Discussion:
[courier-users] How to configure maildir
Alfredo Martino
2002-07-02 07:41:57 UTC
Permalink
Hi to all, I'm a new courier users.

I've installed Courier-imapd in my freebsd server with sendmail.
I've tried to create new user with adduser but this user can't read email via pop3
because the maildir is not correct.

I wanto to know if courier work fine with sendmail or if i have to install qmail.
If courier work fine with sendmail how can I create new user ? I've to create first user
with adduser and then makemaildir ?

Can you help me with a little example for adding new user in my freebsd and configuring
it to read email via pop3.


Thanks in advace.
Alfredo Martino
***@guest.tlsoft.it


".... perche' TIM e' grande ...."
Bill Michell
2002-07-02 08:06:58 UTC
Permalink
Alfredo Martino writes:

> Hi to all, I'm a new courier users.
>
> I've installed Courier-imapd in my freebsd server with sendmail.
> I've tried to create new user with adduser but this user can't read email via pop3
> because the maildir is not correct.
>
> I wanto to know if courier work fine with sendmail or if i have to install qmail.
> If courier work fine with sendmail how can I create new user ? I've to create first user
> with adduser and then makemaildir ?
>
> Can you help me with a little example for adding new user in my freebsd and configuring
> it to read email via pop3.
>
>
The command you need to find out about is maildirmake.

This makes an empty maildir structure for the user.

On at least some systems, it is perfectly feasible to add a suitable maildir
to the dummy user so it will be automatically created for new users with
adduser. I'm not sure about the details for freebsd though...

--
Bill Michell
***@mics.org.uk (home)
M. Jolic
2002-07-02 09:19:01 UTC
Permalink
Hi Gurus,

i've searched the list but nothing found similar.

I've got a problem to log the user WITH domain in the Logs.

The normal way shows the user only.

from imapd.c
[...]
fprintf(stderr, "INFO: LOGIN, user=%s, ip=[%s]\n",
getenv("AUTHENTICATED"), ip);
[...]

maillog:
[...]
Jul 1 22:10:37 HOST imapd-ssl: Connection, ip=[xxx.xxx.xxx.xxx]
Jul 1 22:10:37 HOST imapd-ssl: LOGIN, user=user1, ip=[xxx.xxx.xxx.xxx]
Jul 1 22:10:44 HOST imapd-ssl: LOGOUT, user=user1, ip=[xxx.xxx.xxx.xxx],
headers=76, body=4477
[...]
(btw: HOST is the name of the server)

So, how get we logged the virtual domain?

Like, for ex.:

user1 is a user from virtual domain example.com.

It could look like:

maillog:
[...]
Jul 1 22:10:37 HOST imapd-ssl: Connection, ip=[xxx.xxx.xxx.xxx]
Jul 1 22:10:37 HOST imapd-ssl: LOGIN, user=user1, domain=example.com,
ip=[xxx.xxx.xxx.xxx]
Jul 1 22:10:44 HOST imapd-ssl: LOGOUT, user=user1, domain=example.com,
ip=[xxx.xxx.xxx.xxx], headers=76, body=4477
[...]

thx.

m.
M. Jolic
2002-07-02 10:56:50 UTC
Permalink
>
> make sure your users login with their full name + email address.
>

They do!

Ex: login: ***@example.com pass: pass

> e.g. ***@example.com
>
> that way it gets logged with the vhost.

No.

When somebody uses (for ex.) outlook and the login is defined as:
***@example.com (+ his password) we get this LOG:

Jul 1 22:10:37 HOST imapd-ssl: LOGIN, user=user1, ip=[xxx.xxx.xxx.xxx]

NOT user=***@example.com ::: NO host-info!!!

Do you get something else?

>
> the problem with your vhost setup is, is that you can't have two of the same
> accounts for
> different domains,
>
> e.g. ***@example.com
> ***@foo.bar

This is handled trough vmailmgrd!

We have ***@virtualdom1.com, ***@virtualdom2.com, ***@virtualdom3.com,
***@virtualdom4.com, ***@virtualdom5.com... and so on!

Every user has to login with his FULL login-name like mentioned above.

And everything works as expected!

But we don't see any domain in the logs!

>
> i use ldap is the authentication backend and altered the me file
> /courier/etc to change the domain name to something other then
> the default domain, as it gets appended by default to the username and
> authentication fails.
>
> also i have created a hostedddomains file with domains that i host.

Is 'hosteddomains' handled by courier? If yes, how?

***

Any comments?

thx.

m.

>
> Mail me if you want more info,
>
> With kind regards,
>
> Tjeerd van der Zee
>
> ----- Original Message -----
> From: "M. Jolic" <***@primusavitos.de>
> To: <courier-***@lists.sourceforge.net>
> Sent: Tuesday, July 02, 2002 1:16 PM
> Subject: [courier-users] courier-imap and virtual domains with vmailmgr &
> qmail...
>
>
>> Hi Gurus,
>>
>> i've searched the list but nothing found similar.
>>
>> I've got a problem to log the user WITH domain in the Logs.
>>
>> The normal way shows the user only.
>>
>> from imapd.c
>> [...]
>> fprintf(stderr, "INFO: LOGIN, user=%s, ip=[%s]\n",
>> getenv("AUTHENTICATED"), ip);
>> [...]
>>
>> maillog:
>> [...]
>> Jul 1 22:10:37 HOST imapd-ssl: Connection, ip=[xxx.xxx.xxx.xxx]
>> Jul 1 22:10:37 HOST imapd-ssl: LOGIN, user=user1, ip=[xxx.xxx.xxx.xxx]
>> Jul 1 22:10:44 HOST imapd-ssl: LOGOUT, user=user1, ip=[xxx.xxx.xxx.xxx],
>> headers=76, body=4477
>> [...]
>> (btw: HOST is the name of the server)
>>
>> So, how get we logged the virtual domain?
>>
>> Like, for ex.:
>>
>> user1 is a user from virtual domain example.com.
>>
>> It could look like:
>>
>> maillog:
>> [...]
>> Jul 1 22:10:37 HOST imapd-ssl: Connection, ip=[xxx.xxx.xxx.xxx]
>> Jul 1 22:10:37 HOST imapd-ssl: LOGIN, user=user1, domain=example.com,
>> ip=[xxx.xxx.xxx.xxx]
>> Jul 1 22:10:44 HOST imapd-ssl: LOGOUT, user=user1, domain=example.com,
>> ip=[xxx.xxx.xxx.xxx], headers=76, body=4477
>> [...]
>>
>> thx.
>>
>> m.
>>
>>
>>
>> -------------------------------------------------------
>> This sf.net email is sponsored by:ThinkGeek
>> Welcome to geek heaven.
>> http://thinkgeek.com/sf
>> _______________________________________________
>> courier-users mailing list
>> courier-***@lists.sourceforge.net
>> Unsubscribe: https://lists.sourceforge.net/lists/listinfo/courier-users
>>
Sam Varshavchik
2002-07-02 11:20:41 UTC
Permalink
M. Jolic writes:


> When somebody uses (for ex.) outlook and the login is defined as:
> ***@example.com (+ his password) we get this LOG:
>
> Jul 1 22:10:37 HOST imapd-ssl: LOGIN, user=user1, ip=[xxx.xxx.xxx.xxx]
>
> NOT user=***@example.com ::: NO host-info!!!

File a bug report with microsoft. The IMAP mail client completely controls
the userid it sends to the server.


--
Sam
M. Jolic
2002-07-02 11:55:30 UTC
Permalink
> When somebody uses (for ex.) outlook and the login is defined as:
>> ***@example.com (+ his password) we get this LOG:
>>
>> Jul 1 22:10:37 HOST imapd-ssl: LOGIN, user=user1, ip=[xxx.xxx.xxx.xxx]
>>
>> NOT user=***@example.com ::: NO host-info!!!
>
> File a bug report with microsoft. The IMAP mail client completely controls
> the userid it sends to the server.

But there's only one problem:

Our Users have to login with ***@domain.com with their password and this
has to be transmitted, otherwise our server wouldn't allow to log in.

So, ***@domain.com is transmitted and via the authvmailmgr authenticated
as mentioned on www.inter7.com (->free software->courier-imap).

Maybe this could be a solution:

In imapd.c:

SetENV("DAMNEDVIRTUALUSER")=user;

and further on where the code will log through syslog:

fprintf(stderr, "INFO: LOGIN, user=%s, ip=[%s]\n",
getenv("DAMNEDVIRTUALUSER"), ip);

Is this possible?

m.
Sam Varshavchik
2002-07-02 12:43:06 UTC
Permalink
M. Jolic writes:


> Our Users have to login with ***@domain.com with their password and this
> has to be transmitted, otherwise our server wouldn't allow to log in.
>
> So, ***@domain.com is transmitted and via the authvmailmgr authenticated
> as mentioned on www.inter7.com (->free software->courier-imap).

Sounds like a vmailmgr issue. Courier-IMAP has no issue authenticating
virtual login IDs with fully-qualified domain names.

--
Sam
Roland Schneider
2002-07-02 12:40:05 UTC
Permalink
--M. Jolic wrote on 02.07.2002 15:44 +0200:

> So, ***@domain.com is transmitted and via the authvmailmgr authenticated
> as mentioned on www.inter7.com (->free software->courier-imap).
>
> Maybe this could be a solution:
>
> In imapd.c:
>
> SetENV("DAMNEDVIRTUALUSER")=user;
>
> and further on where the code will log through syslog:
>
> fprintf(stderr, "INFO: LOGIN, user=%s, ip=[%s]\n",
> getenv("DAMNEDVIRTUALUSER"), ip);
>
> Is this possible?

courier logs whatever the auth-module puts into $AUTHENTICATED:

fprintf(stderr, "INFO: LOGOUT, user=%s, ip=[%s],"
" headers=%lu, body=%lu\n",
getenv("AUTHENTICATED"), getenv("TCPREMOTEIP"),


Use one of the auth-module which comes with courier, or fix
vmailmgr (which also has a couple of other problems reported).

Roland
M. Jolic
2002-07-02 13:16:04 UTC
Permalink
> courier logs whatever the auth-module puts into $AUTHENTICATED:
>
> fprintf(stderr, "INFO: LOGOUT, user=%s, ip=[%s],"
> " headers=%lu, body=%lu\n",
> getenv("AUTHENTICATED"), getenv("TCPREMOTEIP"),
>
>
> Use one of the auth-module which comes with courier, or fix
> vmailmgr (which also has a couple of other problems reported).
>

Do you know about other problems? What kind of?

The problem is:

We're using qmail with vmailmgrd (for virtual hosts) with pop & smtp. So we
decided to use courier-imap as an additional module because it's faster than
pop and we like the possbility to keep messages in folders or to mark them
as seen/deleted...

But we don't want to have a conf for vmailmgr (qmail) and another for imap.

We're just talking about a solution with qmail & SQL where every user (vdoms
too) is listed in the SQL-table. There are solution for qmail to handle this
(auth via SQL).

The Benefit is that users who want to connect via imap could be
authenticated through the auth-SQL too.

So we have only ONE database for the whole mail-system.

Is anybody using this? (or similiar?)

milli
Toni Mattila
2002-07-02 13:19:27 UTC
Permalink
Hi,

> > Our Users have to login with ***@domain.com with their password and this
> > has to be transmitted, otherwise our server wouldn't allow to log in.
> >
> > So, ***@domain.com is transmitted and via the authvmailmgr authenticated
> > as mentioned on www.inter7.com (->free software->courier-imap).
> Sounds like a vmailmgr issue. Courier-IMAP has no issue authenticating
> virtual login IDs with fully-qualified domain names.

So far I haven't met any issues with vmailmgr-auth module. It's just that
in logfiles the domain is stripped from the username. So the auth works
100%, only thing is that the info doesn't get in full to the logfile.

Regards,
Toni Mattila
Roland Schneider
2002-07-02 14:03:13 UTC
Permalink
--M. Jolic wrote on 02.07.2002 17:05 +0200:

>> courier logs whatever the auth-module puts into $AUTHENTICATED:
>>
>> fprintf(stderr, "INFO: LOGOUT, user=%s, ip=[%s],"
>> " headers=%lu, body=%lu\n",
>> getenv("AUTHENTICATED"), getenv("TCPREMOTEIP"),
>>
>>
>> Use one of the auth-module which comes with courier, or fix
>> vmailmgr (which also has a couple of other problems reported).
>>
>
> Do you know about other problems? What kind of?

Authentification only works a few times and such things...
Check out the archives and consult the vpopmail-mailinglist.

$ grep -ihE "^Subject:.*(vmail|vpopmail|vchkp)" * | sort | uniq

Subject: RE: [courier-users] Courier-IMAP+vpopmail
Subject: RE: [courier-users] Re: Vpopmail READ THIS
Subject: Re: [courier-users] Qmail, Courier-imap, and vmailmgr
Subject: [courier-users] Courier IMAP + Qmail + Vpopmail + PHP
Subject: [courier-users] Courier imap-1.4.5 and vpopmail-5.3.6
Subject: [courier-users] Courier-IMAP 1.3.8 + Vpopmail 5.2 problems with
default domain
Subject: [courier-users] Courier-IMAP+vpopmail
Subject: [courier-users] Courier-IMAP+vpopmail Msg2
Subject: [courier-users] Re: Vpopmail READ THIS
Subject: [courier-users] Re: Vpopmail READ THIS
Subject: [courier-users] Re: courier-imap needs restart to auth vpopmail users
Subject: [courier-users] User authenticated but unable to read new mail
(vpopmail 5.2 +courier 1.4.4)
Subject: [courier-users] courier-imap needs restart to auth vpopmail users
Subject: [courier-users] vpopmail user confused how Courier-IMAP controls qmail
delivery

I usually scroll quickly whenever v* gets mentioned...

