headers
qpopper-request | 1 May 2003 09:01
Favicon

Qpopper Monthly Help File

         M O N T H L Y   Q P O P P E R   L I S T   H E L P

This is a routine monthly reminder of how to unsubscribe or use other
list commands.

First off, when you subscribe to a mailing list, you almost always get
a first message from the list admin telling you about the mailing list,
and explaining how to unsubscribe.  It is always a good idea to keep
these messages, since you never know when you will need to unsubscribe.
This is particularly useful when you change email addresses, because it
is difficult to unsubscribe from a list after you have a different
mailing address.

Every message sent by the list includes headers that you can just click
on to subscribe, unsubscribe, get help, or see the archives (if the
list has archives).

If you have tried this method, and the mailing list software won't let
you unsubscribe, it is probably because your address has changed.  In 
this case, please send a message to 
<mailto:listmaster <at> lists.pensive.org> stating which list (or lists) 
you want to unsubscribe from, and what you think your previous address
was.  There is a human (me) who will then try to take care of your 
request, often within a few days.

The list server in use (called AutoShare), offers several options which
you can set for yourself.  These options affect how the list sends you
messages.

The available options are:
(Continue reading)

Permalink | Reply | 0 comments |
headers
Ken Lalonde | 1 May 2003 22:10

Re: re outlook problems

If you're using Outlook XP and there is very little latency
between client and server, it sounds like the problem I posted
about last Feb:

  > From: Ken Lalonde <ken <at> globalremit.com>
  > Date: Tue, 04 Feb 2003 13:47:43 -0500
  > To: qpopper <at> lists.pensive.org
  > Subject: Re: I/O error flushing output to client
  >
  > We saw this problem fairly frequently.  The clients in question
  > were running Outlook XP, connecting to a qpopper-4.0.4 server
  > on the same very lightly loaded LAN.  Diddling with chunky-writes
  > or Outlook's timeout values did no good.
  >
  > A google search pointed to a bug in Outlook XP (shocking!),
  > wherein Outlook gets confused if the server responds "too quickly".
  > See the "I frequently get a timeout" question at:
  >   http://www.workgroupmail.com/faq_area.asp?area=Client%20Setup&id=34
  > and:
  >   http://home.hetnet.nl/~ojb-hamster/EnWIP/EnWeb/html/faq18om.htm
  >
  > The URLs above both suggest configuring the server to
  > pause for 50ms between POP commands.  I changed qpopper to do that,
  > which boils down to a "usleep(50*1000)" call in popper.c,
  > just before the pop_get_command() call.
  >
  > Haven't seen the problem since.
  >
  > I can provide a patch on request.
  >
(Continue reading)

Permalink | Reply | 0 comments |
headers
Simon May | 2 May 2003 08:53
Picon

Re: re outlook problems

> Outlook probably just rudely dropped the connection.
I guess that it's a matter of seeing if the sock is still active

Simon
Permalink | Reply | 0 comments |
headers
Simon May | 2 May 2003 08:56
Picon

Re: re outlook problems

> Because the client has disconnected. The output being flushed is to the
> IP connection, not to disk.
Then why is there still a qpopper process hanging in there.

Simon
Permalink | Reply | 0 comments |
headers
Daniel Senie | 2 May 2003 14:16

Re: re outlook problems

At 02:56 AM 5/2/2003, Simon May wrote:
> > Because the client has disconnected. The output being flushed is to the
> > IP connection, not to disk.
>Then why is there still a qpopper process hanging in there.

How does a process find out that the remote end of a TCP session has 
disconnected? When it tries to write to the socket, it gets told "operation 
not permitted..." Now, how long does TCP keep trying before it decides the 
other end of a connection has not only disconnected, but done so rudely by 
not saying goodbye? TCP has a retry mechanism. Until it stops retrying, it 
thinks the session is still up, and because of that, so does qpopper. How 
do you propose we learn faster that the remote has gone away?

You might try Ken Lahonde's method of adding a microsleep. He posted a note 
about that yesterday. Would be interesting to know if slowing down command 
responses did indeed fix your problems too.
Permalink | Reply | 0 comments |
headers
Randall Gellens | 4 May 2003 01:58

Security Vulnerability in Poppassd

A security vulnerability in the poppassd component of the Qpopper 
distribution has been discovered and publicized.  This vulnerability 
could permit an attacker to gain root access.  Note that the flaw 
requires shell access to to a system with poppassd installed. 
Poppassd is not built nor installed by default.

