I've been arguing with this off and on for months trying to upgrade my
old Cyrus installation, but I *cannot* get SASL2 to work.
All I want is for Cyrus to authenticate against /etc/sasldb2.db. That's
it. But it refuses to work.
My imapd.conf:
configdirectory: /var/imap
partition-default: /va
I run Cyrus on an OpenBSD 2.8 machine along with Postfix. About once
every week, Cyrus's lmtpd starts rejecting all connections to the unix
domain socket. The daemon is still running, but it refuses connections.
This happens no matter whether Postfix is delivering straight through lmtp
or if it'
On Thu, 15 Mar 2001, Gilion Goudsmit wrote:
>
> I used the following to send something
> to my mailbox:
>
> cat /etc/printcap | /usr/cyrus/bin/deliver-wrapper giel
> cat /etc/printcap | /usr/cyrus/bin/deliver-wrapper user.giel
> cat /etc/printcap | /usr/cyrus/bin/deliver giel
> cat /etc/printcap |
On Wed, 14 Mar 2001, The Hermit Hacker wrote:
>
> Trying to convince a group that Cyrus is more secure, but my arguments are
> about as lame as can be :(
>
> Does anyone have a URL that I can use in my args, that compares them
> better?
Well, there are two CERT advisories about older versions of
On Mon, 12 Mar 2001, Kaj J. Niemi wrote:
> On Mon, Mar 12, 2001 at 02:28 -0500, Bitt Faulk wrote:
> > On Mon, 12 Mar 2001, Kaj J. Niemi wrote:
> > >
> > > It works nicely however for some strange reason all my users get a folder
> > > named "user."
On Mon, 12 Mar 2001, Kaj J. Niemi wrote:
> On Mon, Mar 12, 2001 at 12:48 -0500, Bitt Faulk wrote:
>
> > A while back, I submitted a patch to provide the ability for user folders
> > (INBOX and others) to be autocreated on first login. I was wondering if
> > anyone was usi
A while back, I submitted a patch to provide the ability for user folders
(INBOX and others) to be autocreated on first login. I was wondering if
anyone was using it and if I could get some feedback from those of you
that are.
Thanks.
-Bitt
On Mon, 5 Mar 2001, Alain Turbide wrote:
>
> If you're using Outlook Express as a mail client then this is a known
> problem with Outlook Express. You might want to change the folder
> synchronization settings to none if you want to prevent this behaviour.
It's not an OE problem, although one of
On Tue, 6 Mar 2001, Luca Olivetti wrote:
>
> En/Na Bitt Faulk ha escrit:
>
> > If you're trying to implement a blavk-box solution, you'll have to go to a
> > lot of effort, using qmail's delivery defaults mechanisms in order to get
> > it to deliver t
Lately, I've noticed that things that my mail client has seen before get
marked as recent again. That is, I'll change folders to something else
and come back to my INBOX and there will be a number of messages that are
still marked as recent. I can't intentionally reproduce the problem, but
I hav
On Mon, 5 Mar 2001, Sam Stern wrote:
>
> I recently Migrated to qmail with maildirs and would like to install Cyrus
> 2.0.12 on my OpenBSD 2.8 system. However, I can't locate any relevant
> information of how Cyrus interoperates with Maildir.
Simply put, it doesn't. Cyrus has its own mailstore a
On Mon, 19 Feb 2001, Oguz YILMAZ wrote:
>
> when trying to connect to the imap server there is 15 seconds delay. I
> have tried 3 times.all connections had 15 seconds delay.
Usually, delays that last a specific amount of time are DNS problems or
identd problems. As I'm pretty sure that Cyrus doe
On Fri, 16 Feb 2001, Darren Nickerson wrote:
>
> Is the "cannot load CA data" important? Have I missed a step in the SSL setup?
Nope. There's a sort of virtual monopoly on certificate authorities (CAs)
whereby apps that have builtin SSL automatically trust the CAs that are
built into them. Thes
This might be obvious to everyone, but I thought I'd post it anyway.
