[courier-users] Re: Courier IMAP running but not working
Sam Varshavchik
> I run the imapd.rc and imapd-ssl.rc. Beautiful; no errors. Try
> telneting to port 143, or 993, though, and
>
> telnet: connect to address 127.0.0.1: Connection refused
Did you remember to run 'make install-configure'?
> One more question, does courier require the maildir name to be
> "Maildir"? I would like to maybe change it to "INBOX" or something
> for asthetic's sake....
If you want to globally change the maildir location from $HOME/Maildir to
$HOME/INBOX, just change imapd.rc and imapd-ssl.rc
--
Sam
_______________________________________________
courier-users mailing list
courie...@lists.sourceforge.net
http://lists.sourceforge.net/lists/listinfo/courier-users
Drew Raines
> Did you remember to run 'make install-configure'?
yes. Except when I did it with `make' it gave me
*** Error code 2
Stop in /home/raines/courier-imap-1.3.7 (line 961 of Makefile).
so I used `gmake' and it seemed to work fine. Would there be a
problem with using gmake?
I think I'm following the INSTALL to the tee, but I can't figure out
why it doesn't bind the port. I checked /etc/services and both 143
and 993 are defined. If it is binded to the port, it's not answering
requests.
> If you want to globally change the maildir location from $HOME/Maildir to
> $HOME/INBOX, just change imapd.rc and imapd-ssl.rc
gee.. real difficult.. :)
--
Andrew A. Raines | dr...@phg.mc.vanderbilt.edu | +1 615 343 5853
program in human genetics | vanderbilt university medical center
Charlie Watts
> * Sam Varshavchik <mr...@courier-mta.com>, 15:21 05/05/2001:
> > Did you remember to run 'make install-configure'?
>
> yes. Except when I did it with `make' it gave me
>
> *** Error code 2
>
> Stop in /home/raines/courier-imap-1.3.7 (line 961 of Makefile).
>
> so I used `gmake' and it seemed to work fine. Would there be a
> problem with using gmake?
>
> I think I'm following the INSTALL to the tee, but I can't figure out
> why it doesn't bind the port. I checked /etc/services and both 143
> and 993 are defined. If it is binded to the port, it's not answering
> requests.
gmake is preferred.
From INSTALL:
* make - The GNU make is recommended. Solaris's make is to be
avoided. xBSD already has a gmake port, install it and use it (use
gmake everywhere this document refers to make).
Chad Leigh, Shire.Net LLC
is inetd already listening on those ports (but with no valid daemon to take
the connection)?
inetd cannot serve those ports when courier is...
Chad
--On Saturday, May 05, 2001 4:20 PM -0500 Drew Raines
<dr...@phg.mc.vanderbilt.edu> wrote:
> * Sam Varshavchik <mr...@courier-mta.com>, 15:21 05/05/2001:
>> Did you remember to run 'make install-configure'?
>
> yes. Except when I did it with `make' it gave me
>
> *** Error code 2
>
> Stop in /home/raines/courier-imap-1.3.7 (line 961 of Makefile).
>
> so I used `gmake' and it seemed to work fine. Would there be a
> problem with using gmake?
>
> I think I'm following the INSTALL to the tee, but I can't figure out
> why it doesn't bind the port. I checked /etc/services and both 143
> and 993 are defined. If it is binded to the port, it's not answering
> requests.
>
>
>
>> If you want to globally change the maildir location from $HOME/Maildir
>> to $HOME/INBOX, just change imapd.rc and imapd-ssl.rc
>
> gee.. real difficult.. :)
>
>
> --
> Andrew A. Raines | dr...@phg.mc.vanderbilt.edu | +1 615 343 5853
> program in human genetics | vanderbilt university medical center
>
Drew Raines
>
> is inetd already listening on those ports (but with no valid daemon to take
> the connection)?
>
> inetd cannot serve those ports when courier is...
no.. in fact I killed inetd just to make sure.