> We're just talking about a solution with qmail & SQL where every user (vdoms
> too) is listed in the SQL-table. There are solution for qmail to handle this
> (auth via SQL).
>
> The Benefit is that users who want to connect via imap could be
> authenticated through the auth-SQL too.

The table can be queried natively with authmysql, check out
the MYSQL_SLECT_CLAUSE for use with splitted login/domain.

Roland
Toni Mattila
2002-07-02 14:39:18 UTC
Permalink
Hi,

> >> Use one of the auth-module which comes with courier, or fix
> >> vmailmgr (which also has a couple of other problems reported).
> >
> > Do you know about other problems? What kind of?
>
> Authentification only works a few times and such things...
> Check out the archives and consult the vpopmail-mailinglist.

Please. Don't confuse two diffrent products. Vpopmail by inter7 and
vmailmgr by Bruce Guenter.

In this case we were just wondering why the authentication module doesn't
pass on the full username, or why isn't the full username logged.

Besides thes the logging-thing, I'm 100% satisfied with vmailmgr. It's
less quirky than vpopmail(i'm running also vpopmail in diffrent setup).

With vpopmail 5.2 and authvchkpw I haven't noticed any problems. I'm
running authpam and authvchkpw in chain(no authdaemon involved). There are
about 600+ domains with about 5-30 users in each.

Regars,
Toni Mattila
Leonard Chan
2002-07-02 23:56:02 UTC
Permalink
Alfredo Martino writes:

> Hi to all, I'm a new courier users.
>
> I've installed Courier-imapd in my freebsd server with sendmail.
> I've tried to create new user with adduser but this user can't read email via pop3
> because the maildir is not correct.
>
> I wanto to know if courier work fine with sendmail or if i have to install qmail.

I don't think sendmail can work directly with courier-imapd.

> If courier work fine with sendmail how can I create new user ? I've to create first user
> with adduser and then makemaildir ?

You should create a Maildir in your skeletion directory. Once done, adduser
should be able to copy the structure to your newly created user's home dir.
"man adduser" will give you more info on skeleton dir.

>
> Can you help me with a little example for adding new user in my freebsd and configuring
> it to read email via pop3.

I just disabled sendmail (sendmail_enable="NONE" in /etc/rc.conf) and
install ports/mail/courier

>
>
> Thanks in advace.
> Alfredo Martino
> ***@guest.tlsoft.it
>
>
> ".... perche' TIM e' grande ...."
>
>
>
> -------------------------------------------------------
> This sf.net email is sponsored by:ThinkGeek
> Welcome to geek heaven.
> http://thinkgeek.com/sf
> _______________________________________________
> courier-users mailing list
> courier-***@lists.sourceforge.net
> Unsubscribe: https://lists.sourceforge.net/lists/listinfo/courier-users
Binand Raj S.
2002-07-03 01:55:03 UTC
Permalink
On Wednesday 03 July 2002 07:25 am, Leonard Chan wrote:
> > I wanto to know if courier work fine with sendmail or if i have to
> > install qmail.
>
> I don't think sendmail can work directly with courier-imapd.

If procmail is the local delivery agent for sendmail, then this
/etc/procmailrc delivers mails to ~/Maildir for all users:

***@mail[~] cat /etc/procmailrc
DEFAULT=$HOME/Maildir/

One can fiddle around with procmail scripts for virtual domains also,
I suppose. sendmail needn't be the issue here. The trailing slash is
important, and when setup correctly, procmail will even create ~/Maildir
for you.

Binand
M. Jolic
2002-07-04 11:09:07 UTC
Permalink
Hi Gurus and everybody who is using qmail with VMAILMGR and Courier-IMAP!

After several tests we've found the problem and solved it!

No, it's not Outlook or anything else:

It's the authvmailmgr-module from vmailmgr.

So, if you want to log the user of a virtual domain who is using
courier-imap here's how to handle this:


The authentication-module of vmailmgr sends the "virtual name" of the domain
only! The solution:

Set an ENV with a name where no prog is using this ENV, except the
(modified) imapd from courier-imap:

look at "authvlib.cc" from the source of vmailmgr:

[...]
if(!lookup_baseuser(fulluser, basepw, virtname))
fail_login("Invalid or unknown base user or domain");
presetenv("VUSER=", virtname);
/*
the following line must be added! The ENV-Name can be different, but
notice to change this in the imapd.c too!
*/
presetenv("FULL_USER=", fulluser);

if(!virtname) {
if(virtual_only)
[...]

Notice: We add the ENV "FULL_USER" to the code so that the full
functionality of vmailmgr for qmail will be untouched!

Then make it with: make
and copy the program authvmailmgr to the auth-dir of courier-imap where your
original authvmailmgr resides.

Go to the source of courier-imap and look into the path imap for imapd.c.
Open the file and look for:
[...]
writes("* BYE Courier-IMAP server shutting down\r\n");
writes(tag);
writes(" OK LOGOUT completed\r\n");
writeflush();
emptytrash();
/*
Original: getenv("AUTHENTICATED")
NOW: getenv("FULL_USER")
*/
fprintf(stderr, "INFO: LOGOUT, USER=%s, ip=[%s],"
" headers=%lu, body=%lu\n",
getenv("FULL_USER"), getenv("TCPREMOTEIP"),
header_count, body_count);
exit(0);
}

if (strcmp(curtoken->tokenbuf, "LIST") == 0
[...]

Now we put the "FULL_USER"-ENV instead of "AUTHENTICATED"-ENV, because
according to vmailmgr the variable "fulluser" contains the domain and the
virtual user!

Next step:

STOP imap (& imap-ssl if your using)
make
make install (or copy the imapd to /usr/lib/courier-imap/bin or whereever
your imap-path is)

After you've overwritten the original imapd-prog: start it again and log
into your mailbox via imap and logout.
Look in the maillog (/var/log/maillog) and you should see something like:

INFO: LOGOUT, USER=domain.com-test, ip=[xxx.xxx.xxx.xxx], headers=0, body=0

Now you have the "alternative" Login from the user:

Login: ***@domain.com -> Becomes: domain.com-test

Login: domain.com-test -> Besomes: domain.com-test

I was surprised why NOBODY is interested to get the size of a virtual user
who is getting mails via imap over vmailmgr-auth!!!!

Any ideas?

milli
Sam Varshavchik
2002-08-27 15:38:04 UTC
Permalink
R'twick Niceorgaw writes:

> Hi all,
> I have setup a webadmin password in /etc/courier/webadmin/password file
> using "make install-webadmin-password", but when I try to login to webadmin,
> I'm getting invalid password. I can see my password is correct in the
> password file.

Make sure the ownership/permission of the webadmin wrapper in cgi-bin are
correct.
R'twick Niceorgaw
2002-08-27 15:49:04 UTC
Permalink
I have it root.root mode 555
is it correct or do i need to change it ?

----- Original Message -----
From: "Sam Varshavchik" <***@courier-mta.com>
To: "Courier-Users List" <courier-***@lists.sourceforge.net>
Sent: Tuesday, August 27, 2002 1:37 PM
Subject: [courier-users] Re: webadmin : invalid password


> R'twick Niceorgaw writes:
>
> > Hi all,
> > I have setup a webadmin password in /etc/courier/webadmin/password file
> > using "make install-webadmin-password", but when I try to login to
webadmin,
> > I'm getting invalid password. I can see my password is correct in the
> > password file.
>
> Make sure the ownership/permission of the webadmin wrapper in cgi-bin are
> correct.
>
>
>
>
>
> -------------------------------------------------------
> This sf.net email is sponsored by: OSDN - Tired of that same old
> cell phone? Get a new here for FREE!
> https://www.inphonic.com/r.asp?r=sourceforge1&refcode1=vs3390
> _______________________________________________
> courier-users mailing list
> courier-***@lists.sourceforge.net
> Unsubscribe: https://lists.sourceforge.net/lists/listinfo/courier-users
>
Sam Varshavchik
2002-08-27 17:46:09 UTC
Permalink
Chris MacLeod writes:

> hmm, ok. Not sure how to go about figuring this one out then.
>
> There is no bounce, the application is using me as a relay (and it's
> allowed) and it makes a connection and then sits there. Nothing else in
> the logs, and no bounced message. But if I point it at a sendmail relay
> it works fine, weird.

Check your DNS setup. Check that the application's IP address resolves
backwards and forwards through DNS. Check that there are no broken
firewalls in between that block the ident port.

It's probably a long DNS/ident timeout on connect.
Chris MacLeod
2002-08-27 19:01:05 UTC
Permalink
On Tue, 2002-08-27 at 15:45, Sam Varshavchik wrote:
> Check your DNS setup. Check that the application's IP address resolves
> backwards and forwards through DNS. Check that there are no broken
> firewalls in between that block the ident port.

nope, not DNS forward and reverse is good.
I have ident turned off at the mta so I doubt that's it unless this
cruft app requiress ident.

The box can relay fine off my MTA it's this crufty application which
seems to hold a grudge against courier (since it works with a sendmail
relay) hence why I'm needing to look at the helo and everything to see
if the problem is there.
Sam Varshavchik
2002-08-27 20:43:04 UTC
Permalink
Chris MacLeod writes:

> The box can relay fine off my MTA it's this crufty application which
> seems to hold a grudge against courier (since it works with a sendmail
> relay) hence why I'm needing to look at the helo and everything to see
> if the problem is there.

You'll need to look at the packet level trace anyway.

I recall a similar bug in some old Perl Mail:: module, which believed it
could accomplish its duty simply by flinging SMTP commands one after another
at the server, without bothering to wait for the server to reply. That's not
going to work very well (sendmail might tolerate this abuse, but it violates
the PIPELINING restrictions and won't work with Courier).
Chris MacLeod
2002-08-28 13:37:04 UTC
Permalink
On Tue, 2002-08-27 at 18:42, Sam Varshavchik wrote:
> I recall a similar bug in some old Perl Mail:: module, which believed it
> could accomplish its duty simply by flinging SMTP commands one after another
> at the server, without bothering to wait for the server to reply. That's not
> going to work very well (sendmail might tolerate this abuse, but it violates
> the PIPELINING restrictions and won't work with Courier).

That sounds eerily similar to what I'm seeing. I started doing some
sniffing with tcpdump and noticed that the helo comes in but that's it
nothing after that, whereas off the sendmail relay it all comes in at
once.

C
Chris MacLeod
2002-08-28 13:56:08 UTC
Permalink
On Tue, 2002-08-27 at 18:42, Sam Varshavchik wrote:

> I recall a similar bug in some old Perl Mail:: module, which believed it
> could accomplish its duty simply by flinging SMTP commands one after another
> at the server, without bothering to wait for the server to reply. That's not
> going to work very well (sendmail might tolerate this abuse, but it violates
> the PIPELINING restrictions and won't work with Courier).

Sam - Have you got a link to the PIPELINING restrictions you are
referring to, I'm guessing that it's part of the rfc on smtp.

C
Sam Varshavchik
2002-08-28 14:56:05 UTC
Permalink
Chris MacLeod writes:

> On Tue, 2002-08-27 at 18:42, Sam Varshavchik wrote:
>
>> going to work very well (sendmail might tolerate this abuse, but it violates
>> the PIPELINING restrictions and won't work with Courier).
>
> Sam - Have you got a link to the PIPELINING restrictions you are
> referring to, I'm guessing that it's part of the rfc on smtp.

It's RFC 2197. All it is, is a precise definition when SMTP commands can
be sent without waiting for a reply to the previous SMTP command.

All it says is that the client must wait for a reply to EHLO/HELO, DATA,
NOOP, and QUIT before sending the next command.

If the client spews MAIL FROM/RCPT TO/DATA without waiting for a reply to
EHLO/HELO then it's not going to work with Courier.

Incidentally, a side effect of this is that Courier won't receive spam from
misconfigured HTTP proxies. There's a contemporary fad where spamware scans
for misconfigured HTTP caches which will proxy an HTTP CONNECT for anyone,
connect to the cache, send CONNECT mailserver.example.com:25, then sends an
SMTP command block formatted as HTTP request headers, followed by a blank
line, and the spam payload. The stupid cache then attempts to connect to
the destination server and throw the whole chunk at it. Stupid sendmail
relays will casually process the input stream a line at a time, and
obediently deliver the spam payload. Courier will flush the input after a
HELO and not even see the rest of the crap.
Benjamin Schleinzer
2002-08-28 04:04:04 UTC
Permalink
Did you configure courier with or without IPv6 support. On some maschines
IPv& support is broken and can result in the strangest errors. Try to
recompile courier without IPv6 support and try again it might solve the
problem.


> nope, not DNS forward and reverse is good.
> I have ident turned off at the mta so I doubt that's it unless this
> cruft app requiress ident.
>
> The box can relay fine off my MTA it's this crufty application which
> seems to hold a grudge against courier (since it works with a sendmail
> relay) hence why I'm needing to look at the helo and everything to see
> if the problem is there.
Sam Varshavchik
2002-09-02 09:20:08 UTC
Permalink
Peter Rose writes:

> Sep 1 19:35:44 server authdaemond.pgsql: authdaemon:
> modules="authcustom authcram authuserdb authpgsql authpam", daemons=5
> Sep 1 19:36:59 server authdaemond.pgsql: authdaemon:
> modules="authcustom authcram authuserdb authpgsql authpam", daemons=5
> Sep 1 19:36:59 server modprobe: can't locate module net-pf-10
>
> Is this likely to be because of the missing net-pf-10 module or does it
> sound like a PAM thing? I recall I had a similar problem tweaking PAM

You have the postgres module enabled, but you haven't configured it.
Disable the postgres module.
Sam Varshavchik
2002-09-04 13:19:02 UTC
Permalink
Dirk Broodryk writes:

> Hi..
>
> Thanks for the response..
>
> basicly I want to get rid of the
> dirkb%***@ferreiragroup.co.za
> the % bit is wrong. Courier adds that bit by it's own..

