a remarkable load, all qpoppers. using server mode. I think they
>  	stuck something in there.

Which version of 3?  There was a bug in 3.0 and 3.0.1, fixed in 
3.0.2, that caused poor performance in many instances.

Date: Sun, 25 Jun 2000 18:39:18 -0500 (CDT)
From: Jason Englander <jason at interl dot net>
Subject: Re: 3.1b3 buglet?

On Thu, 22 Jun 2000, Steven Champeon wrote:

> I'm a bit concerned, though, that popper checks for the existence of, and
> reads the file, every time. Is this really the sort of thing that changes
> on a regular basis? "Oh, it's Tuesday, better move the host that's serving
> as the DRAC host to another IP" ;) I'd rather see it as a command-line arg.

I realized that after my last message.  Configurability = good, increased
I/O = bad.  ...I don't think I'll ever be changing it, and even if I were
to do so, even a --with-drac-host=localhost flag for autoconf would be
good.  A command line arg, gooder.

  Jason

-- 
Jason Englander <jason at interl dot net>
Systems Administrator - InterLink L.C.




From: "ganizani phiri" <ganizani at malawi dot net>
Subject: messages preceeded by ! any danger?
Date: Mon, 26 Jun 2000 08:28:37 +0300

When I peep through the mail box via email remover I see some messages
preceeded by an eclamation mark like

! 50K Date Time     From    Subject

Is there any danger in this.

Ganizani


Date: Mon, 26 Jun 2000 10:28:47 -0300
From: Rodrigo Luiz Anami <rodrigoa at bestway.com dot br>
Subject: Procmail

Hi there !

There is any type of conflict or concurrence between procmail and qpopper 
that can cause high load. Anyone has experimented it ?

[]'s
_________________________________________________________________
Eng° Rodrigo Luiz Anami                   rodrigoa at bestway.com dot br
Administrador de Sistemas    Voz:(19) 736 3721 Cel:(19) 9123 9995
Best Way Internet Provider                (19) 736 3700 (Campinas)
http://www.bestway.com.br                (19) 736 3727 (FAX)
webmaster at bestway.com dot br          0800 112262 (Outras Localidades)


Subject: Re: Procmail
Date: Mon, 26 Jun 2000 15:43:56 +0200 (MET DST)
From: Eric Luyten <Eric.Luyten at vub.ac dot be>

> There is any type of conflict or concurrence between procmail and qpopper 
> that can cause high load. 

Really ? Please describe the problem.

>                            Anyone has experimented it ?

No. We are running procmail 3.14 and Qpopper3.0b34 (slightly patched) 
on a very busy server. No problemos.

I noticed (wallclock reading) that procmail 3.14 takes about half the
time to complete message delivery compared to the mail.local that comes 
with Solaris 7.      Which means the 'lock' lives a lot less long.

> []'s

Brackets 2 U 2


Eric.

Date: Mon, 26 Jun 2000 10:58:25 -0300
From: Rodrigo Luiz Anami <rodrigoa at bestway.com dot br>
Subject: Re: Procmail

At 15:43 26/06/00 +0200, you wrote:
> > There is any type of conflict or concurrence between procmail and=
 qpopper
> > that can cause high load.
>
>Really ? Please describe the problem.

We have a Slackware Linux machine running qpopper 3.0.2, sendmail 8.10.2 
and procmail 3.14. We have about 6k users and sometimes our linux box 
reaches loads about 4 and 5. When I use top command, it displays some 
popper and procmail processess in status D at the same time that load is 
high. Our hardware is a Net Finity 5000 with Pentium III 600 Mhz, 512 Mb 
RAM and 3 UW2 Scsi 18 GB disks. Two of them are in RAID level one.

[]'s



> >                            Anyone has experimented it ?
>
>No. We are running procmail 3.14 and Qpopper3.0b34 (slightly patched)
>on a very busy server. No problemos.
>
>I noticed (wallclock reading) that procmail 3.14 takes about half the
>time to complete message delivery compared to the mail.local that comes
>with Solaris 7.      Which means the 'lock' lives a lot less long.
>
> > []'s
>
>Brackets 2 U 2
>
>
>Eric.