I compiled everything with gmake now. 'gmake check' worked, so did
install and install-configure. I run the rc scripts and it starts
without flaw. Alas, nothing doing. I still can get to ports 143 and
993.
I even tried nmap-ing the machine and it doesn't scan anything on
those ports either. I fear that OpenBSD in its "security by default"
has disabled something of which I'm unaware.
Drew Raines
> without flaw. Alas, nothing doing. I still can get to ports 143 and
> 993.
um, should be "..canNOT get to ports 143 and 993."
Drew Raines
>
> None whatsoever... Sorry! LOL ... good luck.
>
> Hrm ... when inetd binds to ports [ telnet, forex ] what does that look
> like in netstat -a ? Does that say "tcp" or "tcp6" ? Perhaps tcp6 is
> normal and it is just a firewalling problem.
>
well I snipped some of the output. There's an ssh for tcp and tcp6 so
I'm assuming the only reason I can use ssh is because it uses tcp. I
need to figure out how to get imap on tcp and not tcp6. I'm a total
openbsd newbie.
-d
# netstat -a
Active Internet connections (including servers)
Proto Recv-Q Send-Q Local Address Foreign Address (state)
tcp 0 0 mays.ssh ci903556-a.nash1.1956 ESTABLISHED
tcp 0 0 *.smtp *.* LISTEN
tcp 0 472 mays.ssh ci903556-a.nash1.1445 ESTABLISHED
tcp 0 0 *.ssh *.* LISTEN
tcp 0 0 localhost.sunrpc *.* LISTEN
tcp 0 0 *.sunrpc *.* LISTEN
udp 0 0 mays.ntp *.*
udp 0 0 localhost.ntp *.*
udp 0 0 *.ntp *.*
udp 0 0 localhost.sunrpc *.*
udp 0 0 *.sunrpc *.*
udp 0 0 *.syslog *.*
tcp6 0 0 *.imaps *.* LISTEN
tcp6 0 0 *.imap *.* LISTEN
tcp6 0 0 *.ssh *.* LISTEN
Larry Moore
I have compiled 1.3.8 with OpenBSD 2.7 and I can't connect to port 143 or
993 however "netstat -anf inet|grep 143" reveals that courier is listening
on tcp6 so my ip-version 4 connections don't make it.
I used to have courier-imap working on this system but I can't remeber
which version - I think it was 1.3.0.
Anyway, I have compiled and installed versions of Courier-IMAP down to
1.3.5. I'll have to find the sources for 1.3.0 and try that again before
deciding its to do with header files on my system that are the cause of the
problem.
Cheers,
Larry.
----- Original Message -----
From: "Drew Raines" <dr...@phg.mc.vanderbilt.edu>
To: <courie...@lists.sourceforge.net>
Sent: Sunday, May 06, 2001 7:05 AM
Subject: Re: [courier-users] Re: Courier IMAP running but not working
> * "Chad Leigh, Shire.Net LLC" <ch...@shire.net>, 17:52 05/05/2001:
> >
> > is inetd already listening on those ports (but with no valid daemon to
take
> > the connection)?
> >
> > inetd cannot serve those ports when courier is...
>
> no.. in fact I killed inetd just to make sure.
>
> I compiled everything with gmake now. 'gmake check' worked, so did
> install and install-configure. I run the rc scripts and it starts
> without flaw. Alas, nothing doing. I still can get to ports 143 and
> 993.
>
> I even tried nmap-ing the machine and it doesn't scan anything on
> those ports either. I fear that OpenBSD in its "security by default"
> has disabled something of which I'm unaware.
>
>
> --
> Andrew A. Raines | dr...@phg.mc.vanderbilt.edu | +1 615 343 5853
> program in human genetics | vanderbilt university medical center
>
Sam Varshavchik
> I'm assuming the only reason I can use ssh is because it uses tcp. I
> need to figure out how to get imap on tcp and not tcp6.
Why? IPv6 socket API is completely backwards compatible to IPv4.