No it doesn't.
Sam Varshavchik
2002-09-06 11:04:05 UTC
Permalink
Alexei Batyr' writes:

> Sam,
>
> I remember that some time ago there was two types of "bad" exit codes in
> .courier file - some of them caused sending DSNs and some others not. Now

Not exactly. Some of them caused an immediate rejection, others resulted
in the mail being requeued for a later re-delivery attempt.

> any of the listed in dot-corier manpage codes (64, 65, 67, 68, 69, 70, 76,
> 77, 78, or 112) generate DSN. What's changed?

Nothing. These codes are listed as ones that generate immediate bounces.

> Is there now a possibility to
> set up .courier (or .mailfilter) exit code so that courier reply only with
> SMTP error without DSN? E.g. "catchall" .courier-default file just reply
> with "550 User unknown" when receiving mail for non-existent local users?

.courier is read after the mail has already been accepted, so SMTP is no
longer a factor, by this time.

Courier rejects mail to nonexistent recipients by default.
Sam Varshavchik
2002-09-06 12:54:02 UTC
Permalink
Alexei Batyr' writes:

> On Friday, September 06, 2002 3:02 PM, Roland Schneider wrote:
>
>> messages processed by .courier-dotfiles (or .mailfilter)
>> have already been received and accepted by the server.
>
> Wasn't older versions of courier capable to reply with defined "550 ..."
> after receiving message (completion of DATA command)?

Right, and it still does. But .courier files have nothing to do with it.
Sam Varshavchik
2002-09-09 19:14:04 UTC
Permalink
Sam Carleton writes:

> On Mon, Sep 09, 2002 at 01:12:26PM -0700, Gordon Messmer wrote:
>> On Mon, 2002-09-09 at 12:28, Sam Carleton wrote:
>> >
>> > Ok, thank you for pointing out there was a difference. I got it to
>> > work! It is compiled and installed! Now for the big Q, why didn't
>> > my approach work?
>>
>> I don't know that courier's configure scripts specifically search the
>> $prefix for libraries. --prefix and --libdir are not equivalent to
>> setting CPPFLAGS and LDFLAGS. The former adjust were courier will
>> install to. The latter adjust where the compiler and linker look for
>> libraries already installed.
>
> So it was not that I went wrong, it is that the configuration does not
> use the -libdir and --includedir. How does someone (me) go about
> changing this little bug so that others don't beat there head against
> the wall assuming that --libdir and --includedir are utilized?

There's nothing to fix. --libdir and --includedir specify something that's
completely different.
Sam Varshavchik
2002-09-09 20:30:03 UTC
Permalink
Donald Nash writes:

> if (strcmp(p->as, "\\RECENT") == 0 &&
> current_mailbox_info.msgs[msgnum].recentflag)
> p->value=1;
> if (f.seen && flags->seen)
> p->value=1;
> if (f.answered && flags->answered)
> p->value=1;
> if (f.deleted && flags->deleted)
> p->value=1;
> if (f.flagged && flags->flagged)
> p->value=1;
> }
> break;
>
> I notice that there is no testing for the \\DRAFT flag. If I'm reading the
> code right, this would indicate that it isn't possible for Courier to
> search for draft messages. Was this intentional or just an oversight?

It's an oversight. The \Draft flag support was added some time ago, and
this reference slipped through the cracks...
Sam Varshavchik
2002-09-11 16:51:05 UTC
Permalink
Theodore Knab writes:

> Thanks for your speedy response.
>
> I have previous raised the maximum connections per IP address to 10.
>
> Where would I find the maximum accepted number of connections ?
>
> ***@imap:~$ cat /etc/courier/imapd | grep -i 'max\|limit'
> ##NAME: MAXDAEMONS:0
> # Maximum number of IMAP servers started
> #MAXDAEMONS=1000
> MAXDAEMONS=10000

That's MAXDAEMONS. 10K should be adequate.
Sam Varshavchik
2002-09-11 18:02:01 UTC
Permalink
Tim Lynch writes:

> yes, i too am looking to seperate esmtp logs from imapd logs.
>
> there doesn't appear to be any way to specify facility with courierlogger,
> according to the man page. haven't dug into the source, although i suspect
> it is gnu logger.c.

Typical syslog facilities are rather crude.

There are various syslog replacements floating somewhere out there which
can use regexps to spread log messages from different programs into
different files.
Tim Lynch
2002-09-11 20:36:04 UTC
Permalink
thanks, i am aware of syslog, it's various flavors, the various
replacements, and etc.

are there any suggestions besides:
modify courierlogger to make facility configurable
use something other than stock syslog

notice this courier inconsistency:
in esmtpd the stderrlogger can be changed, like so:
TCPDOPTS="-stderrlogger=/usr/lib/courier/sbin/courierlogger -nodnslookup -no
identlookup"

in imapd the above creates redundant -stderrlogger args in couriertcpd.

----- Original Message -----
From: "Sam Varshavchik" <***@courier-mta.com>
To: <courier-***@lists.sourceforge.net>
Sent: Wednesday, September 11, 2002 1:00 PM
Subject: [courier-users] Re: Courier pop log and imap log


: Tim Lynch writes:
:
: > yes, i too am looking to seperate esmtp logs from imapd logs.
: >
: > there doesn't appear to be any way to specify facility with
courierlogger,
: > according to the man page. haven't dug into the source, although i
suspect
: > it is gnu logger.c.
:
: Typical syslog facilities are rather crude.
:
: There are various syslog replacements floating somewhere out there which
: can use regexps to spread log messages from different programs into
: different files.
:
:
:
: -------------------------------------------------------
: In remembrance
: www.osdn.com/911/
: _______________________________________________
: courier-users mailing list
: courier-***@lists.sourceforge.net
: Unsubscribe: https://lists.sourceforge.net/lists/listinfo/courier-users
:
Sam Varshavchik
2002-09-11 19:22:03 UTC
Permalink
Edwin Culp writes:

> Quoting Marc Lindahl <***@bowery.com>:
>
> | In the webadmin app - there's a section for 'free mail' servers (under BOFH
> | mail filters, or Incoming ESMTP, I don't remember which off hand) - put
> | yahoo.com in there and it'll only accept mail from yahoo.com's authentic
> | mail relays. Amazing how much spam it stops.
>
> Marc,
>
> Thanks. I can certainly understand how it could reduce spam. I see
> a lot with freemail addresses so I filed this away for later application.
> Being an old guy who learned with ed and then graduated to vi, I had
> never used the web interface to courier configuration so I kept this
> email until I had a chance to set it up. I did and added yahoo and
> hotmail, then I went looking for where it was saved. I found it in
> courier/webadmin/added/bofh. It includes what I have in my
> courier/bofh file with an additional freemail line. AFAIK, there
> should only be one bofh file so my question is
>
> "Which bofh will take precedence or are they both checked with each new mail?"

You forgot to select "install new configuration".
Edwin Culp
2002-09-11 20:49:05 UTC
Permalink
Quoting Sam Varshavchik <***@courier-mta.com>:

| >
| > "Which bofh will take precedence or are they both checked with each new
| mail?"
|
| You forgot to select "install new configuration".

Thanks, Sam.

Duh! It did exactly what I had hoped. I'll quietly go back to vi now. This web
stuff is too complicated for me;-).

Thanks,

ed
Sam Varshavchik
2002-09-14 22:06:19 UTC
Permalink
Anand Buddhdev writes:

> I think that's wrong behaviour. If courier's esmtpd is bound only to
> 10.0.0.2, it should not make assumptions about what is running on
> 10.0.0.1. Why shouldn't I be allowed to run 2 different SMTP daemons on

The esmtp client doesn't really know anything about what the esmtp server
does.

> different IP addresses on the same box, or even on the same IP but
> different ports, and not pass email between them?

Good point, this needs to be investigated.
Sam Varshavchik
2002-09-15 12:49:03 UTC
Permalink
Patrick O'Reilly writes:

> Sam,
>
> thanks for your response.
>
> The sending MTA is none other than sendmail, and I have had the same
> reaction from two separate sendmail servers. Sendmail seems to understand
> that "550 Invalid User" is fatal, but my "550 other ad-hoc text" gets put
> into the mailq on the sendmail server. Does sendmail inspect the entire
> string?

I don't know.
Sam Varshavchik
2002-09-15 12:51:02 UTC
Permalink
Peter C. Norton writes:

> I don't see where setting any info for the loginid is possible. It
> seems like I do the following:
>
> SELECT PGSQL_UID_FIELD, PGSQL_GID_FIELD, ... WHERE lower(id)='loginid'
>
> by setting
>
> PGSQL_LOGIN_FIELD id
>
> But I can't see how, besides altering the source I can set up my
> query to do a lower() on the loginid string.

PGSQL_LOGIN_FIELD lower(id)
Sam Varshavchik
2002-09-16 19:41:04 UTC
Permalink
Edwin Culp writes:

> db3-3.3.11,1
> autoconf213-2.13.000227_2
> automake14-1.4.5_1
> gmake-3.79.1_3
> libtool-1.3.4_4
> m4-1.4_1
> sysconftool-0.13
> perl-5.6.1_8
> openldap-2.0.25_1
> p5-Net-CIDR-0.04
> courier-ldap-0.39.3
> I don't think you can get much more basic than that.

Most of the above are build dependencies only. autoconf, automake, gmake,
libtool, m4, and sysconftool are not required at runtime, only at buildtime.
Sam Varshavchik
2002-09-19 19:44:02 UTC
Permalink
wolfgang writes:

> Given the following header (longer than 69 chars):
>
> From: "1234567890123456789012345678901234567890"
> <***@123domain.tld>
>
> Courier will change it to
>
> From:
> "1234567890123456789012345678901234567890" <***@123domain.tld>
>
>
> This will only happen if the receipient are not local, therfore the mail
> is handled by esmtpd only. If I Send the same message to a local
> receipient the header will remain unchanged.
>
> 1. Is this a known behavior ?
> a) BUG
> b) feature

It's not a bug. Both are perfectly valid, and completely equivalent,
headers.
wolfgang
2002-09-20 16:21:06 UTC
Permalink
Sam Varshavchik writes:

> It's not a bug. Both are perfectly valid, and completely equivalent,
> headers.
>
Nice to read this!
Just for interrest, could you explaine why courier rewrites the header ?

The reason I ask this are that one Provider will not accept such a
header an don't pass on the mail to the recipient.

Could you please point me to the relevant RFC to show him that's an
valid header.

Thanks.
Wolfgang Käß
Patrick O'Reilly
2002-09-20 20:30:02 UTC
Permalink
>> It's not a bug. Both are perfectly valid, and completely equivalent,
>> headers.
>>
>Nice to read this!
>Just for interrest, could you explaine why courier rewrites the header ?
>
>The reason I ask this are that one Provider will not accept such a
>header an don't pass on the mail to the recipient.
>
>Could you please point me to the relevant RFC to show him that's an
>valid header.
>

http://docs.perimeter.co.za/RFC/rfc822.txt (My Copy - you may validate it
anywhere you choose :)

Look for section "B.1. SYNTAX" . It explains that field body may consist
of any text separated by white space and CRLF in any arbitrary combination
(well, almost any).

---
Regards,
Patrick O'Reilly.
___ _ __
/ _ )__ __ (_)_ __ ___ _/ /____ __
/ __/ -_) _) / ~ ) -_), ,-/ -_) _)
/_/ \__/_//_/_/~/_/\__/ \__/\__/_/
http://www.perimeter.co.za
wolfgang
2002-09-21 13:03:01 UTC
Permalink
Sam Varshavchik writes:
>
> >> It's not a bug. Both are perfectly valid, and completely equivalent,
> >> headers.

Patrick O'Reilly wrote:
>
> Look for section "B.1. SYNTAX" . It explains that field body may consist
> of any text separated by white space and CRLF in any arbitrary combination
> (well, almost any).
>
Thanks for pointing me to rfc822.
But you are missleaded, it's not the question what a field-body may
consits of.

The question are where (pos) can a header be 'folded' ?

rfc822 says:
the general rule is that wherever there
may be linear-white-space (NOT simply LWSP-chars), a CRLF
immediately followed by AT LEAST one LWSP-char may instead be
inserted.

The first 'linear-white-space' are between the colon ':' and the
field-body, so Courier will be ok.

Wolfgang Käß
Sam Varshavchik
2002-09-21 20:50:03 UTC
Permalink
wolfgang writes:

> Sam Varshavchik writes:
>
>> It's not a bug. Both are perfectly valid, and completely equivalent,
>> headers.
> >
> Nice to read this!
> Just for interrest, could you explaine why courier rewrites the header ?

Mail headers on locally-sent mail are generally rewritten for the purpose of
adding any default domains. As part of the process, excessively long header
lines are wrapped.

> The reason I ask this are that one Provider will not accept such a
> header an don't pass on the mail to the recipient.
>
> Could you please point me to the relevant RFC to show him that's an
> valid header.