_________________________________________________________________
Eng° Rodrigo Luiz Anami                   rodrigoa at bestway.com dot br
Administrador de Sistemas    Voz:(19) 736 3721 Cel:(19) 9123 9995
Best Way Internet Provider                (19) 736 3700 (Campinas)
http://www.bestway.com.br                (19) 736 3727 (FAX)
webmaster at bestway.com dot br          0800 112262 (Outras Localidades)


Date: Mon, 26 Jun 2000 17:10:28 -0700 (PDT)
From: David Schlussel <david at bestoffer dot com>
Subject: compile error: undefined symbol __inet_ntoa

Hello,

I've gotten past a few compile errors, but this one stumps me.  I've got
the '-lnsl -lsocket -lresolv -lmail ' flags set, which should include
inet_ntoa, but somehow it's not being found

I'm on solaris 2.7, using gcc (2.8.1?), bind 8.2.2

Anyone else run into this?

-David


From: "Kenneth Porter" <shiva at well dot com>
Date: Tue, 27 Jun 2000 10:06:12 -0700
Subject: Re: compile error: undefined symbol __inet_ntoa

On Mon, 26 Jun 2000 17:10:28 -0700 (PDT), David Schlussel wrote:

>I've gotten past a few compile errors, but this one stumps me.  I've got
>the '-lnsl -lsocket -lresolv -lmail ' flags set, which should include
>inet_ntoa, but somehow it's not being found
>
>I'm on solaris 2.7, using gcc (2.8.1?), bind 8.2.2

If it's a compile error, does that imply *not* a linker error? If so,
just grep your headers to see where that symbol is defined. It looks
like an internal symbol used by a macro. Defining additional libraries
only helps at link time. Looks like you're missing a necessary header.

Ken
mailto:shiva at well dot com
http://www.sewingwitch.com/ken/
http://www.harrybrowne2000.org/



Date: Tue, 27 Jun 2000 10:11:12 -0700 (PDT)
From: David Schlussel <david at bestoffer dot com>
Subject: Re: compile error: undefined symbol __inet_ntoa

hi, and thanks for the speedy response...

i'm not sure about the differences, it looks like a linker error at
further examination...

# make
....
gcc  pop_dele.o pop_dropcopy.o pop_get_command.o pop_get_subcommand.o
pop_init.o pop_last.o pop_list.o pop_log.o pop_lower.o pop_msg.o
pop_parse.o pop_pass.o pop_quit.o pop_rset.o pop_send.o pop_stat.o
pop_updt.o pop_user.o pop_xtnd.o pop_xmit.o popper.o pop_bull.o
xtnd_xlst.o pop_uidl.o mktemp.o pop_rpop.o pop_apop.o md5.o pop_auth.o
pop_pope.o pop_extend.o scram.o hmac.o base64.o pop_util.o get_sub_opt.o
msg_ptr.o -o popper -lnsl -lsocket -lresolv -lmail  -lkrb   -lcrypt
../mmangle/libmangle.a ../common/libcommon.a
Undefined                       first referenced
 symbol                             in file
__inet_ntoa                         pop_init.o
ld: fatal: Symbol referencing errors. No output written to popper
make[1]: *** [popper] Error 1
make[1]: Leaving directory `/opt/src/qpopper3.0.2/popper'
make: *** [popper_server] Error 2

it only appears to have make a call to inet_ntoa here:

./popper/pop_init.c:         
            kdata.pinst, kdata.prealm, inet_ntoa(addr->sin_addr) );
./popper/pop_init.c:    
            p->ipaddr = (char *) strdup ( inet_ntoa ( cs.sin_addr ) );

thanks again

david

On Tue, 27 Jun 2000, Kenneth Porter wrote:

> On Mon, 26 Jun 2000 17:10:28 -0700 (PDT), David Schlussel wrote:
> 
> >I've gotten past a few compile errors, but this one stumps me.  I've got
> >the '-lnsl -lsocket -lresolv -lmail ' flags set, which should include
> >inet_ntoa, but somehow it's not being found
> >
> >I'm on solaris 2.7, using gcc (2.8.1?), bind 8.2.2
> 
> If it's a compile error, does that imply *not* a linker error? If so,
> just grep your headers to see where that symbol is defined. It looks
> like an internal symbol used by a macro. Defining additional libraries
> only helps at link time. Looks like you're missing a necessary header.
> 
> Ken
> mailto:shiva at well dot com
> http://www.sewingwitch.com/ken/
> http://www.harrybrowne2000.org/
> 
> 


From: "Kenneth Porter" <shiva at well dot com>
Date: Tue, 27 Jun 2000 10:38:09 -0700
Subject: Re: compile error: undefined symbol __inet_ntoa

On Tue, 27 Jun 2000 10:11:12 -0700 (PDT), David Schlussel wrote:

>Undefined                       first referenced
> symbol                             in file
>__inet_ntoa                         pop_init.o
>ld: fatal: Symbol referencing errors. No output written to popper

>./popper/pop_init.c:         
>            kdata.pinst, kdata.prealm, inet_ntoa(addr->sin_addr) );
>./popper/pop_init.c:    
>            p->ipaddr = (char *) strdup ( inet_ntoa ( cs.sin_addr ) );

The curious thing to me is that the symbol has two leading underscores,
which is generally a clue that the C symbol has one, and is a worker
function serving a macro.

Under Linux (RH6.1), inet_ntoa is a function declared in
/usr/include/arpa/inet.h (#include <arpa/inet.h>). You might try
compiling pop_init.c and adding the -E or -S switches to just
preprocess or compile to assembly and inspect the result to see if the
correct symbols are getting used.

BTW, the man page for the API shows what header to include, but I don't
see the library name. Anyone know how to determine which library
provides a given API? (At least for gcc.)

Ken
mailto:shiva at well dot com
http://www.sewingwitch.com/ken/
http://www.harrybrowne2000.org/



Date: Tue, 27 Jun 2000 11:30:12 -0700 (PDT)
From: David Schlussel <david at bestoffer dot com>
Subject: Re: compile error: undefined symbol __inet_ntoa

thanks again for your continued help.

when i installed bind, i moved aside inet.h to allow bind's inet.h,
and they have fairly different definitions

--

include# grep ntoa arpa/inet.h.old /usr/local/bind/include/arpa/inet.h

arpa/inet.h.old:extern char *inet_ntoa(struct in_addr);
arpa/inet.h.old:char    *inet_ntoa();

bind/include/arpa/inet.h:#define     inet_ntoa __inet_ntoa
bind/include/arpa/inet.h:#define     inet_nsap_ntoa __inet_nsap_ntoa
bind/include/arpa/inet.h:/*const*/ char      *inet_ntoa
         __P((struct in_addr));
bind/include/arpa/inet.h:char                *inet_nsap_ntoa
         __P((int, const u_char *, char *));

---

That would explain the __, i see no other mention of it under
/usr/local/bind

i'd like to use the new build of bind.  i'm looking into the bind mailing
list archives.  in one spot they mention linking in libbind.a.  my c is
incredibly rusty and i'm not sure how to do that. 

thanks for the help

-david


On Tue, 27 Jun 2000, Kenneth Porter wrote:

> On Tue, 27 Jun 2000 10:11:12 -0700 (PDT), David Schlussel wrote:
> 
> >Undefined                       first referenced
> > symbol                             in file
> >__inet_ntoa                         pop_init.o
> >ld: fatal: Symbol referencing errors. No output written to popper
> 
> >./popper/pop_init.c:         
> >            kdata.pinst, kdata.prealm, inet_ntoa(addr->sin_addr) );
> >./popper/pop_init.c:    
> >            p->ipaddr = (char *) strdup ( inet_ntoa ( cs.sin_addr ) );
> 
> The curious thing to me is that the symbol has two leading underscores,
> which is generally a clue that the C symbol has one, and is a worker
> function serving a macro.
> 
> Under Linux (RH6.1), inet_ntoa is a function declared in
> /usr/include/arpa/inet.h (#include <arpa/inet.h>). You might try
> compiling pop_init.c and adding the -E or -S switches to just
> preprocess or compile to assembly and inspect the result to see if the
> correct symbols are getting used.
> 
> BTW, the man page for the API shows what header to include, but I don't
> see the library name. Anyone know how to determine which library
> provides a given API? (At least for gcc.)
> 
> Ken
> mailto:shiva at well dot com
> http://www.sewingwitch.com/ken/
> http://www.harrybrowne2000.org/
> 
> 


