gmane.mail.qmail.toaster.devel

[qmailtoaster-devel] -help

Thibault Richard — 2013-05-17T11:58:19

[qmailtoaster-devel] Greeting failed, but delivery success

Clement Thomas — 2013-03-25T15:15:17

Hi,
    We use qmail with ezmlm for mailinglist service

[root< at >xx-yyy ~]# rpm -qa|egrep '(ezmlm|qmail)'
qmail-toaster-1.03-1.3.25
ezmlm-toaster-0.53.324-1.3.3

There was postfix SMTP failure with one of the MX to which qmail tried 
mail delivery and got a 4xx reply. Log says ZConnected to , but greeting 
failed, server says "All ports busy". Despite this message, the delivery 
was success. exact log at 
http://pastie.org/private/stwlr6jmvhiljzbed0i3ag All the mail delivery 
tried via the above MX is lost, but qmail log reports success with above 
ZConnected message.  Z means delivery deferral, if i am not wrong.

Regards,
Clement

[qmailtoaster-devel] dovecot lda discussion

Eric Shubert — 2012-09-24T16:57:14

This thread just popped up on the vpopmail list today.
http://www.mail-archive.com/vchkpw< at >inter7.com/msg28208.html

We'll be going to use dovecot-lda eventually, perhaps after an initial 
dovecot implementation, so as not to change too many things at once.

Does anyone have any thoughts about this? Just throwing this out here at 
this point. I don't expect anything to happen with it for several months.

[qmailtoaster-devel] vqadmin 1.4 BROKEN (?)

Dan McAllister — 2012-08-07T15:21:34

Greetings QMT developers...

I've been off working on other things and just got back to doing some 
QMT testing this morning... and am finding that vqadmin version 1.4 is 
broken for 64-bit arch.
The error appears in the configure portion of the make, where in spite 
of being called with a --with cnt5064 flag, the configure call is being 
made with
   --host=x86_64-unknown-linux-gnu
and
   --build=x86_64-unknown-linux-gnu
flags that are later determined to be unsupported.

I have confirmed that the error persists on both 1.4.0 and 1.4.1

I don't know who worked on these... and I haven't checked them against a 
32-bit build (is anyone still using 32-bit COS?) or against COS5 yet... 
but I'm assuming this is a simple error to fix, I just don't know who to 
look to to get it fixed!

Thanks to all


Dan McAllister

[qmailtoaster-devel] Roadmap update (short term)

Eric Shubert — 2012-08-06T17:41:04

Just an update to keep everyone in the loop here.

The 1.4.x versions are upgrades to upstream versions. This phase is 
essentially complete. Bharath will be making a few more 1.4.x versions 
that are compatible with the stock php53 and php6 as well. Some packages 
have a 1.4.x version, and some do not.

All packages will have a 1.5.0 version, which will be the first Cos6 
compliant version. There will also be an updated QTP package which 
corresponds with these. I'm beginning work on these this week, and with 
any luck will have them available before mid-month.

In addition to adding cnt_60 sections to the spec files, I'll be 
removing the oldest cruft for distro versions that are no longer 
supported. If anyone really needs to go back to that stuff, the older 
srpms are in the archive. I would have liked to have had that stuff in 
the initial git repos, but I don't think that value is as great as 
getting Cos6 support available sooner. I still might grab the older 
versions when the time comes to populate the github repos.

I'll be leaving the current Suse and Mandriva stuff in there, and will 
also add sections for the latest Fedora releases. There will be no 
testing done on these before they're released though. Only CentOS 5/6 
will be tested and officially supported for the time being.

That's all for now. Comment and questions are welcome.

Thanks.

[qmailtoaster-devel] QTP Bug for CentOS 6

Dan McAllister — 2012-07-26T17:19:01

Greetings Developers:

I had occasion today to try to install spamdyke on a FRESH install of 
CentOS 6 via QTP -- and found that QTP is rejecting my CentOS 6 as 
unsupported.
However, looking into the scripts themselves, they are supposed to 
support CentOS & RHEL 6.x (according to notes within the scripts).