RFC 2822 states that header lines may be wrapped wherever whitespace is
present.

--
Sam
Sam Varshavchik
2002-09-24 00:59:02 UTC
Permalink
Huaikun Lin writes:

> Hi Sam
>
> At 07:50 PM 9/22/2002 -0400, Sam Varshavchik wrote:
>>Huaikun Lin writes:
>>
>>>Hi
>>>I am doing transfer mailboxes from "mailbox" format to "Maildir" format,
>>>some POP/IMAP customers
>>>email client set up "Leave messages on server". Some mailboxess are as
>>>big as 80MB.
>>>When I transfer them to "Maildir" format,how can I mark the "read"
>>>messages as "read"?
>>
>>Append ":2,S" to the message's filename.
>
> I know that.
>
>
>>>If I don't do that, the customers have to download the whole messages
>>>(50~80MB) again,this will
>>>cause big trouble.
>>>Anyone has solution?
>>
>>Appending :2,S may not be the complete solution. Message UIDs will also
>>change, and some clients may use that as an excuse to re-download the mail
>>anyway. You'll definitely need to test your client software to verify.
> I tested that didn't work.
>
> Email clients will re-download everything.
>
> I tested Eudora,Outlook Express and Microsoft Outlook. If I set up leave
> message on server using Eudora and
> read all messages, when I used Outlook Express or Microsoft Outlook,they
> still re-download everything treat all
> messages un-read.

You will then need to obtain the folder UID and message UIDs of your
existing mail, then arrange to set the same for the mail in the maildirs.

This is a small project of moderate complexity. It's not complicated, but
will require non-trivial amounts of scripting and programming.
Huaikun Lin
2002-09-24 18:27:04 UTC
Permalink
Hi Sam

At 10:58 PM 9/23/2002 -0400, Sam Varshavchik wrote:
>Huaikun Lin writes:
>
>>Hi Sam
>>At 07:50 PM 9/22/2002 -0400, Sam Varshavchik wrote:
>>>Huaikun Lin writes:
>>>
>>>>Hi
>>>>I am doing transfer mailboxes from "mailbox" format to "Maildir"
>>>>format, some POP/IMAP customers
>>>>email client set up "Leave messages on server". Some mailboxess are as
>>>>big as 80MB.
>>>>When I transfer them to "Maildir" format,how can I mark the "read"
>>>>messages as "read"?
>>>
>>>Append ":2,S" to the message's filename.
>>I know that.
>>
>>>>If I don't do that, the customers have to download the whole messages
>>>>(50~80MB) again,this will
>>>>cause big trouble.
>>>>Anyone has solution?
>>>
>>>Appending :2,S may not be the complete solution. Message UIDs will also
>>>change, and some clients may use that as an excuse to re-download the
>>>mail anyway. You'll definitely need to test your client software to verify.
>> I tested that didn't work.
>>Email clients will re-download everything.
>>I tested Eudora,Outlook Express and Microsoft Outlook. If I set up leave
>>message on server using Eudora and
>>read all messages, when I used Outlook Express or Microsoft Outlook,they
>>still re-download everything treat all
>>messages un-read.
>
>You will then need to obtain the folder UID and message UIDs of your
>existing mail, then arrange to set the same for the mail in the maildirs.

I can see the message IDs for the mail in the existing mailbox.

But I don't know what is the folder UID? How to get them?

Lin
Sam Varshavchik
2002-09-24 21:04:02 UTC
Permalink
Huaikun Lin writes:

>>
>>You will then need to obtain the folder UID and message UIDs of your
>>existing mail, then arrange to set the same for the mail in the maildirs.
>
> I can see the message IDs for the mail in the existing mailbox.
>
> But I don't know what is the folder UID? How to get them?

I think the UW-IMAP server uses ctime in the inode as the folder UID.
Sam Varshavchik
2002-09-26 19:59:03 UTC
Permalink
Patrick O'Reilly writes:

> That sounds good - I want to do so, but I can't find out "how-to" mark
> recipients as delivered...
>
> I've looked in 'courier' and 'courierfilter', and reformail and
> reformime, and mailq and cancelmsg and a whole bunch of other man
> pages... :(
>
> A little clue please?

http://www.courier-mta.org/queue.html

"Control File Format" section. I suggest examining a couple of records in
a live queue, until you're familiar with it.
Alessandro Vesely
2002-09-27 07:06:02 UTC
Permalink
Sam Varshavchik writes:
> Patrick O'Reilly writes:
>> ..."how-to" mark recipients as delivered...
>
> http://www.courier-mta.org/queue.html
>
> "Control File Format" section. I suggest examining a couple of
> records in a live queue, until you're familiar with it.

That's great, but it in turn raises a couple of questions.

1) What was that discussion about rewriting messages?
One cannot rewrite the message but can rewrite the control file?
Then, it would seem advisable that (in the minority of cases)
when the filter should alter the message, it just marks any
recipients as delivered and then resends the altered message.
(Preferably from a different process.)

2) The control file looks quite intricate for an occasional filter
writer. Is it planned to have some API, such as get_sender, get_next
recipient and the like? Could that be the goal of some sub-project?

In case it can be done, I insist that we should devise some
(standard?) mechanism that will allow local filters to use the
funcionality of a global filter. In that way, a postmaster can
factor functions often used in local filters and implement them
in C - more efficient, more reliable, no metacharacter headaches.
Local filters use those function just as any other mailfilter
function they call.

Cool?
Ale
Sam Varshavchik
2002-09-27 23:11:02 UTC
Permalink
Andrew writes:

> I've never built the full courier distribution, only courier IMAP; and
> I've always installed from the tarball, never built an RPM. I found it
> in the authlib directory. It doesn't get installed by default but stays
> in the source tree. Perhaps it doesn't make it into the RPMs.

authtest is not required in a properly configured system. Hence it's not
included in the binary RPMs.
Sam Varshavchik
2002-09-27 23:29:03 UTC
Permalink
Enrique Vadillo writes:
> Now, since the previous config did not eliminate messages 7 days old
> (i had messages 2 months old there!) i have changed my etc/imapd line to:
>
> IMAP_EMPTYTRASH=BASURERO:7,ENVIADOS:7
>
> do you think this will make it work? some users that wrote back to me said
> that you always respond simply saying 'prove it' and that they simply
> believe that the automated 'EMPTYTRASH' thingy is not working..

You still haven't shown anything that actually shows that it's not working.
Sam Varshavchik
2002-09-29 21:17:03 UTC
Permalink
Barry Hensley writes:

> fast that I almost couldn't read it but, it said something about webmail
> users not being able to change their passwords. Have you heard of
> anything like this?

Ignore it. It's complaining that the expect package is not installed, which
is not used here.
Barry Hensley
2002-09-29 21:47:03 UTC
Permalink
Yes, it did say something about the expect package.

Thanks for the feedback. Now, I've just got to get it up and running.

Barry

----- Original Message -----
From: "Sam Varshavchik" <***@courier-mta.com>
To: <courier-***@lists.sourceforge.net>
Sent: Sunday, September 29, 2002 6:16 PM
Subject: [courier-users] Re: --with-locking-method error


> Barry Hensley writes:
>
> > fast that I almost couldn't read it but, it said something about webmail
> > users not being able to change their passwords. Have you heard of
> > anything like this?
>
> Ignore it. It's complaining that the expect package is not installed,
which
> is not used here.
>
>
>
> -------------------------------------------------------
> This sf.net email is sponsored by:ThinkGeek
> Welcome to geek heaven.
> http://thinkgeek.com/sf
> _______________________________________________
> courier-users mailing list
> courier-***@lists.sourceforge.net
> Unsubscribe: https://lists.sourceforge.net/lists/listinfo/courier-users
>
Sam Varshavchik
2002-10-01 21:00:42 UTC
Permalink
nadim writes:

> Well things never get better it seems. In my previous mail I wrote that
>
> ls /usr/lib/courier-imap/libexec/authlib/ :
> authdaemon authdaemond authdaemond.ldap authdaemond.plain
>
> the authshadow or authpam where not there, why I don't know.

Try reading INSTALL. It explains everything.
Sam Varshavchik
2002-10-01 21:04:59 UTC
Permalink
nadim writes:

> On Wednesday 02 October 2002 00:03, Sam Varshavchik wrote:
>> Steve Shockley writes:
>> > Recently, OpenBSD has changed Apache to run chroot /var/www. Before I
>> > spend hours tracing it out, what's the feasibility of getting either
>> > webmail or webadmin to run in a chroot?
>>
>> 0% possibility of success. How exactly are you plainning to have the
>> chrooted webmail binary read maildirs outside of the chroot jail?
>
> Wouldn't it be possible to have a home directory in the new root which
> would be a like to the real /home?

So, if:

1) You move all the necessary dynamic libraries into the chroot jail

2) You move all the mail accounts into chroot jail

3) Since sending mail involves running a small shell script stub, you'll
need to move at least /bin/sh into the chroot jail too

and

4) What you end up in your chroot jail isn't much different than what
outside of the chroot jail looks like,

then

what exactly was the point to this chroot jail in the first place?
Sam Varshavchik
2002-10-03 00:04:25 UTC
Permalink
Zenon Panoussis writes:

> echo "***@localhost" >/home/foo/.courier-default
> echo "***@localhost" >/home/otheraccount/.courier-default

This happens to be exactly what webadmin does, when told to do something of
this sort.

> It's just the first one I figured which served my purposes.

It would've been much easier just to use webadmin to set this stuff up.
Zenon Panoussis
2002-10-03 00:38:05 UTC
Permalink
Sam Varshavchik wrote:
>

> This happens to be exactly what webadmin does, when told to do something
> of this sort.

>> It's just the first one I figured which served my purposes.

> It would've been much easier just to use webadmin to set this stuff up.

I was tempted severly, but I like to know the systems I run
and I would still not be able to understand aliases if webadmin
had made them for me. I think it's an excellent tool, but only
for long-term convenience; not for first-time education. The
latter requires rubbing ;)

Z
Alexei Batyr'
2002-10-03 10:12:01 UTC
Permalink
On Thursday, October 03, 2002 6:02 AM, Sam Varshavchik wrote:

> Zenon Panoussis writes:
>
>> echo "***@localhost" >/home/foo/.courier-default
>> echo "***@localhost" >/home/otheraccount/.courier-default
>
> This happens to be exactly what webadmin does, when told to do
> something of this sort.
>
>> It's just the first one I figured which served my purposes.
>
> It would've been much easier just to use webadmin to set this stuff up.

Wouldn't it be even easier if courier just doesn't strip local domain parts
from _aliases_ but still does it for _real_ local recipients? What could be
wrong in such behaviour?

--
Alexei.
Bill Michell
2002-10-03 10:50:03 UTC
Permalink
Alexei Batyr' writes:


> Wouldn't it be even easier if courier just doesn't strip local domain parts
> from _aliases_ but still does it for _real_ local recipients? What could be
> wrong in such behaviour?
>
It would confuse me.

Currently, for domains listed in locals, you *never* append the domain, for
all other domains, you *always* append it. Simple.

--
Bill Michell
***@mics.org.uk (home)
Sam Varshavchik
2002-10-03 09:48:06 UTC
Permalink
Zenon Panoussis writes:

>
> Sam Varshavchik wrote:
>>
>
>> Lookout may not have the brain cells to know how to do LOGIN or
>> CRAM-MD5. Try PLAIN.
>
> Talking about which, is there any difference between
> "LOGIN CRAM-MD5 PLAIN" and "PLAIN CRAM-MD5 LOGIN"?
> Are the methods tried in the order they appear on in
> a pre-programmed way?

It's the client's call.
Alexei Batyr'
2002-10-03 14:12:05 UTC
Permalink
On Thursday, October 03, 2002 7:10 PM, Bill Michell wrote:

> Alexei Batyr' writes:
>
>> On Thursday, October 03, 2002 4:48 PM, Bill Michell wrote:
>>
>>> Currently, for domains listed in locals, you *never* append the
>>> domain, for all other domains, you *always* append it. Simple.
>>
>> Simple idea: strip local domain part _after_ looking up alias
>> database instead of doing it _before_.
> But why?

Let's look at practical example: in my installation there is approx. 10
local domains for different publications of our publishing house. All users
are real mail server users. Some of them work for several magazines, so they
want messages addressed to, e.g., ***@pcmag.ru, ***@pcweek.ru and ***@crn.ru to
go to the same mailbox. That's why I'm using "unidomain" scheme, where all
hosted domains are local and everybody can use any domain part of his
address.

However in some cases I need to make difference between domains, e.g. alias
"news" should exist for all my domains but messages to ***@pcmag.ru and
***@crn.ru should be redirected to the different real mailboxes. Now
instead of simply using alias database entries

***@pcmag.ru: user1
***@crn.ru: user2
...

I need to use following sysconfdir/aliasdir/courier-news file:

|case "${HOST}" in \
"pcmag.ru") sendmail -t -f "$SENDER" user1 ;; \
"crn.ru") sendmail -t -f "$SENDER" user2 ;; \
...
*) sendmail -t -f "$SENDER" userX ;; \
esac

Not very simple and straightforward, isn't it?

--
Alexei.
Sam Varshavchik
2002-10-03 19:42:07 UTC
Permalink
Alexei Batyr' writes:


> Let's look at practical example: in my installation there is approx. 10
> local domains for different publications of our publishing house. All users
> are real mail server users. Some of them work for several magazines, so they
> want messages addressed to, e.g., ***@pcmag.ru, ***@pcweek.ru and ***@crn.ru to
> go to the same mailbox. That's why I'm using "unidomain" scheme, where all
> hosted domains are local and everybody can use any domain part of his
> address.