From: "Kenneth Porter" <shiva at well dot com>
Date: Tue, 27 Jun 2000 12:01:12 -0700
Subject: Re: compile error: undefined symbol __inet_ntoa

On Tue, 27 Jun 2000 11:30:12 -0700 (PDT), David Schlussel wrote:

>when i installed bind, i moved aside inet.h to allow bind's inet.h,
>and they have fairly different definitions

Ouch, looks like the bind release breaks builds expecting to use the
resolver in libc.

>i'd like to use the new build of bind.  i'm looking into the bind mailing
>list archives.  in one spot they mention linking in libbind.a.  my c is
>incredibly rusty and i'm not sure how to do that. 

Just add "-lbind" to the list of libraries in the link command line.

Meanwhile, qpopper's autoconfig needs to check for the presence of the
new bind library and, if found, add it to the list of libraries to link
against. I've not yet messed with autoconfig, so perhaps someone with
experience there can propose the patch?

Ken
mailto:shiva at well dot com
http://www.sewingwitch.com/ken/
http://www.harrybrowne2000.org/



Date: Tue, 27 Jun 2000 11:59:56 -0700 (PDT)
From: David Schlussel <david at bestoffer dot com>
Subject: Re: compile error: undefined symbol __inet_ntoa

thanks!  that did the trick.  i tried -llibbind, but it blew up.

-david

On Tue, 27 Jun 2000, Kenneth Porter wrote:

> On Tue, 27 Jun 2000 11:30:12 -0700 (PDT), David Schlussel wrote:
> 
> >when i installed bind, i moved aside inet.h to allow bind's inet.h,
> >and they have fairly different definitions
> 
> Ouch, looks like the bind release breaks builds expecting to use the
> resolver in libc.
> 
> >i'd like to use the new build of bind.  i'm looking into the bind mailing
> >list archives.  in one spot they mention linking in libbind.a.  my c is
> >incredibly rusty and i'm not sure how to do that. 
> 
> Just add "-lbind" to the list of libraries in the link command line.
> 
> Meanwhile, qpopper's autoconfig needs to check for the presence of the
> new bind library and, if found, add it to the list of libraries to link
> against. I've not yet messed with autoconfig, so perhaps someone with
> experience there can propose the patch?
> 
> Ken
> mailto:shiva at well dot com
> http://www.sewingwitch.com/ken/
> http://www.harrybrowne2000.org/
> 
> 


Date: Tue, 27 Jun 2000 16:07:03 -0400
From: Linda Pelleu Antil <linda.pelleu at nist dot gov>
Subject: How to upgrade properly

I have not seen this discussed on the list so I imagine I must be doing
something wrong here and hope you can give me guidance.  We have an environment
where we have some users who leave mail on the server and some who do not.  We
recently upgraded from qpopper 2.53 to qpopper3.1b3; when we put the new
version of popper in place (and restarted inetd), the users who had left mail
on the server downloaded those messages again, even though they had been
retrieved before (saw this with both Eudora and Netscape clients).  This
appeared to happen just the first time connecting to the new server, then
normal performance returned.  

I'd like to understand why this happened so we can perform upgrades in the
future without this happening.  We configured qpopper3.1b3 with
--enable-servermode, --enable-old-uidl, and --disable-status, among other
things.

Thanks for your help-
Linda


Subject: Re: How to upgrade properly
Date: Tue, 27 Jun 2000 16:11:03 -0400
From: Ken Hornstein <kenh at cmf.nrl.navy dot mil>

>I have not seen this discussed on the list so I imagine I must be doing
>something wrong here and hope you can give me guidance.  We have an environment
>where we have some users who leave mail on the server and some who do not.  We
>recently upgraded from qpopper 2.53 to qpopper3.1b3; when we put the new
>version of popper in place (and restarted inetd), the users who had left mail
>on the server downloaded those messages again, even though they had been
>retrieved before (saw this with both Eudora and Netscape clients).  This
>appeared to happen just the first time connecting to the new server, then
>normal performance returned.  