Work-arounds include removing the poppassd executable, or changing 
the permissions on the poppassd executable to prevent execution 
except by root.  Either of these methods should enable you to avoid 
the vulnerability.
Permalink | Reply | 1 comment |
headers
Simon Byrnand | 5 May 2003 00:31
Picon

Re: re outlook problems

At 08:16 2/05/03 -0400, Daniel Senie wrote:
>At 02:56 AM 5/2/2003, Simon May wrote:
>> > Because the client has disconnected. The output being flushed is to the
>> > IP connection, not to disk.
>>Then why is there still a qpopper process hanging in there.
>
>How does a process find out that the remote end of a TCP session has 
>disconnected? When it tries to write to the socket, it gets told 
>"operation not permitted..." Now, how long does TCP keep trying before it 
>decides the other end of a connection has not only disconnected, but done 
>so rudely by not saying goodbye? TCP has a retry mechanism. Until it stops 
>retrying, it thinks the session is still up, and because of that, so does 
>qpopper. How do you propose we learn faster that the remote has gone away?

It's not something that can be done at a Qpopper level....setting the 
system TCP retry timers lower helps a lot though. On linux I use:

echo 3 >/proc/sys/net/ipv4/tcp_retries1
echo 5 >/proc/sys/net/ipv4/tcp_retries2

(Original values are 7 and 15)

Cuts down the average timeout period from 15 minutes or more to a couple of 
minutes.

Regards,
Simon
Permalink | Reply | 2 comments |
headers
G G Papazoglou | 7 May 2003 21:03
Picon

local mailer location

************************************************************
ATTENTION PLEASE
Sorry. Your message did NOT get posted or forwarded
You cannot post to a list, unless you are a subscriber
The list server address is autoshare <at> lists.pensive.org
************************************************************

Hi and please forgive me for I know this isn't actually a qpopper issue, but
I suppose an issue of the local mailer,
but thought I'd find an answer in this group since I use qpopper 4.0.4.

I use Solaris 8, and by default the local mailer stores incoming mail
to /var/mail, whereas eg sent items are stored in $home/mail/sent-mail.
Could someone please let me know how can I change the local mailer's
setting to be storing incoming mail to eg. $home/mail/inbox?

And of course, should I change something in qpopper to be able to
start reading inboxes from the new location instead of /var/mail?
If so, does this have to be during compilation or there is a conf
file that can be edited?

Thank you
Grigorios
Permalink | Reply | 0 comments |
headers
Chuck Yerkes | 12 May 2003 07:07

Re: re outlook problems

Quoting Simon Byrnand (simon <at> igrin.co.nz):
...
> It's not something that can be done at a Qpopper level....setting the 
> system TCP retry timers lower helps a lot though. On linux I use:
> 
> echo 3 >/proc/sys/net/ipv4/tcp_retries1
> echo 5 >/proc/sys/net/ipv4/tcp_retries2
> 
> (Original values are 7 and 15)
> 
> Cuts down the average timeout period from 15 minutes or more to a couple of 
> minutes.

/me wonders what Stevens and the RFCs say about these reductions.
(I don't know, but I have fixed TCP settings that were reduced well
below the RFC requiremnts that cause problems on all but the most
ideal connections).

As someone said, you need a sniff of the offending traffic.
If that's snoop or tcpdump from the host to port 110 of the server,
then you get actual information to debug.

If it's easy to reproduce, then it won't take much time, if not,
you may end up running long snoops.  Perhaps you start up a new
one every period and rotate snoop files (saving the last couple).
Or you just snoop with a machine with a very large disk.

There are, by my recollection, at least 6 unique Outlook behaviors,
at least from the attachment handling perspective.
(Continue reading)

Permalink | Reply | 0 comments |
headers
Mubashir Cheema | 15 May 2003 21:52

qpopper not reading message status set by others?

I am new to qpopper.  So please forgive me if this is a FAQ.  Although I
have read the FAQ and browsed through the mailing list archive and did not
find an answer to this question.

I am running qpopper on a RedHat 9 system using xinetd.  Mail is read by
users using imap, pop or local mail client like pine and mutt.

It seems that qpopper is not reading the status correctly or not at all
when it is set by other software.  So if the user uses reads their email
locally using pine and then later connects using pop and only interested
in downlading new messages, qpopper ends up serving them all the messages
that it itself has not marked read.  uw-imap does not do this.

I have briefuly scanned the qpopper administrator guide but did not see an
obvious solution to this problem.

If anybody has any solutions to this problem, I'd be happy to hear from
them.

Thanks.

-Cheema
Permalink | Reply | 0 comments |

Gmane