Here's what I found:
On line 356 of the *qtp-whatami* script, version 6 is missing from the 
case statement...
     case $relnum in
     4 | 5 )
should read
   case $relnum in
     4 | 5 | 6 )

NOTE: the check for 6 appears elsewhere, so it appears to have just been 
an oversight...

Once the extra 4 chars were added, spamdyke installed via QTP just fine...

Dan

[qmailtoaster-devel] clamav 0.97.5 changes

Eric Shubert — 2012-07-18T17:27:11

I hope to have clamav-toaster-0.97.5-1.4.1 available by the first of 
next week. I intend to make 2 changes in addition to the upstream 
version update.

First, I hope to redo the patch file so that it works with the tighter 
default fuzz factor that COS6 uses. I prefer doing this rather than 
adding an option to the patch command.

More significantly, I intend to remove the database files from the rpm, 
and modify the spec file such that on a new install it invokes freshclam 
in order to obtain the signature database files. This will substantially 
reduce the size of the rpm, saving a good deal of bandwidth for updates 
(which has bitten us in the past).

If anyone has a problem with this, now's the time to speak up.

Thanks.

[qmailtoaster-devel] Clamav 0.97.5 for testing

Bharath Chari — 2012-06-28T16:24:31

Hi all,

I've uploaded the new clamav 0.97.5 src rpm to my server for TESTING:

http://arachnis.com/qmt/clamav/clamav-toaster-0.97.5-1.4.1.src.rpm

Please use in production only after Eric promotes it to the mirrors.

Eric : No changes made to spec file except for version numbers/packaging 
information. I've included the main.cvd in this version too. As 
discussed, at some point we'll stop doing that.

Bharath

[qmailtoaster-devel] Download script with SHA1 verification

Ron Pacheco — 2012-06-16T14:43:33

Devel,

Attached is my first shot at a download script that:

(1) Makes 10 retry attempts on a file; in the long run probably 
overkill, but right now with some mirror issues, it should be enough 
attempts to resolve any problems.

(2) Verifies a download against an SHA1 checksum. A checksum fail is 
simply considered a download fail and the download will be reattempted 
up to the retry limit. (On a checksum fail, both the srpm and sha1 file 
are removed before the retry, under the premise that either one could 
have been corrupt on download.)

I tested this extensively on my own servers, but I obviously could not 
test against the live servers since there are no SHA1 checksum files on 
the live download mirrors at present. Well, not that I know of. ;)

The script assumes the standard convention of the SHA1 file being name 
file.ext.sha1 for a file named file.ext.

Cheers,

Ron
#!/bin/sh

# To sha1sum or not to sha1sum
if [ "$1" = "-nosha1" ]; then
USESHA1=no
echo "NOTE: Downloads will NOT be verified with sha1 checksums!"
sleep 3
else
USESHA1=yes
fi

# Verify that sha1sum is available
if [ "$USESHA1" = "yes" ] ; then
which sha1sum >/dev/null 2>&1
if [ $? -ne 0 ]; then
echo "Cannot find sha1sum. Installing sha1sum is highly recommended,"
echo "but you may run this script again with the -nosha1 option."
exit 1
fi
fi

# Got wget?
which wget >/dev/null 2>&1
if [ $? -ne 0 ]; then
echo "Please install wget before proceeding."
exit 1
fi

# Subroutine to download file and verify sha1 checksum
getfile() {

# max retries; 10 seemed reasonable
retries=10

# success flag
finished=no

while [ "$finished" = "no" ]; do

# bail if max retries reached
if [ $retries -eq 0 ]; then
echo "Failed downloading: http://$1/$2"
echo "Please check your connection."
exit 1
else
retries=`expr $retries - 1`
fi