I _believe_ that is because the UIDL generation algorithm changed between
2.53 and 3.x, and clients that rely on UIDL to store mail on the server
would see those messages as "new" mail.

--Ken

Date: Tue, 27 Jun 2000 16:28:23 -0400
From: Linda Pelleu Antil <linda.pelleu at nist dot gov>
Subject: Re: How to upgrade properly

At 04:11 PM 6/27/00 -0400, Ken Hornstein wrote:
>>I have not seen this discussed on the list so I imagine I must be doing
>>something wrong here and hope you can give me guidance.  We have an 
>environment
>>where we have some users who leave mail on the server and some who do not. 
We
>>recently upgraded from qpopper 2.53 to qpopper3.1b3; when we put the new
>>version of popper in place (and restarted inetd), the users who had left mail
>>on the server downloaded those messages again, even though they had been
>>retrieved before (saw this with both Eudora and Netscape clients).  This
>>appeared to happen just the first time connecting to the new server, then
>>normal performance returned.  
>
>I _believe_ that is because the UIDL generation algorithm changed between
>2.53 and 3.x, and clients that rely on UIDL to store mail on the server
>would see those messages as "new" mail.
>
>--Ken

Thanks, Ken.  If that's the case then at least I should not have to worry about
this on our next upgrade.  I guess I had thought the --enable-old-uidl would
address that issue...

Linda


Subject: Re: How to upgrade properly
Date: Tue, 27 Jun 2000 16:29:33 -0400
From: Ken Hornstein <kenh at cmf.nrl.navy dot mil>

>Thanks, Ken.  If that's the case then at least I should not have to worry about
>this on our next upgrade.  I guess I had thought the --enable-old-uidl would
>address that issue...

Remember, that was a _guess_ :-)  It's hard to know for sure (I would wonder
why you're upgrading your production server to a beta release ...)

--Ken

From: rek2 <rek2 at linuxbusca dot com>
Subject: hello all!
Date: Wed, 28 Jun 2000 00:50:03 -0400

HEllo all! my name is 
Christian Fernandez i Work like a linux system administrator in screamdesign.com
i been using qpopper for 6 month now...
and today I just updated from the old series to the 3.2 one
and got lot of questions
-why now instaed of login the stats on /var/log/messages it does in
/var/log/maillog?  how can i have it like in the old series? 
-why the stats are not showing the messagges download or ready to download like
before? is that a bug?

I have some more... but  i will ask them when this ones comes clear
BTW I use qpopper with qmail and  $user/Mailbox directory if this helps

and sorry about my english im from spain.

Christian

Date: Wed, 28 Jun 2000 09:08:15 -0400
From: "jmcintyr" <jmcintyr at i-plus dot net>

problem, from time to time my outlook client users swear they have not received their mail but it is still sitting in mail file. most users retain their mail on the server for several days. how can i force, fool, outlook into downloading all the mail agai
n ?

my configuration is 10.2 hpux on a d250 running qpopper 3.0.2 and sendmail 8.9.3.

thanks...jim mc....

From: "Jack Barnett" <jbarnett at axil.netmate dot com>
Subject: qpopper 3.0.2 does compile on Solaris 2.3?
Date: Wed, 28 Jun 2000 09:15:49 -0500

--inx~/pop3/qpopper3.0.2 -->make
cd ./popper  && make  all
make[1]: Entering directory
`/export/home3/jbarnett/pop3/qpopper3.0.2/popper'
gcc -c -I.. -I.. -I. -I../mmangle -I../common \
-g -O2 -fpcc-struct-return  -DHAVE_CONFIG_H  -DSOLARIS2 -DSPEC_POP_AUTH -DGN
U_PA
SS -DNO_GETLINE -DCONTENT_LENGTH=1 -DUNIX pop_dropcopy.c -o pop_dropcopy.o
pop_dropcopy.c: In function `do_drop_copy':
pop_dropcopy.c:747: `EDQUOT' undeclared (first use this function)
pop_dropcopy.c:747: (Each undeclared identifier is reported only once
pop_dropcopy.c:747: for each function it appears in.)