I wanted to allow encrypted imap connections from everywhere, but
unencrypted sessions only from my internal network. TCP Wrappers seemed
an obvious solution, but since the current Cyrus code sets the daemon name
to check to th
On Fri, 9 Feb 2001, James Presley wrote:
>
> So I was reading the archives and saw someone had just copied the
> /usr/local/BerkeleyDB.3.2/lib/libdb.a file to the /usr/lib directory
> effectivly overwritting the old one. So I tried that and holy crap it
> worked! But when I tried to make I got
>
Oops. Forgot to tell you what needs to be set. In /etc/imapd.conf, you
need to set autocreatequota to something other than 0. If you want no
quotas, then set it to -1. If you want to create folders other than
just the INBOX, set autocreatefolders to them, separated by spaces. I
didn't impleme
Bitt Faulk wrote:
>
> I'll post the patch a little later tonight or tomorrow after I've had
> a little bit of a chance to test it against 2.0.11. Might as well
> add your suggestion first :-)
Okay. Done. It's not fully tested, but I'm pretty sure it's r
I've brought this up several times before, but I'd like to offer it
again. I think that the current ACL permission of CREATE for being able
to delete folders is better than the old right of DELETE, but I think
that it's still wrong.
Let me propose an example. Say that there is a shared folder c
Tristan Ball wrote:
>
> Yes, definately. Infact, a configurable list of mailboxes to create
> would be the ideal, as currently my user creation scripts also create
> an INBOX.trash.
That shouldn't be too hard to add on. It's a pretty good idea, too.
There needs to be another configuration opti
Bitt Faulk wrote:
>
> The master/Makefile links master with libwrap, and it doesn't need
> it. On my OpenBSD system, this actually causes runtime linking
> errors, as libwrap expects deny_severity and allow_severity to be
> supplied to it. The only place those symbols
I created a patch a long time ago (around 1.5.16) that allows an
administrator to set up their server so that INBOXes are automatically
created upon IMAP login. It seems to still work with a few mods under
the 2.0.* line.
Basically, it requires autocreatequota to exist in imapd.conf, and if it
d
Lawrence Greenfield wrote:
>
> Aha! This is it. limit_fds() calls syslog() before we reserve fd 3
> and 4 for communicating with the children.
>
> Thanks very much. Please try the attached patch. (The important
> part is that limit_fds() is called after reserving fds 0-4.)
Yup. That's it.
I found the problem. Sort of.
The usage of a static number for the STATUS_FD and LISTEN_FDs seem to be
causing problems. I can't define exactly how its doing it, but those
descriptors are being redefined somehow, and the data ends up being sent
to the wrong place. It also looks like there migh
More research:
I patched master.c to spew out its file descriptors using lsof via a
system() call, before and after the dup2()s. It's fine. However, the
imapd process that's running has those fds differently. It looks like
something was royally trodden upon. So, while I think that the rlim_t
The master/Makefile links master with libwrap, and it doesn't need it.
On my OpenBSD system, this actually causes runtime linking errors, as
libwrap expects deny_severity and allow_severity to be supplied to it.
The only place those symbols exist is in master/service.o, but that's
not linked in
Bitt Faulk wrote:
>
> Lawrence Greenfield wrote:
> >
> > Could you please try this patch?
>
> That seems to do it. :-)
I take it back. It doesn't have any apparent affect. I must be smoking
crack.
-Bitt
Lawrence Greenfield wrote:
>
> Could you please try this patch?
That seems to do it. :-)
-Bitt
Bitt Faulk wrote:
>
> A simple test program, listed below, seems to work better with rlim_t
> than int. All that I did was change the argument to limit_fds from
> an int to a rlim_t (line 6). Here's the outputs with int and rlim_t:
Forgot to mention that making that sim
Lawrence Greenfield wrote:
>
> Can someone with a FreeBSD/OpenBSD machine please read the
> documentation for setrlimit() and find out what's going wrong with the
> current code?
Well, the debug output says:
Feb 5 19:50:15 inky master[12991]: set maximum file descriptors to
256/0
Which would
Bitt Faulk wrote:
>
> Bitt Faulk wrote:
> >
> > I've compiled Cyrus 2.0.11 under OpenBSD. The master program runs,
> > but it will spawn each subserver only once.
>
> A little followup
>
> I've been doing some more research, and it looks l
Bitt Faulk wrote:
>
> I've compiled Cyrus 2.0.11 under OpenBSD. The master program runs,
> but it will spawn each subserver only once.
A little followup
I've been doing some more research, and it looks like the fd being used
to signal back to the master process is n
This is an odd one
I've compiled Cyrus 2.0.11 under OpenBSD. The master program runs, but
it will spawn each subserver only once. That is, I can telnet to port
143 and login and do stuff with the imap protocol once. But if it try
it again, it answers the port, but never prints out the bann
32 matches
Mail list logo