And that's the broken part. You _should_ be using hosteddomains here, and
use aliases for the shared accounts.

> |case "${HOST}" in \
> "pcmag.ru") sendmail -t -f "$SENDER" user1 ;; \
> "crn.ru") sendmail -t -f "$SENDER" user2 ;; \
> ...
> *) sendmail -t -f "$SENDER" userX ;; \
> esac
>
> Not very simple and straightforward, isn't it?

Of course not. Aliases -- when properly implemented -- are much simpler.
Alexei Batyr'
2002-10-04 06:42:02 UTC
Permalink
On Friday, October 04, 2002 1:41 AM, Sam Varshavchik wrote:

>> using "unidomain" scheme, where all hosted domains are local and
>> everybody can use any domain part of his address.
>
> And that's the broken part. You _should_ be using hosteddomains
> here, and use aliases for the shared accounts.

Probably you're right - it'd be more correct implementation, but... I'm lazy
and busy, so I prefer simply tell people that they can use any domain and
forget about this issue instead of having to ask every new (and often old,
about 300 ones) emloyee which aliases (s)he want.

So question remains: What could be wrong if in such configuration courier
strip local domain part _after_ looking up alias database instead of doing
it _before_?

--
Alexei.
Sam Varshavchik
2002-10-03 19:49:02 UTC
Permalink
cc writes:

> 250-oval Ok.
> 250-XVERP=Courier
> 250-XEXDATA
> 250-XSECURITY=NONE,STARTTLS
> 250-PIPELINING
> 250-8BITMIME
> 250-SIZE
> 250 DSN
>
> So I guess the answer is no, it doesn't support AUTH. What gives? When I
> originally built Courier I used: ./configure --prefix=/opt/courier
> --with-authmysql

Probably because you did not enable ESMTP authentication in the
configuration file.
cc
2002-10-03 20:22:04 UTC
Permalink
Sam Varshavchik writes:

>>
>> So I guess the answer is no, it doesn't support AUTH. What gives? When
>> I originally built Courier I used: ./configure --prefix=/opt/courier
>> --with-authmysql
>
> Probably because you did not enable ESMTP authentication in the
> configuration file.

Well I've just (re)read the INSTALL instructions and I don't see any mention
of enabling or disabling it.

Bizzarely, it has started working. I used the provided script to create a
trial certificate for doing SSL for IMAP, SMTP & POP3 and all of a sudden
ESMTP listed AUTH as a command. It wasn't just because I restarted esmptd
either, as I'd done that many many times as I tried different things. So
right now, it works, I just wish I knew why so that when I install it on a
production server I can be sure I can make it work again.
Bill Schindler
2002-10-03 20:36:04 UTC
Permalink
On Thursday, October 3, 2002, at 03:21 PM, cc wrote:

> Sam Varshavchik writes:
>> Probably because you did not enable ESMTP authentication in the
>> configuration file.
>
> Well I've just (re)read the INSTALL instructions and I don't see any
> mention of enabling or disabling it.
cc
2002-10-04 20:48:03 UTC
Permalink
Bill Schindler writes:

>
Jonas Printzén (Hemma)
2002-10-04 06:38:04 UTC
Permalink
Hello all!

SOLUTION!

I found that if I build the courier RPM's from source I get a special
RPM with all the smtp-auth stuff! Without it all the config files can be
just fine but you get no auth on esmtp!!!! Also you get no errors or such
when starting esmtp nor anything in the logs stating the auth can't run.

1) So! Install the courier-smtpauth package
2) Enable the authentication in the esmtp config file.
AUTHMODULES="authdaemon"
ESMTPAUTH="LOGIN CRAM-MD5" (works with KMail and OutLockExpress)

3) Restart the esmtp service.
4) Verify with running telnet <smtpserverip> <smtpport> and send the EHLO <myhostname>
If the reply containg an 250-AUTH .... line, auth is OK!

5) Go ahead and send mail! :o)

TIP!

If you whant to protect your server from auth eavesdropping don't allow auth-smtp on
port 25. Let this be for local delivery only. Setup the esmpt-ssl to force SSL/TLS and
allow auth. This way you have better protection on your authentication process.
I always try to avoid any auth on non-encrypted connection...

/Jonas
Bill Michell
2002-10-04 14:39:04 UTC
Permalink
> -----Original Message-----
> From: courier-users-***@lists.sourceforge.net
> [mailto:courier-users-***@lists.sourceforge.net]On Behalf Of Alexei
> Batyr'
> Sent: 04 October 2002 09:41
> To: courier-***@lists.sourceforge.net
> Subject: [courier-users] Re: Aliases redirect to local account
>
>
> On Friday, October 04, 2002 1:41 AM, Sam Varshavchik wrote:
>
> >> using "unidomain" scheme, where all hosted domains are local and
> >> everybody can use any domain part of his address.
> >
> > And that's the broken part. You _should_ be using hosteddomains
> > here, and use aliases for the shared accounts.
>
> Probably you're right - it'd be more correct implementation,
> but... I'm lazy
> and busy, so I prefer simply tell people that they can use
> any domain and
> forget about this issue instead of having to ask every new
> (and often old,
> about 300 ones) emloyee which aliases (s)he want.
>
> So question remains: What could be wrong if in such
> configuration courier
> strip local domain part _after_ looking up alias database
> instead of doing
> it _before_?
>
And I say again - it would xonfuse me. locals *never* have the domain
appended, hosteddomains *always* do. Simple.

--
Bill Michell
***@mics.org.uk
Alexei Batyr'
2002-10-05 09:57:02 UTC
Permalink
On Friday, October 04, 2002 8:38 PM, Bill Michell wrote:

>> So question remains: What could be wrong if in such configuration courier
>> strip local domain part _after_ looking up alias database instead of
doing it _before_?
>>
> And I say again - it would xonfuse me. locals *never* have the domain
> appended, hosteddomains *always* do. Simple.

Any confusing or breaking existing configuration would be avoided if
courieresmtpd looks up aliases for locals twice: Firstly with full recipient
address, as for hosteddomains (if found, follow alias instructions and exit)
and secondly (if not found) with stripped local domain part. I'd try to make
patch but not sure I could find appropriate piece of code.

--
Alexei.
Sam Varshavchik
2002-10-21 19:31:04 UTC
Permalink
Sanjay X. Patel writes:

> Can someone pleas tell me if there are any special file permission for
> ant of the files in the etc folder?

All permissions will be set correctly by make install and make
install-configure.
Sam Varshavchik
2002-10-25 10:12:05 UTC
Permalink
Brian Candler writes:

> bindir=${exec_prefix}/bin
>
> This is fine except it breaks POP3 STLS / IMAP STARTTLS, because pop3d.rc /
> imapd.rc clears the environment before starting the daemon, which in turn
> looks for the COURIERTLS environment variable.
>
> I only use the imap/sqwebmail parts of Courier, so I can't speak for
> esmtpd-ssl.
>
> Unless couriertls is considered a more "experimental" part of Courier than
> the rest of it, I am not sure why the path to it is put a configuration
> variable in the first place.
>
> Why not just put
>
> COURIERTLS=@bindir@/couriertls

Oh, I see...

You can't use @bindir@ directly due to the simple fact that it's default
value is, literally, ${exec_prefix}/bin

You literally need to initialize these things in the rc script by hand...
Brian Candler
2002-10-25 14:21:03 UTC
Permalink
On Fri, Oct 25, 2002 at 08:11:45AM -0400, Sam Varshavchik wrote:
> > Unless couriertls is considered a more "experimental" part of Courier than
> > the rest of it, I am not sure why the path to it is put a configuration
> > variable in the first place.
> >
> > Why not just put
> >
> > COURIERTLS=@bindir@/couriertls
>
> Oh, I see...
>
> You can't use @bindir@ directly due to the simple fact that it's default
> value is, literally, ${exec_prefix}/bin

I see. But it does work for @bindir@/pop3d for example:

@SETENV@ - @SHELL@ -c " set -a ; . @sysconfdir@/pop3d ; \
. @sysconfdir@/pop3d-ssl ; \
TLS_PROTOCOL=$TLS_STARTTLS_PROTOCOL ; \
export TLS_PROTOCOL ;
@libexecdir@/couriertcpd -address=$ADDRESS \
-stderrlogger=@libexecdir@/courierlogger \
-stderrloggername=pop3d \
-maxprocs=$MAXDAEMONS -maxperip=$MAXPERIP \
-pid=$PIDFILE $TCPDOPTS \
$PORT @sbindir@/pop3login $LIBAUTHMODULES \
here >>>>>> @bindir@/pop3d Maildir"

which becomes ${exec_prefix}/bin/pop3d, but exec_prefix is expanded in the
quoted string before the shell is called.

My suggestion before was wrong - setting COURIERTLS at the top of the .rc
script won't work, because it still isn't inherited by the cleaned-up
environment which @SETENV@ produces.

But ISTM that the knowledge of where couriertls lives belongs in the .rc
script (just as with the other components - couriertcpd, pop3login, pop3d
etc) rather than in a config file.

As long as the definition comes _within_ the quoted shell command line I
think it should work: i.e.

[pop3d.rc]

@SETENV@ - @SHELL@ -c " set -a ; . @sysconfdir@/pop3d ; \
. @sysconfdir@/pop3d-ssl ; \
TLS_PROTOCOL=$TLS_STARTTLS_PROTOCOL ; \
export TLS_PROTOCOL ;
>>> COURIERTLS=@bindir@/couriertls ; export COURIERTLS ; \

In the case of pop3d-ssl.rc and imapd-ssl.rc, I think you should just be
able to replace $COURIERTLS with @bindir@/couriertls

Cheers,

Brian.
Sam Varshavchik
2002-10-25 20:58:03 UTC
Permalink
Brian Candler writes:

> In the case of pop3d-ssl.rc and imapd-ssl.rc, I think you should just be
> able to replace $COURIERTLS with @bindir@/couriertls

Well, I've decided to fix all four scripts the same way, for consistency's
sake. I've thrown up a pair of new tarballs,
courier-0.40.0.20021025.1.tar.bz2 and courier-imap-1.6.0.20021025.tar.bz2
respectively. Quick feedback is appreciated; I'd like to replace the
released files with these tarballs.
Brian Candler
2002-10-28 10:15:05 UTC
Permalink
On Fri, Oct 25, 2002 at 06:57:35PM -0400, Sam Varshavchik wrote:
> > In the case of pop3d-ssl.rc and imapd-ssl.rc, I think you should just be
> > able to replace $COURIERTLS with @bindir@/couriertls
>
> Well, I've decided to fix all four scripts the same way, for consistency's
> sake. I've thrown up a pair of new tarballs,
> courier-0.40.0.20021025.1.tar.bz2 and courier-imap-1.6.0.20021025.tar.bz2
> respectively. Quick feedback is appreciated; I'd like to replace the
> released files with these tarballs.

I've rolled out courier-imap-1.6.0.20021025 and sqwebmail-3.4.0-20021026 and
a basic functionality test seems to show everything working fine.

Cheers,

Brian.
Sam Varshavchik
2002-10-27 02:03:02 UTC
Permalink
Devin Bayer writes:

> Jeff Potter wrote:
>> man makehosteddomains
>> > This may be a stupid question, but I am running a mail server that hosts
>> > several virtual domains through MySQL. I have everything running well,
>> > but would like to implement a "default" alias that would collect any
>> > mail that does not have a valid recipient for the domain and send it to
>> > a particular, modifiable account.
>
> Does something that serves the same function as this exist for
> non-hosteddomains. i.e. Can I have a fallback account for a domain
> listed in locals?

Yes, see dot-courier(5)
Sam Varshavchik
2002-10-27 12:34:02 UTC
Permalink
Michael Carmack writes:

> On Sun, Oct 27, 2002 at 08:02:26AM +0000, Michael Carmack wrote:
>>
>> I just saw this again tonight, and I didn't think to get an trace
>> before I fixed the problem. When it happens again--a couple weeks
>> maybe--I'll post more info. In the meantime, any suggestions?
>
> It happened sooner than I expected. The strace (minus the library
> bits) is fairly short, so I pasted it below:

The error occurs in the forked child process. The parent's trace is
meaningless.

Use strace -f next time.