make[1]: *** [pop_dropcopy.o] Error 1
make[1]: Leaving directory `/export/home3/jbarnett/pop3/qpopper3.0.2/popper'
make: *** [popper_server] Error 1

The system is a Solaris 2.3 (Sparc) running gcc 2.7.21 and it was configured
with

./configure --enabled-log-login

Older version of qpopper have compiled on this platform and this same tar
ball compiled cleanly and smoothly on a Solaris 2.5.1 (Sparc) system without
problem, you think it is something to do with Solaris 2.3?  The qpopper
version is 3.0.2.  Anyone have some suggestions?

thanks,
Jack





Date: Wed, 28 Jun 2000 09:29:16 -0500
From: "J.D. Bronson" <jdb at xpec dot com>
Subject: trouble with latest beta

I am running Solaris 8 (sparc) and using Qpopper /APOP.

I am noticing (intermittantly) that when a user pops mail, some messages
are within others...

For example:

I just received an email from a friend that contained a joke...after the
joke and his signature line, there was another email (FULL HEADERS too)
that was from the qpopper list!

Now I am taking these complaints seriously.....

Any thoughts on this?






-- 
J.D. Bronson // jeff at xpec.com // http://www.xpec dot com/jeff
Aurora Health Care // Information Systems // Telecom Site Manager
Milwaukee Wisconsin USA // Office: 414.978.8282 // Fax: 414.328.8282

Date: Wed, 28 Jun 2000 15:54:38 +0100 (BST)
From: "Yvonne.Hodson" <yhodso01 at bcuc.ac dot uk>
Subject: AUTHFILE in 3.0.2 on Solaris2.6

I am trying to compile 3.0.2 on Solaris 2.6 with the AUTHFILE option.
In 2.53 -DAUTHFILE="/.../...." was added to the O_DEFS in Makefile and worked fine.
In 3.0.2 there is no O_DEFS but similar OS_DEFS is set in configure.in
If I add -DAUTHFILE="/.../.." there, configure works fine but make aborts with
a syntax error
pop_pass.c line 93 syntax error before or at /
line 93 is
93      fp = fopen ( AUTHFILE, "r" );

Am I adding the macro in the wrong place? Where should it be?
Any help appreciated.

Yvonne Hodson
Computer Support Services
BCUC

Date: Wed, 28 Jun 2000 11:00:15 -0400 (EDT)
From: Ayan George <ayan at kiwi.ayan dot net>
Subject: -1 hidden messages.

Hello,

I subscribed to this list for information about a problem with qpop
v3.0.2 running on a production.

After authenticating, some users are recieving an +OK message
indicating they have -1 hidden messages.  I haven't been able to
identify the circumstances under which this occurs for Windows
users but it happens consistently for Mac users who use Outlook
Express.

I'm sure this has been discussed in the past so feel free to point
me to appropriate resources (list archives, RFCs, &c).

I'd like to know:

	1) What are hidden messages?  I don't see anything about
	them in any RFCs that I've read.

	2) How are messages marked as hidden?  ie, Where is this
	information stored?

	3) Is there a patch or a workaround for the -1 hidden
	messages response?

This version of popper is running on a production system with an
approximately 22k user burden.  I'd like to avoid running the beta
version if I can.

Any advice will be appreciated -- any advice regarding my mail
issues that is.

Thanks in advanced,

Ayan

From: Alexis Albinet <albinet at fluxus dot net>
Date: Wed, 28 Jun 2000 16:54:28 +0200 (CEST)
Subject: dot-lock file not deleted

Hi all,

I use sendmail 8.10 and qpopper 3.0.2 on FreeBSD 3.3.

When I compile the source of qpopper 3.0.2 and then use it, there is no problem.

If I use the FreeBSD ports of qpopper 3.0.2, the dot lock files (.login.pop) are not
removed after the end of pop session.