> tcp6 0 0 *.imaps *.* LISTEN
> tcp6 0 0 *.imap *.* LISTEN
> tcp6 0 0 *.ssh *.* LISTEN
According to RFC 2553 (3.7) these sockets should accept both IPv4 and IPv6
connections.
--
Sam
Drew Raines
>
> Why? IPv6 socket API is completely backwards compatible to IPv4.
>
hmm.. then why does openssh bind listen with tcp and tcp6? wouldn't
it just use tcp6... I use that as an example because that's "oem"
openbsd.
>
> According to RFC 2553 (3.7) these sockets should accept both IPv4 and IPv6
> connections.
hmm... well maybe somebody's not implementing something correctly.
From all these mailing lists I'm on, I'm starting to believe that
you're the only one that reads RFCs anyway :)
--
Andrew A. Raines | dr...@phg.mc.vanderbilt.edu | +1 615 343 5853
program in human genetics | vanderbilt university medical center
_______________________________________________
Sam Varshavchik
>>
>> According to RFC 2553 (3.7) these sockets should accept both IPv4 and IPv6
>> connections.
>
> hmm... well maybe somebody's not implementing something correctly.
> From all these mailing lists I'm on, I'm starting to believe that
> you're the only one that reads RFCs anyway :)
If your IPv6 stack is broken, --without-ipv6 will disable it.
--
Sam
Drew Raines
>
> If your IPv6 stack is broken, --without-ipv6 will disable it.
>
>
After configuring and compiling with that option, I still get the same
result. Here's what I did in detail:
$ rm -rf courier-imap-1.3.8
$ tar zxf courier-imap-1.3.8.tar.gz
$ cd courier-imap-1.3.8
$ ./configure --without-ipv6; gmake; gmake check
$ su
# gmake install
# gmake install-configure
# /usr/lib/courier-imap/libexec/imapd.rc start
# netstat -a | grep imap
tcp6 0 0 *.imap *.* LISTEN
do I need to do something else?
--
Andrew A. Raines | dr...@phg.mc.vanderbilt.edu | +1 615 343 5853
program in human genetics | vanderbilt university medical center
_______________________________________________
Sam Varshavchik
> * Sam Varshavchik <mr...@courier-mta.com>, 21:53 05/05/2001:
>>
>> If your IPv6 stack is broken, --without-ipv6 will disable it.
>>
>>
>
> After configuring and compiling with that option, I still get the same
> result. Here's what I did in detail:
>
> $ rm -rf courier-imap-1.3.8
> $ tar zxf courier-imap-1.3.8.tar.gz
> $ cd courier-imap-1.3.8
> $ ./configure --without-ipv6; gmake; gmake check
> $ su
> # gmake install
> # gmake install-configure
> # /usr/lib/courier-imap/libexec/imapd.rc start
> # netstat -a | grep imap
> tcp6 0 0 *.imap *.* LISTEN
>
>
> do I need to do something else?
Check if RFC1035_IPV6 is set to 0 or 1 in rfc1035/rfc1035.h
--
Sam
Larry Moore
now permits it to listen for version 4 connections and is working fine from
the basic tests I have performed including SSL.
Cheers,
Larry.
----- Original Message -----
From: "Drew Raines" <dr...@phg.mc.vanderbilt.edu>
To: <courie...@lists.sourceforge.net>
Sent: Sunday, May 06, 2001 12:28 PM
Subject: Re: [courier-users] Re: Courier IMAP running but not working
> * Sam Varshavchik <mr...@courier-mta.com>, 21:53 05/05/2001:
> >
> > If your IPv6 stack is broken, --without-ipv6 will disable it.
> >
> >
>
> After configuring and compiling with that option, I still get the same
> result. Here's what I did in detail:
>
> $ rm -rf courier-imap-1.3.8
> $ tar zxf courier-imap-1.3.8.tar.gz
> $ cd courier-imap-1.3.8
> $ ./configure --without-ipv6; gmake; gmake check
> $ su
> # gmake install
> # gmake install-configure
> # /usr/lib/courier-imap/libexec/imapd.rc start
> # netstat -a | grep imap
> tcp6 0 0 *.imap *.* LISTEN
>
>
> do I need to do something else?