>
>
>
> old_mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0
> x40015000
> old_mmap(NULL, 925116, PROT_READ|PROT_EXEC, MAP_PRIVATE, 3, 0) = 0x40016000
> mprotect(0x400f0000, 32188, PROT_NONE) = 0
> old_mmap(0x400f0000, 20480, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED, 3, 0xd9
> 000) = 0x400f0000
> old_mmap(0x400f5000, 11708, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED|MAP_ANON
> YMOUS, -1, 0) = 0x400f5000
> close(3) = 0
> mprotect(0x40016000, 892928, PROT_READ|PROT_WRITE) = 0
> mprotect(0x40016000, 892928, PROT_READ|PROT_EXEC) = 0
> personality(0 /* PER_??? */) = 0
> getpid() = 9030
> setgid(10001) = -1 EPERM (Operation not permitted)
> getuid() = 1
> setuid(1) = 0
> rt_sigaction(SIGPIPE, {SIG_IGN}, {SIG_DFL}, 8) = 0
> getuid() = 1
> brk(0) = 0x80506d0
> brk(0x80506f8) = 0x80506f8
> brk(0x8051000) = 0x8051000
> pipe([3, 4]) = 0
> pipe([5, 6]) = 0
> fork() = 9031
> sendmail: ERR: Permission denied
> --- SIGCHLD (Child exited) ---
> close(3) = 0
> close(6) = 0
> fcntl(4, F_SETFD, FD_CLOEXEC) = 0
> fcntl(4, F_GETFL) = 0x1 (flags O_WRONLY)
> fstat64(4, {st_mode=S_IFIFO|0600, st_size=0, ...}) = 0
> old_mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0
> x400f8000
> _llseek(4, 0, 0xbffff9b4, SEEK_CUR) = -1 ESPIPE (Illegal seek)
> fcntl(5, F_SETFD, FD_CLOEXEC) = 0
> fcntl(5, F_GETFL) = 0 (flags O_RDONLY)
> fstat64(5, {st_mode=S_IFIFO|0600, st_size=0, ...}) = 0
> old_mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0
> x400f9000
> _llseek(5, 0, 0xbffff9b4, SEEK_CUR) = -1 ESPIPE (Illegal seek)
> rt_sigaction(SIGINT, {0x804ade0, [], SA_RESTART|0x4000000}, {SIG_DFL}, 8) = 0
> rt_sigaction(SIGTERM, {0x804ade0, [], SA_RESTART|0x4000000}, {SIG_DFL}, 8) = 0
> rt_sigaction(SIGHUP, {0x804ade0, [], SA_RESTART|0x4000000}, {SIG_DFL}, 8) = 0
> rt_sigaction(SIGALRM, {0x8049e14, [], SA_RESTART|0x4000000}, {SIG_DFL}, 8) = 0
> alarm(1800) = 0
> write(4, "root\n", 5) = -1 EPIPE (Broken pipe)
> --- SIGPIPE (Broken pipe) ---
> fstat64(1, {st_mode=S_IFCHR|0622, st_rdev=makedev(136, 5), ...}) = 0
> old_mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0
> x400fa000
> ioctl(1, SNDCTL_TMR_TIMEBASE, {B38400 opost isig icanon echo ...}) = 0
> write(1, "432 Service temporarily unavaila"..., 37432 Service temporarily unavai
> lable.
> ) = 37
> munmap(0x400f8000, 4096) = 0
> munmap(0x400fa000, 4096) = 0
> _exit(75) = ?
>
>
>
>
> -------------------------------------------------------
> This SF.net email is sponsored by: ApacheCon, November 18-21 in
> Las Vegas (supported by COMDEX), the only Apache event to be
> fully supported by the ASF. http://www.apachecon.com
> _______________________________________________
> courier-users mailing list
> courier-***@lists.sourceforge.net
> Unsubscribe: https://lists.sourceforge.net/lists/listinfo/courier-users
Sam Varshavchik
2002-10-29 22:17:02 UTC
Permalink
Michael Carmack writes:

> On Sun, Oct 27, 2002 at 09:33:34AM -0500, Sam Varshavchik wrote:
>> >>
>> >>I just saw this again tonight, and I didn't think to get an trace
>> >>before I fixed the problem. When it happens again--a couple weeks
>> >>maybe--I'll post more info. In the meantime, any suggestions?
>> >
>> >It happened sooner than I expected. The strace (minus the library
>> >bits) is fairly short, so I pasted it below:
>>
>> The error occurs in the forked child process. The parent's trace is
>> meaningless.
>
> Ok, here's a trace from the user. This is only the bit around the point
> of failure. The full text is a bit much, but I can put it on the web if
> this isn't enough.
>
> Thanks.
>
>
> [pid 2876] read(5, "TZif\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\1\0\0\0\1\0"...,
> 44) = 44
> [pid 2876] fstat64(5, {st_mode=S_IFREG|0444, st_size=101, ...}) = 0
> [pid 2876] old_mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS
> , -1, 0) = 0x40177000
> [pid 2876] read(5, "\0\0\0\0\0\0Local time zone must be se"..., 4096) = 57
> [pid 2876] close(5) = 0
> [pid 2876] munmap(0x40177000, 4096) = 0
> [pid 2876] rt_sigaction(SIGPIPE, {0x4012ffac, [], 0x4000000}, {SIG_IGN}, 8) = 0
> [pid 2876] send(3, "<19>Oct 29 11:35:51 submit: No s"..., 53, 0) = 53


You need to look a little bit earlier than this. About 20-50 lines more.
Furthermore you need to use the -s option, because all the text messages are
cut off by default.
Sam Varshavchik
2002-11-01 10:55:04 UTC
Permalink
Michael Carmack writes:

> On Tue, Oct 29, 2002 at 07:16:08PM -0500, Sam Varshavchik wrote:
>>
>> You need to look a little bit earlier than this. About 20-50 lines more.
>> Furthermore you need to use the -s option, because all the text messages
>> are cut off by default.
>
> It got a bit long, so here it is on the web:
>
> http://antidote.karmak.org/sendmail-trace
>
> This seems to be happening more frequently , or maybe I'm just paying
> more attention now.

This is the problem:

[pid 9655] chdir("/pkg/courier/0.39.3/.i686-pc-linux-gnu/.karmak.1/var/run/tmp") = 0
[pid 9655] mkdir("103613", 0770) = -1 EACCES (Permission denied)
[pid 9655] _exit(0)

The permissions or the ownership of this directory are fscked up.

Check the permissions on the sendmail binary and that tmp subdirectory.
The tmp subdirectory must be owned by courier's uid/gid, and have mode 0770.
The sendmail binary must be uid root.

[***@ny root]# ls -l /usr/lib/courier/bin/sendmail
-r-s--x--x 1 root daemon 41334 Oct 25 23:16 /usr/lib/courier/bin/sendmail

A new spool subdirectory is created every couple of hours. I think you
lost the setuid bit on sendmail, meaning that every couple of hours
everything will break until you run sendmail as root, which will
finally be able to run the submit child process as the courier user,
and create the new spool subdirectory.
Sam Varshavchik
2002-10-29 10:44:05 UTC
Permalink
Robin Bowes writes:

> Sam,
>
>> FAM is optional, however it does provide for some value-added
>> functionality. In the event that the additional value-added functionality
>> is expected, the dependency provides a hint that the required prerequisites
>> are missing.
>
> Ok. Can you document this in the php build instructions?

What does PHP have to do with anything?

> libstdc++-devel = 2.96 is needed by (installed) gcc-c++-2.96-110
>
> So, to install FAM, I need to upgrade, amongst other things, php and gcc.

Only because you're trying to install FAM from RH 8.0 on RH 7.2.

> I revisited this following your reply and found that I actually have
> libstdc++.so.5 installed (from libstdc++-3.2-0.1) and I managed to
> install fam using the "--nodeps" option.

Congratulations -- you have a completely broken server.
Robin Bowes
2002-10-29 13:00:07 UTC
Permalink
Sam,

> Robin Bowes writes:
>
>> Sam,
>>
>>> FAM is optional, however it does provide for some value-added
>>> functionality. In the event that the additional value-added functionality
>>> is expected, the dependency provides a hint that the required
prerequisites
>>> are missing.
>>
>> Ok. Can you document this in the php build instructions?
>
> What does PHP have to do with anything?

Okay, Okay, spot the typo. For "php" read "rpm". See my previous mail for
suggested enhancements to rpm build documentation.

>
>> libstdc++-devel = 2.96 is needed by (installed) gcc-c++-2.96-110
>>
>> So, to install FAM, I need to upgrade, amongst other things, php and gcc.
>
> Only because you're trying to install FAM from RH 8.0 on RH 7.2.

See one of my previous replies. My server is somewhere between 7.2 and 8.0
following my abortive attempts to use apache2 (went back to apache1 as php
isn't 100% on apache2). I'm not trying to justify it - just trying to
explain why I might not want to install FAM on my box.

>
>> I revisited this following your reply and found that I actually have
>> libstdc++.so.5 installed (from libstdc++-3.2-0.1) and I managed to
>> install fam using the "--nodeps" option.
>
> Congratulations -- you have a completely broken server.

Hmmm. What leads you to that conclusion? Ok, so it's not going to win any
prizes for the cleanest package set but "completely broken"? I think not.

All of this is getting away from the thrust of my original point which is
that courier-imap does not *need* FAM - it is an option that adds
functionality, but is not essential. I came across this fact by chance
when browsing through the Changelog.

Personally, I would remove fam-devel from the Deps. list in the spec file
but, failing that, I think it would be useful to document the fact that
this package is optional in the rpm build instructions.

Again, see my previous mail for suggested enhancements to rpm build
documentation.

Cheers,

R.
--
Robin Bowes | http://robinbowes.com
Sam Varshavchik
2002-10-29 20:57:08 UTC
Permalink
Robin Bowes writes:

>>> libstdc++.so.5 installed (from libstdc++-3.2-0.1) and I managed to
>>> install fam using the "--nodeps" option.
>>
>> Congratulations -- you have a completely broken server.
>
> Hmmm. What leads you to that conclusion? Ok, so it's not going to win any
> prizes for the cleanest package set but "completely broken"? I think not.

The package dependencies are not picked at random. --nodeps isn't an option
that automatically waves a magic wand and makes everything all right. It's
there for cases where whoever built the package screwed up, and did not
specify the correct dependencies.

If you install a package that explicitly lists a dependency on a particular
version of the standard C runtime library, and you force-install it, you now
have a binary API incompatibility with the runtime C library. If you're
lucky, the package will simply not work, and refuse to run right off the
bat. But that's rare. The typical result are random, subtle faults; such
as daemon processes crashing at random times, or locking up completely. A
typical ABI change involves an extra, or a different argument to some
library call. Now, the library will attempt to read the argument that's
not provided by the userapp, and end up reading random garbage, with
completely unpredictable results.

Anyone who uses --nodeps on a production server needs to have his head
examined.

> All of this is getting away from the thrust of my original point which is
> that courier-imap does not *need* FAM - it is an option that adds
> functionality, but is not essential.

It doesn't, but that has nothing to do with what's listed in the spec file.
The spec file provides instructions for building the package on a released
distribution. Since all recent versions of Red Hat have FAM, it is listed
as a dependency, since having it installed results in enhanced
functionality at run time.
Robin Bowes
2002-10-30 13:02:02 UTC
Permalink
Sam,

> Anyone who uses --nodeps on a production server needs to have his head
> examined.

I agree. However, my box is not a production server, it's one I experiment
with at home.

>> All of this is getting away from the thrust of my original point which is
>> that courier-imap does not *need* FAM - it is an option that adds
>> functionality, but is not essential.
>
> It doesn't, but that has nothing to do with what's listed in the spec file.
> The spec file provides instructions for building the package on a released
> distribution. Since all recent versions of Red Hat have FAM, it is listed
> as a dependency, since having it installed results in enhanced
> functionality at run time.

It was my understanding that the dependencies in the spec file tell rpm
what packages are required to build/install the package. Nothing to do
with a specific distribution.

FAM may be available to "all recent versions of Red Hat" but that doesn't
mean that it is installed. xinetd is available to "all recent versions of
Red Hat" but that doesn't mean it is installed on every server.

Anyway, regardless of this point, and as I said earlier, it's not for me
to tell you how to build your software. I can only suggest what I, as a
user, feel would enhance my experience of using it. If you re-read my
mail, I have not suggested that you remove the dependency (though I
personally would) but rather have suggested that it would make life easier
if alll such dependencies are listed in the rpm build instructions.

Regards,

R.
--
Robin Bowes | http://robinbowes.com
Sam Varshavchik
2002-10-29 20:59:03 UTC
Permalink
Alessandro Vesely writes:

> My filter complains it cannot understand such MIME structure,
> it dumps a copy of the message and blocks it. So I have to
> manually look at it, run reformime -r and send it over again.
> Reformime returns a valid MIME structure.

I doubt that the output from reformime includes all the attachments,
unmolested. Count the boundaries in the reformatted mail.
Alessandro Vesely
2002-10-31 10:51:03 UTC
Permalink
Sam wrote:
> Alessandro Vesely writes:
>
>> My filter complains it cannot understand such MIME structure,
>> it dumps a copy of the message and blocks it. So I have to
>> manually look at it, run reformime -r and send it over again.
>> Reformime returns a valid MIME structure.
>
> I doubt that the output from reformime includes all the attachments,
> unmolested. Count the boundaries in the reformatted mail.
>

In that case it does, because the last attachment was absolutely
(i.e. without even a blank line) empty. In general yes, a boundary
may live as text in the body of another entity. Will some "smart"
client recognize it and decode the attachment?

When MIME came out it standardized and enhanced the practice of
hand writing boundaries, typically like the following:
--8<--- cut here ----

The standard says the boundary value should never appear in the
body, but it is not clear how to handle a message when that
prescription is not observed.

Salutations
Ale
Sam Varshavchik
2002-10-31 11:27:03 UTC
Permalink
Alessandro Vesely writes:

> Sam wrote:
>> Alessandro Vesely writes:
>>
>>> My filter complains it cannot understand such MIME structure,
>>> it dumps a copy of the message and blocks it. So I have to
>>> manually look at it, run reformime -r and send it over again.
>>> Reformime returns a valid MIME structure.
>>
>> I doubt that the output from reformime includes all the attachments,
>> unmolested. Count the boundaries in the reformatted mail.
>>
>
> In that case it does, because the last attachment was absolutely
> (i.e. without even a blank line) empty. In general yes, a boundary
> may live as text in the body of another entity. Will some "smart"
> client recognize it and decode the attachment?
>
> When MIME came out it standardized and enhanced the practice of
> hand writing boundaries, typically like the following:
> --8<--- cut here ----
>
> The standard says the boundary value should never appear in the
> body, but it is not clear how to handle a message when that
> prescription is not observed.