Any idea for this problem ?

Thanks in advance, 

Alexis


Date: Wed, 28 Jun 2000 11:38:09 -0400
From: Andrew Brent <dbrent at linkexchange dot com>
Subject: connection refused

I am rebuilding my mail server and have tried installing qpopper 3.0.2.  I 
previously used 3.0b35 with no problems.  This machine is a fresh install 
(FreeBSD 4.0).  here is what i get when i try to connect:

Trying xxx.xx.xxx.x...
Connected to xxx.xx.xxx.x.
Escape character is '^]'.
Connection closed by foreign host.

Any idea on what might be wrong?

I have tried this line in my inetd.conf

pop3    stream  tcp     nowait.400      root 
/usr/local/qpopper3.0.2/popper/ popper

i added the 400 in case it was the looping problem.  Sorry if this is in 
the archives, i am going through them but it is slow with no search tool.

-drew
*****************************************
Andrew Brent
bCentral
http://www.bCentral.com
*****************************************


From: Joseph Peterson <jay at qtm dot net>
Subject: RE: -1 hidden messages.
Date: Wed, 28 Jun 2000 11:49:26 -0400

I experienced that problem when the user's home directory was not owned
by them, chowning the directory fixed the problem.
This was also in conjunction with users who have used pine and have the
'FOLDER INTERNAL DATA' message (that is hidden from pop3 users)

-----Original Message-----
From: Ayan George [mailto:ayan at kiwi.ayan dot net]
Sent: Wednesday, June 28, 2000 11:00 AM
To: qpopper at lists.pensive dot org
Subject: -1 hidden messages.


Hello,

I subscribed to this list for information about a problem with qpop
v3.0.2 running on a production.

After authenticating, some users are recieving an +OK message
indicating they have -1 hidden messages.  I haven't been able to
identify the circumstances under which this occurs for Windows
users but it happens consistently for Mac users who use Outlook
Express.

I'm sure this has been discussed in the past so feel free to point
me to appropriate resources (list archives, RFCs, &c).

I'd like to know:

	1) What are hidden messages?  I don't see anything about
	them in any RFCs that I've read.

	2) How are messages marked as hidden?  ie, Where is this
	information stored?

	3) Is there a patch or a workaround for the -1 hidden
	messages response?

This version of popper is running on a production system with an
approximately 22k user burden.  I'd like to avoid running the beta
version if I can.

Any advice will be appreciated -- any advice regarding my mail
issues that is.

Thanks in advanced,

Ayan

Date: Wed, 28 Jun 2000 12:17:54 -0400 (EDT)
From: Ayan George <ayan at kiwi.ayan dot net>
Subject: Re: connection refused

The FreeBSD inetd is linked with libwrap and if invoked with the -w
flag checks connections against the /etc/hosts.allow file.

First be sure that inetd  is running with the -w flags (tcpwrappers
are a good thing) then make sure your /etc/hosts.allow file
allows connections.  Assuming you want to allow connections from
anywhere, you can add the line:

popper : ALL : allow

to the top of the file.

Definately read the hosts_access(5) man page for more info.

-Ayan

[ Quoted message from Andrew Brent, received Jun 28, 11:38am.]

> I am rebuilding my mail server and have tried installing qpopper 3.0.2.  I 
> previously used 3.0b35 with no problems.  This machine is a fresh install 
> (FreeBSD 4.0).  here is what i get when i try to connect:
> 
> Trying xxx.xx.xxx.x...
> Connected to xxx.xx.xxx.x.
> Escape character is '^]'.
> Connection closed by foreign host.
> 
> Any idea on what might be wrong?
> 
> I have tried this line in my inetd.conf
> 
> pop3    stream  tcp     nowait.400      root 
> /usr/local/qpopper3.0.2/popper/ popper
> 
> i added the 400 in case it was the looping problem.  Sorry if this is in 
> the archives, i am going through them but it is slow with no search tool.
> 
> -drew
> *****************************************
> Andrew Brent
> bCentral
> http://www.bCentral.com
> *****************************************
> 

[ End message from Andrew Brent. ]