>
> --
> Andrew A. Raines | dr...@phg.mc.vanderbilt.edu | +1 615 343 5853
> program in human genetics | vanderbilt university medical center
>
Larry Moore
I wonder if you are stilll using old binaries in /usr/lib/courier-imap.
Might be worth going into this directory and removing all sub-directories
before you install the binaries.
It appears that the behaviour with tcp6 occurred after version 1.3.0 of
courier-imap because if I build this version it only listens on tcp version
4 socket.
Here is how I configured it on my system (I am using bash for my shell).
Firstly I made an OBJ directory in courier-imap-1.3.8 source directory then
made the directory obsd2.7-sparc32 (OS & hardware platform specific
directory as I build for Solaris too) in OBJ.
Next, cd obsd2.7-sparc32 then ran configure as such (you may want to alter
the socks option and -m in LDFLAGS to suit you particular CPU),
-------------------------------------------------------
export CFLAGS="-mv8 -I/usr/local/include"
export CPPFLAGS=-I/usr/local/include
export LDFLAGS=-L/usr/local/lib
(N.B. the next entry was all on one line)
../../configure --without-ipv6 --with-db=db --with-socks --sysconfdir=/etc/
courier --with-userdb=/etc/courier/userdb --localstatedir=/var/courier --pre
fix=/usr/local --libexecdir=/usr/local/libexec/courier-imap --datadir=/usr/l
ocal/share/courier-imap
make && gmake check
su
make install && gmake install-configure
(adjust /etc/courier/impad.cnf and pop3d.cnf to suit)
/usr/local/sbin/mkimpadcert
/usr/local/sbin/mkpop3dcert
mkdir /var/courier
/usr/local/libexec/courier-imap/imapd.rc start
/usr/local/libexec/courier-imap/impad-ssl.rc start
---------------------------------------------------------
Hope this is of help to you.
Cheers,
Larry.
----- Original Message -----
From: "Drew Raines" <dr...@phg.mc.vanderbilt.edu>
To: "Charlie Watts" <cew...@frontier.net>
Cc: <courie...@lists.sourceforge.net>; <mi...@openbsd.org>
Sent: Sunday, May 06, 2001 9:02 AM
Subject: [courier-users] Re: Courier IMAP running but not working
> * Charlie Watts <cew...@frontier.net>, 18:48 05/05/2001:
> >
> > None whatsoever... Sorry! LOL ... good luck.
> >
> > Hrm ... when inetd binds to ports [ telnet, forex ] what does that look
> > like in netstat -a ? Does that say "tcp" or "tcp6" ? Perhaps tcp6 is
> > normal and it is just a firewalling problem.
> >
>
> well I snipped some of the output. There's an ssh for tcp and tcp6 so
> I'm assuming the only reason I can use ssh is because it uses tcp. I
> need to figure out how to get imap on tcp and not tcp6. I'm a total
> openbsd newbie.
>
> -d
>
>
> # netstat -a
> Active Internet connections (including servers)
> Proto Recv-Q Send-Q Local Address Foreign Address (state)
> tcp 0 0 mays.ssh ci903556-a.nash1.1956
ESTABLISHED
> tcp 0 0 *.smtp *.* LISTEN
> tcp 0 472 mays.ssh ci903556-a.nash1.1445
ESTABLISHED
> tcp 0 0 *.ssh *.* LISTEN
> tcp 0 0 localhost.sunrpc *.* LISTEN
> tcp 0 0 *.sunrpc *.* LISTEN
> udp 0 0 mays.ntp *.*
> udp 0 0 localhost.ntp *.*
> udp 0 0 *.ntp *.*
> udp 0 0 localhost.sunrpc *.*
> udp 0 0 *.sunrpc *.*
> udp 0 0 *.syslog *.*
> tcp6 0 0 *.imaps *.* LISTEN
> tcp6 0 0 *.imap *.* LISTEN
> tcp6 0 0 *.ssh *.* LISTEN
>
Drew Raines
>
> Check if RFC1035_IPV6 is set to 0 or 1 in rfc1035/rfc1035.h
>
Indeed, it is set to 0.