It's quite clear about it: "--fooblah" is fine, as long as fooblah does not
match any of the actual multipart mime boundary delimiter strings. And my
parser does exactly that. Literally. If the parser recognizes a delimiter
string, it's going to close that MIME section. But, anything else that
looks like a boundary delimiter string is going to get ignored.





--
Sam
Juha Saarinen
2002-10-30 18:14:13 UTC
Permalink
On Wed, 30 Oct 2002, Robin Bowes wrote:

> FAM may be available to "all recent versions of Red Hat" but that
> doesn't mean that it is installed. xinetd is available to "all recent
> versions of Red Hat" but that doesn't mean it is installed on every
> server.

I think you're missing the point here. Sam and other software packagers
make reasonable assumptions based on standard installations of e.g. RHL.
FAM and Xinetd are installed by RHL by default with RHL.

Of course, some people remove these or add something else. In doing so,
another assumption is reasonable: that these people know what they're
doing, and are capable of working with a non-standard distribution instead
of griping about it on mailing lists.


--
Juha Saarinen
Robin Bowes
2002-10-30 21:07:03 UTC
Permalink
Juha,

> On Wed, 30 Oct 2002, Robin Bowes wrote:
>
>> FAM may be available to "all recent versions of Red Hat" but that
>> doesn't mean that it is installed. xinetd is available to "all recent
>> versions of Red Hat" but that doesn't mean it is installed on every
>> server.
>
> I think you're missing the point here. Sam and other software packagers
> make reasonable assumptions based on standard installations of e.g. RHL.
> FAM and Xinetd are installed by RHL by default with RHL.
>
> Of course, some people remove these or add something else. In doing so,
> another assumption is reasonable: that these people know what they're
> doing, and are capable of working with a non-standard distribution instead
> of griping about it on mailing lists.

Au contraire, I think *you* (and Sam) are missing the point.

Sam writes software which doesn't cost me anything. I am grateful for that.

He can and does release/build it it whatever way he sees fit. That's his
prerogative.

As you reasonably assume, I *am* capable of working with a non-standard
distribution and I'm not griping about the fact that it creates certain
challenges related to software dependencies.

All I'm saying is that a bit of documentation would make life easier.

For example, somewhere in the RPM build instructions, it wouldn't be
unreasonable to say "In order to build this package you need the following
packages installed: blah, blah, FAM, blah"

After all, it's *optional*, so if I'm building from source (./configure ;
make ; make install) then the configure script will detect that I don't
have (for example) FAM installed and alter the build config accordingly.

Is that unreasonable? Am I really griping?

R.

PS. One small point regarding FAM and Redhat; I started with a "standard"
Redhat 7.2 install. FAM was not included. Sure, it's on the disks, but if
you choose a minimal install it does not get installed.
--
Robin Bowes | http://robinbowes.com
Jesse Keating
2002-10-30 21:18:04 UTC
Permalink
On Wed, 30 Oct 2002 23:06:38 -0000 (GMT)
"Robin Bowes" <robin-***@robinbowes.com> wrote:

# For example, somewhere in the RPM build instructions, it wouldn't be
# unreasonable to say "In order to build this package you need the
# following packages installed: blah, blah, FAM, blah"
#
# After all, it's *optional*, so if I'm building from source
# (./configure ; make ; make install) then the configure script will
# detect that I don't have (for example) FAM installed and alter the
# build config accordingly.

You mean the Build Requires: part of the spec file? lists all the
packages that you would need to build the whole thing. What more do
you want?

--
Jesse Keating
j2Solutions.net
Mondo DevTeam (www.mondorescue.org)

Was I helpful? Let others know:
http://svcs.affero.net/rm.php?r=jkeating
Robin Bowes
2002-10-31 15:37:03 UTC
Permalink
Jesse,

> On Wed, 30 Oct 2002 23:06:38 -0000 (GMT)
> "Robin Bowes" <robin-***@robinbowes.com> wrote:
>
> # For example, somewhere in the RPM build instructions, it wouldn't be
> # unreasonable to say "In order to build this package you need the
> # following packages installed: blah, blah, FAM, blah"
> #
> # After all, it's *optional*, so if I'm building from source
> # (./configure ; make ; make install) then the configure script will
> # detect that I don't have (for example) FAM installed and alter the
> # build config accordingly.
>
> You mean the Build Requires: part of the spec file? lists all the
> packages that you would need to build the whole thing. What more do
> you want?

The rpm build instructions say:

"To build a binary RPM package, enter:

rpm -ta courier-imap-1.4.6.tar.gz"

They also say:

"IMPORTANT: even though building an RPM saves you the headaches of
manually compiling the code, you should still read the INSTALL file as it
contains many important details you need to know."

There is no mention of the spec file in the INSTALL file.

So, unless I extract the tarball and

a. extract the tarball, and
b. know what a spec file is, and
c. know what to look for in the spec file,

then I will never know about the Build Requires section.

Let's face it, one of the main attractions of the rpm paradigm is that it
means that, ahem, less technically adept people can still build and
install software. So potentially, the users who are perhaps most likely to
be using the rpm are those that are least likely to know about the
BuildPreReq section in the spec file.

What it all boils down to is that I'd like to see reference to the spec
file in the rpm build instructions.

Ideally, I'd like to see a list of build dependencies as well, but I
appreciate that would introduce additional maintenance activities to
ensure the spec file and the build instructions are in sync.

I hope that makes my point clear.

R.
--
Robin Bowes | http://robinbowes.com
Jesse Keating
2002-10-31 16:57:08 UTC
Permalink
On Thu, 31 Oct 2002 17:35:56 -0000 (GMT)
"Robin Bowes" <robin-***@robinbowes.com> wrote:

# Let's face it, one of the main attractions of the rpm paradigm is
# that it means that, ahem, less technically adept people can still
# build and install software. So potentially, the users who are
# perhaps most likely to be using the rpm are those that are least
# likely to know about the BuildPreReq section in the spec file.

Thats assuming you are using binary rpms. When you get into dealing
with sourcerpms, or sources with a spec file then you do have to have
some knowledge about software compiling. The BuildReq section is
evoked when you issue a rpmbuild -ta or -ba, you never have to open up
the spec file to make sure you have all that it requires, as it will
tell you.

# What it all boils down to is that I'd like to see reference to the
# spec file in the rpm build instructions.

man rpm

# Ideally, I'd like to see a list of build dependencies as well, but I
# appreciate that would introduce additional maintenance activities to
# ensure the spec file and the build instructions are in sync.

I think you need to read about RPMS a bit more. When you run rpmbuild
-ta file.tar.gz it looks at the spec file, and checks for
dependencies. Since you didn't have fam, it told you that you needed
fam and fam-devel to continue. Thats BuildReqs in action. If you
wanted to look at the spec file, you can.

--
Jesse Keating
j2Solutions.net
Mondo DevTeam (www.mondorescue.org)

Was I helpful? Let others know:
http://svcs.affero.net/rm.php?r=jkeating
Sam Varshavchik
2002-10-30 22:41:02 UTC
Permalink
Russell Premont writes:

> I updated my paths to include openssl directory and lib directory but now I
> get this error:
>
> ar cru libspipe.a spipe.o
> ranlib libspipe.a
> gcc -DHAVE_CONFIG_H -I. -I. -I. -I./.. -I.. -Wall -g -O2 -c
> libcouriertls.c
> In file included from libcouriertls.c:8:
> libcouriertls.h:23:32: openssl/ssl.h: No such file or directory
> libcouriertls.h:24:32: openssl/err.h: No such file or directory

These files, OpenSSL include files, are either not installed, or you're
pointing to the wrong directory.

You do not point CPPFLAGS to the include/openssl directory, you point it to
the include directory.
Sam Varshavchik
2002-10-31 06:03:07 UTC
Permalink
Binand Raj S. writes:

> On Wed, Oct 30, 2002 at 05:06:48PM +0000, Brian Candler wrote:
>> +char *q=malloc(sizeof("UID=")+strlen(userid));
>
> And later...
>
>> + strcat(strcpy(q, "UID="), userid);
>
> Just FYI, it looks to me there is an off by one error here.

No there isn't.
Russell Premont
2002-10-31 14:03:04 UTC
Permalink
I apologize but I am a newbie to this and I do not understand you reply. I
have modified my environment to the following:

bash-2.05$ env
TZ=US/Eastern
HOSTNAME=Star5
LD_LIBRARY_PATH=/usr/local/lib:/usr/lib:/usr/local/ssl/lib:/usr/local/ssl/in
clude:/usr/local/ssl:/usr/openwin/lib:/usr/dt/lib:/usr/ccs/lib:/usr/ucblib:/
usr/local/mysql/lib/mysql
MACHTYPE=sparc-sun-solaris2.9
SHLVL=1
SHELL=/bin/sh
HOSTTYPE=sparc
OSTYPE=solaris2.9
TERM=xterm
PATH=/usr/local/bin:/usr/local/lib:/usr/local/ssl:/usr/local/ssl/lib:/usr/lo
cal/ssl/include:/usr/ccs/bin/:/usr/local/sbin:/lib:/usr/lib:/bin:/sbin:/usr/
bin:/usr/sbin:/usr/ucb
SSH_TTY=/dev/pts/2
_=/bin/env

I have also changed my configuration options to the following:
--prefix=/usr/local/courier --with-ipv6 --with-db=gdbm --with-mailuser=couri
er-mta --with-mailgroup=courier-mta --with-shellpath=/usr/local/bin:/usr/loc
al/lib:/usr/local/ssl:/usr/local/ssl/lib:/usr/local/ssl/include:/usr/ccs/bin
/:/usr/local/sbin:/lib:/usr/lib:/bin:/sbin:/usr/bin:/usr/sbin:/usr/ucb --wit
h-paranoid-smtpext --with-random=/dev/urandom --with-certdb=pathname --with-
aspell=/usr/local/bin/aspell --enable-unicode --enable-https=auto --enable-h
ardtimeout=2400 --enable-softtimeout=1200 --with-cachedir=/usr/local/courier
/var/webmail-logincache --with-trashquota --disable-autorenamesent --enable-
mimetypes


but I still get the following:

libcouriertls.h:23:32: openssl/ssl.h: No such file or directory
libcouriertls.h:24:32: openssl/err.h: No such file or directory
In file included from libcouriertls.c:8:
libcouriertls.h:60: parse error before '*' token
libcouriertls.h:117: parse error before '*' token
libcouriertls.h:117: warning: type defaults to `int' in declaration of
`tls_crea
libcouriertls.h:117: warning: data definition has no type or storage class
libcouriertls.h:118: parse error before '*' token
libcouriertls.h:125: parse error before '*' token
libcouriertls.h:125: parse error before '*' token

I have looked in the system and both files are in
/usr/local/ssl/include/openssl. How can I determine where courier is trying
to look for them?

Thank You

Russell Premont

-----Original Message-----
From: courier-users-***@lists.sourceforge.net
[mailto:courier-users-***@lists.sourceforge.net]On Behalf Of Sam
Varshavchik
Sent: Wednesday, October 30, 2002 7:41 PM
To: Courier-Users
Subject: [courier-users] Re: Compile error


Russell Premont writes:

> I updated my paths to include openssl directory and lib directory but now
I
> get this error:
>
> ar cru libspipe.a spipe.o
> ranlib libspipe.a
> gcc -DHAVE_CONFIG_H -I. -I. -I. -I./.. -I.. -Wall -g -O2 -c
> libcouriertls.c
> In file included from libcouriertls.c:8:
> libcouriertls.h:23:32: openssl/ssl.h: No such file or directory
> libcouriertls.h:24:32: openssl/err.h: No such file or directory

These files, OpenSSL include files, are either not installed, or you're
pointing to the wrong directory.

You do not point CPPFLAGS to the include/openssl directory, you point it to
the include directory.




-------------------------------------------------------
This sf.net email is sponsored by: Influence the future
of Java(TM) technology. Join the Java Community
Process(SM) (JCP(SM)) program now.
http://ads.sourceforge.net/cgi-bin/redirect.pl?sunm0004en
Matt Hyclak
2002-10-31 14:25:10 UTC
Permalink
On Thu, Oct 31, 2002 at 11:01:09AM -0500, Russell Premont enlightened us:
> I apologize but I am a newbie to this and I do not understand you reply. I
> have modified my environment to the following:
>
> bash-2.05$ env
> TZ=US/Eastern
> HOSTNAME=Star5
> LD_LIBRARY_PATH=/usr/local/lib:/usr/lib:/usr/local/ssl/lib:/usr/local/ssl/in
> clude:/usr/local/ssl:/usr/openwin/lib:/usr/dt/lib:/usr/ccs/lib:/usr/ucblib:/
> usr/local/mysql/lib/mysql
> MACHTYPE=sparc-sun-solaris2.9

For what it's worth, LD_LIBRARY_PATH is supposed to be deprecated from
Solaris 8 onward. crle(1) has replaced it. That could be part of the problem.

> but I still get the following:
>
> libcouriertls.h:23:32: openssl/ssl.h: No such file or directory
> libcouriertls.h:24:32: openssl/err.h: No such file or directory
> In file included from libcouriertls.c:8:
> libcouriertls.h:60: parse error before '*' token
> libcouriertls.h:117: parse error before '*' token
> libcouriertls.h:117: warning: type defaults to `int' in declaration of
> `tls_crea
> libcouriertls.h:117: warning: data definition has no type or storage class
> libcouriertls.h:118: parse error before '*' token
> libcouriertls.h:125: parse error before '*' token
> libcouriertls.h:125: parse error before '*' token
>
> I have looked in the system and both files are in
> /usr/local/ssl/include/openssl. How can I determine where courier is trying
> to look for them?
>

config.log should show you what it was trying to do when it failed.