---
#define RFC1035_IPV6 0
---
--
Andrew A. Raines | dr...@phg.mc.vanderbilt.edu | +1 615 343 5853
program in human genetics | vanderbilt university medical center
_______________________________________________
Sam Varshavchik
> Notice that ssh is listening on both tcp and tcp6.
Someone should really fix this. It's quite clear that an IPv6 socket must
accept both IPv4 and IPv6 connections.
--
Sam
Yozo TODA
> Someone should really fix this. It's quite clear that an IPv6 socket must
> accept both IPv4 and IPv6 connections.
OpenBSD does not conform to RFC2553 because of security reasons.
see the section "Interaction between IPv4/v6 sockets" of
the manpage inet6(4) of OpenBSD.
use (http://www.openbsd.org/cgi-bin/man.cgi) if you need...
aha, I found NetBSD also has the same document as OpenBSD.
I wonder this comes from KAME project?
actually I see some of KAME developers saying (on some mailing list)
IPv4 mapped addressing is very ugly and confusing idea.
you should open two sockets, one for IPv4, and one for IPv6.
anyone know the status of IETF activity related to IPv6 API?
I suppose RFC2553 will be "fixed" in the future?
-- yozo.
Sam Varshavchik
>
>> Someone should really fix this. It's quite clear that an IPv6 socket must
>> accept both IPv4 and IPv6 connections.
>
> OpenBSD does not conform to RFC2553 because of security reasons.
> see the section "Interaction between IPv4/v6 sockets" of
> the manpage inet6(4) of OpenBSD.
> use (http://www.openbsd.org/cgi-bin/man.cgi) if you need...
>
> aha, I found NetBSD also has the same document as OpenBSD.
> I wonder this comes from KAME project?
I have no idea where that came from. I can't find anything in RFC 2553 that
suggests anything close to what OpenBSD's man page claims it says. I
certainly can't find anything in sections 3.6, 3.7 and 3.8 - which discuss
the interoperability between IPv4 and IPv6 within the same system (and
including the wildcard bind).
> actually I see some of KAME developers saying (on some mailing list)
> IPv4 mapped addressing is very ugly and confusing idea.
> you should open two sockets, one for IPv4, and one for IPv6.
I find nothing ugly and confusing about IPv4-mapped addresses. The way I
see it, AF_INET4 is just a subset of AF_INET6 addresses. Since you can't
have multiple binds on the same address (unless both sockets are marked
SO_REUSEADDR), if you have an outstanding wildcard bind in AF_INET4, you
should not be able to do a wildcard bind on the same port in AF_INET6, since
you'll essentially be trying to bind something a portion of which is already
bound. However, you should be able to bind the same port for a specific
AF_INET6 address (not IPV4MAPPED), since it's not in the scope of the
wildcard AF_INET4 bind. The opposite is true - you shouldn't be able to do
a wildcard bind in AF_INET4 if there's already an existing wildcard bind in
AF_INET6 address space (which will end up getting all PF_INET4 connections).
In fact, you shouldn't be able to bind anything on the same port in
AF_INET4.
This is, in fact, how it works in Linux's stack. In Linux you can't create
an AF_INET6 and AF_INET4 wildcards on the same port. Either one will lock
the other one out. If you want to handle both IPv4 and IPv6 you don't go
and create two sockets. RFC 2553 already tells you that PF_INET6 sockets
will take PF_INET4 connections, so if you want both PF_INET4 and PF_INET6,
just create a PF_INET6 socket and move on with your life. WTF do you need a
PF_INET4 socket for??
>
> anyone know the status of IETF activity related to IPv6 API?
> I suppose RFC2553 will be "fixed" in the future?
I really don't see anything in RFC 2553 that's in an urgent need of fixing.
--
Sam