Matt

--
Matt Hyclak
Department of Mathematics
Ohio University
(740) 593-1263
Russell Premont
2002-10-31 14:40:05 UTC
Permalink
Configure runs fine. I get the error when I run make. For some reason
courier cant find ssl.h or err.h and I am not sure what to do. Sam sent me
thi reply last night

>Russell Premont writes:

>> I updated my paths to include openssl directory and lib directory but now
I
>> get this error:
>>
>> ar cru libspipe.a spipe.o
>> ranlib libspipe.a
>> gcc -DHAVE_CONFIG_H -I. -I. -I. -I./.. -I.. -Wall -g -O2 -c
>> libcouriertls.c
>> In file included from libcouriertls.c:8:
>> libcouriertls.h:23:32: openssl/ssl.h: No such file or directory
>> libcouriertls.h:24:32: openssl/err.h: No such file or directory
>
>These files, OpenSSL include files, are either not installed, or you're
>pointing to the wrong directory.
>
>You do not point CPPFLAGS to the include/openssl directory, you point it to
>the include directory.

but as I said I am a newbie to this and do not know how to proceed.

Thank You,
Russell Premont

-----Original Message-----
From: Matt Hyclak [mailto:***@math.ohiou.edu]
Sent: Thursday, October 31, 2002 11:24 AM
To: Russell Premont
Cc: Courier Users Mailing List
Subject: Re: [courier-users] Re: Compile error


On Thu, Oct 31, 2002 at 11:01:09AM -0500, Russell Premont enlightened us:
> I apologize but I am a newbie to this and I do not understand you reply. I
> have modified my environment to the following:
>
> bash-2.05$ env
> TZ=US/Eastern
> HOSTNAME=Star5
>
LD_LIBRARY_PATH=/usr/local/lib:/usr/lib:/usr/local/ssl/lib:/usr/local/ssl/in
>
clude:/usr/local/ssl:/usr/openwin/lib:/usr/dt/lib:/usr/ccs/lib:/usr/ucblib:/
> usr/local/mysql/lib/mysql
> MACHTYPE=sparc-sun-solaris2.9

For what it's worth, LD_LIBRARY_PATH is supposed to be deprecated from
Solaris 8 onward. crle(1) has replaced it. That could be part of the
problem.

> but I still get the following:
>
> libcouriertls.h:23:32: openssl/ssl.h: No such file or directory
> libcouriertls.h:24:32: openssl/err.h: No such file or directory
> In file included from libcouriertls.c:8:
> libcouriertls.h:60: parse error before '*' token
> libcouriertls.h:117: parse error before '*' token
> libcouriertls.h:117: warning: type defaults to `int' in declaration of
> `tls_crea
> libcouriertls.h:117: warning: data definition has no type or storage class
> libcouriertls.h:118: parse error before '*' token
> libcouriertls.h:125: parse error before '*' token
> libcouriertls.h:125: parse error before '*' token
>
> I have looked in the system and both files are in
> /usr/local/ssl/include/openssl. How can I determine where courier is
trying
> to look for them?
>

config.log should show you what it was trying to do when it failed.

Matt

--
Matt Hyclak
Department of Mathematics
Ohio University
(740) 593-1263
cml
2002-10-31 18:01:07 UTC
Permalink
Your problem is either you don't have openssl installed, or it's installed in
another place than the one configure expects.

If you did install, try setting a symlink:
ln -s /wherever-your-ssl-is/include/openssl /usr/include/openssl

Solved it for me...

sytse

On Thursday 31 October 2002 17:38, Russell Premont wrote:
> Configure runs fine. I get the error when I run make. For some reason
> courier cant find ssl.h or err.h and I am not sure what to do. Sam sent me
> thi reply last night
>
> >Russell Premont writes:
> >> I updated my paths to include openssl directory and lib directory but
> >> now
>
> I
>
> >> get this error:
> >>
> >> ar cru libspipe.a spipe.o
> >> ranlib libspipe.a
> >> gcc -DHAVE_CONFIG_H -I. -I. -I. -I./.. -I.. -Wall -g -O2 -c
> >> libcouriertls.c
> >> In file included from libcouriertls.c:8:
> >> libcouriertls.h:23:32: openssl/ssl.h: No such file or directory
> >> libcouriertls.h:24:32: openssl/err.h: No such file or directory
> >
> >These files, OpenSSL include files, are either not installed, or you're
> >pointing to the wrong directory.
> >
> >You do not point CPPFLAGS to the include/openssl directory, you point it
> > to the include directory.
>
> but as I said I am a newbie to this and do not know how to proceed.
>
> Thank You,
> Russell Premont
>
> -----Original Message-----
> From: Matt Hyclak [mailto:***@math.ohiou.edu]
> Sent: Thursday, October 31, 2002 11:24 AM
> To: Russell Premont
> Cc: Courier Users Mailing List
> Subject: Re: [courier-users] Re: Compile error
>
> On Thu, Oct 31, 2002 at 11:01:09AM -0500, Russell Premont enlightened us:
> > I apologize but I am a newbie to this and I do not understand you reply.
> > I have modified my environment to the following:
> >
> > bash-2.05$ env
> > TZ=US/Eastern
> > HOSTNAME=Star5
>
> LD_LIBRARY_PATH=/usr/local/lib:/usr/lib:/usr/local/ssl/lib:/usr/local/ssl/i
>n
>
> clude:/usr/local/ssl:/usr/openwin/lib:/usr/dt/lib:/usr/ccs/lib:/usr/ucblib:
>/
>
> > usr/local/mysql/lib/mysql
> > MACHTYPE=sparc-sun-solaris2.9
>
> For what it's worth, LD_LIBRARY_PATH is supposed to be deprecated from
> Solaris 8 onward. crle(1) has replaced it. That could be part of the
> problem.
>
> > but I still get the following:
> >
> > libcouriertls.h:23:32: openssl/ssl.h: No such file or directory
> > libcouriertls.h:24:32: openssl/err.h: No such file or directory
> > In file included from libcouriertls.c:8:
> > libcouriertls.h:60: parse error before '*' token
> > libcouriertls.h:117: parse error before '*' token
> > libcouriertls.h:117: warning: type defaults to `int' in declaration of
> > `tls_crea
> > libcouriertls.h:117: warning: data definition has no type or storage
> > class libcouriertls.h:118: parse error before '*' token
> > libcouriertls.h:125: parse error before '*' token
> > libcouriertls.h:125: parse error before '*' token
> >
> > I have looked in the system and both files are in
> > /usr/local/ssl/include/openssl. How can I determine where courier is
>
> trying
>
> > to look for them?
>
> config.log should show you what it was trying to do when it failed.
>
> Matt
>
> --
> Matt Hyclak
> Department of Mathematics
> Ohio University
> (740) 593-1263
>
>
>
> -------------------------------------------------------
> This sf.net email is sponsored by: Influence the future
> of Java(TM) technology. Join the Java Community
> Process(SM) (JCP(SM)) program now.
> http://ads.sourceforge.net/cgi-bin/redirect.pl?sunm0004en
> _______________________________________________
> courier-users mailing list
> courier-***@lists.sourceforge.net
> Unsubscribe: https://lists.sourceforge.net/lists/listinfo/courier-users
Russell Premont
2002-10-31 19:05:02 UTC
Permalink
Thanks for the help. That did the trick. I did not know where courier was
looking for those files. I updated my paths but it still did not work.

Than you again,
Russell Premont

-----Original Message-----
From: courier-users-***@lists.sourceforge.net
[mailto:courier-users-***@lists.sourceforge.net]On Behalf Of cml
Sent: Thursday, October 31, 2002 3:00 PM
To: Russell Premont; Courier-Users
Cc: Matt Hyclak
Subject: Re: [courier-users] Re: Compile error


Your problem is either you don't have openssl installed, or it's installed
in
another place than the one configure expects.

If you did install, try setting a symlink:
ln -s /wherever-your-ssl-is/include/openssl /usr/include/openssl

Solved it for me...

sytse

On Thursday 31 October 2002 17:38, Russell Premont wrote:
> Configure runs fine. I get the error when I run make. For some reason
> courier cant find ssl.h or err.h and I am not sure what to do. Sam sent me
> thi reply last night
>
> >Russell Premont writes:
> >> I updated my paths to include openssl directory and lib directory but
> >> now
>
> I
>
> >> get this error:
> >>
> >> ar cru libspipe.a spipe.o
> >> ranlib libspipe.a
> >> gcc -DHAVE_CONFIG_H -I. -I. -I. -I./.. -I.. -Wall -g -O2 -c
> >> libcouriertls.c
> >> In file included from libcouriertls.c:8:
> >> libcouriertls.h:23:32: openssl/ssl.h: No such file or directory
> >> libcouriertls.h:24:32: openssl/err.h: No such file or directory
> >
> >These files, OpenSSL include files, are either not installed, or you're
> >pointing to the wrong directory.
> >
> >You do not point CPPFLAGS to the include/openssl directory, you point it
> > to the include directory.
>
> but as I said I am a newbie to this and do not know how to proceed.
>
> Thank You,
> Russell Premont
>
> -----Original Message-----
> From: Matt Hyclak [mailto:***@math.ohiou.edu]
> Sent: Thursday, October 31, 2002 11:24 AM
> To: Russell Premont
> Cc: Courier Users Mailing List
> Subject: Re: [courier-users] Re: Compile error
>
> On Thu, Oct 31, 2002 at 11:01:09AM -0500, Russell Premont enlightened us:
> > I apologize but I am a newbie to this and I do not understand you reply.
> > I have modified my environment to the following:
> >
> > bash-2.05$ env
> > TZ=US/Eastern
> > HOSTNAME=Star5
>
>
LD_LIBRARY_PATH=/usr/local/lib:/usr/lib:/usr/local/ssl/lib:/usr/local/ssl/i
>n
>
>
clude:/usr/local/ssl:/usr/openwin/lib:/usr/dt/lib:/usr/ccs/lib:/usr/ucblib:
>/
>
> > usr/local/mysql/lib/mysql
> > MACHTYPE=sparc-sun-solaris2.9
>
> For what it's worth, LD_LIBRARY_PATH is supposed to be deprecated from
> Solaris 8 onward. crle(1) has replaced it. That could be part of the
> problem.
>
> > but I still get the following:
> >
> > libcouriertls.h:23:32: openssl/ssl.h: No such file or directory
> > libcouriertls.h:24:32: openssl/err.h: No such file or directory
> > In file included from libcouriertls.c:8:
> > libcouriertls.h:60: parse error before '*' token
> > libcouriertls.h:117: parse error before '*' token
> > libcouriertls.h:117: warning: type defaults to `int' in declaration of
> > `tls_crea
> > libcouriertls.h:117: warning: data definition has no type or storage
> > class libcouriertls.h:118: parse error before '*' token
> > libcouriertls.h:125: parse error before '*' token
> > libcouriertls.h:125: parse error before '*' token
> >
> > I have looked in the system and both files are in
> > /usr/local/ssl/include/openssl. How can I determine where courier is
>
> trying
>
> > to look for them?
>
> config.log should show you what it was trying to do when it failed.
>
> Matt
>
> --
> Matt Hyclak
> Department of Mathematics
> Ohio University
> (740) 593-1263
>
>
>
> -------------------------------------------------------
> This sf.net email is sponsored by: Influence the future
> of Java(TM) technology. Join the Java Community
> Process(SM) (JCP(SM)) program now.
> http://ads.sourceforge.net/cgi-bin/redirect.pl?sunm0004en
> _______________________________________________
> courier-users mailing list
> courier-***@lists.sourceforge.net
> Unsubscribe: https://lists.sourceforge.net/lists/listinfo/courier-users



-------------------------------------------------------
This sf.net email is sponsored by: Influence the future
of Java(TM) technology. Join the Java Community
Process(SM) (JCP(SM)) program now.
http://ads.sourceforge.net/cgi-bin/redirect.pl?sunm0004en
Sam Varshavchik
2002-10-31 23:06:06 UTC
Permalink
Kris Kelley writes:

> Just to be thorough: Due to a different problem, we had some message
> files generated between 25 October and 28 October end up with incorrect
> timestamps. This resulted in the INTERNALDATE for these messages being
> incorrect, even though the message headers had correct timestamps in all
> the relevant places.

No.

> Could that have somehow aggravated this UID
> problem? I'm ready to file this one under MS stupidity, but I want to
> make sure there aren't any other influences I have control over that
> might be factors.

See what the Maildir/courierimapuiddb file says (make sure you check the
right folder's courierimapuiddb file). That's where messages' UIDs are
saved. If you do not see any dupes there, OE definitely pulled it out of
its ass.
Sam Varshavchik
2002-11-01 03:53:02 UTC
Permalink
Russell Premont writes:

> Here are the results of the diff. testsuite1.txt is the results of running
> LC_ALL=C; export LC_ALL; ./testsuite | ./testsuitefix.pl

Look closely, and try again. Before running a diff, you have to sort the
output first:

>> > make[1]: Entering directory `/export/home/russ/courier-0.39.3/imap'
>> > LC_ALL=C; export LC_ALL; ./testsuite | ./testsuitefix.pl | sort |
> cmp -s - ./testsuite.txt
Loading...