# if we don't have the file already, wget it
if [ ! -f $2 ]; then
wget http://$1/$2
fi

# did we wget the file?
if [ -f $2 ]; then

# are we checking the sha1 sum?
if [ "$USESHA1" = "yes" ]; then

# if we don't have the sha1 checksum file, wget it
if [ ! -f $2.sha1 ]; then
wget http://$1/$2.sha1
fi

# if the sha1 file was retrieved, verify the checksum
if [ -f $2.sha1 ]; then
sha1sum -c $2.sha1
if [ $? -eq 0 ]; then
finished=yes
fi
fi

# if we have both the srpm and sha1 files but the checksum failed,
# then blow them both away for another attempt
if [ -f $2.sha1 -a "$finished" = "no" ]; then
rm -f $2 $2.sha1
fi

else
finished=yes
fi
fi
done
}

QT_RPMLIST="http://www.qmailtoaster.com/info/current.txt"
QT_MIRRORS="mirrors.qmailtoaster.com"
QT_RPMS=`wget -q -O - ${QT_RPMLIST}`

# Make sure we have the list
if [ -z "${QT_RPMS}" ] ; then
   echo "Qmail Toaster source RPM list could not be downloaded from"
   echo $QT_RPMLIST
   echo "Please check your connection and try again."
   exit 1
fi

# Download the packages
for SRPM in $QT_RPMS; do
getfile $QT_MIRRORS $SRPM
done

---------------------------------------------------------------------
To unsubscribe, e-mail: qmailtoaster-devel-unsubscribe< at >qmailtoaster.com
For additional commands, e-mail: qmailtoaster-devel-help< at >qmailtoaster.com

[qmailtoaster-devel] Linux Today - Mandriva To Use Mageia For Their Business Servers

Eric Shubert — 2012-05-23T14:58:00

Nigel,

http://www.linuxtoday.com/infrastructure/mandriva-to-use-mageia-for-their-business-servers.html
http://blog.mageia.org/en/2011/06/01/mageia-1/

Perhaps we should be keeping an eye on mageia instead of mandriva?


Johannes,

Does OBS have support for mageia yet? Any plans that you know of?

[qmailtoaster-devel] qmailadmin ezmlm dependency

Eric Shubert — 2012-05-23T02:38:16

I don't use ezmlm, but prefer mailman instead. I happened to attempt to 
remove ezmlm from a QMT host, and yum wanted to remove qmailadmin as 
well because it's dependent on ezmlm.

I'd like to sever this dependency if possible. Would someone care to 
have a look at qmailadmin to see if it's feasible for qmailadmin to 
exist sans ezmlm?

Thanks.

[qmailtoaster-devel] email addresses < at >qmailtoaster.com

Eric Shubert — 2012-05-22T16:04:26

It's been suggested that we set up email addresses for development team 
members. I'd like to get people's thoughts about this.

I think that having email addresses < at >qmailtoaster.com shows a bit of 
solidarity, so from the image standpoint, perhaps it's a good thing. At 
the same time though, this is a community project, and seeing peoples' 
email addresses from widely scattered domains reflects that, which is 
something that I like to see.

I don't know how the rest of you typically do email, but having an 
additional email account doesn't do much for me personally. Most 
communication happens via the lists, and I use gmane.org w/ TBird for 
that. So having another email account < at >qmt is just one more place to 
look for messages (undesirable).

Perhaps we could simply set up forwards < at >qmailtoaster.com instead of 
full accounts. That might achieve the best of both sides.

I don't think that everyone necessarily needs to do the same thing 
either. If you'd like an account or forward name, or nothing at all, 
that's ok. I'm just pointing out that the option is there if we'd like 
to use it.

So let's hear from you. What would you like?

[qmailtoaster-devel] qmail-spp

Eric Shubert — 2012-05-16T16:44:03

I just came across this on the spamdyke list:
http://qmail-spp.sourceforge.net/

It looks very useful, and I'd like to consider getting it into QMT at 
some point. I particularly like the plugin for limiting MAIL FROM rates.

There are 30 plugins in all, many of which provide functionality which 
QMT already has. I'm wondering though if this might provide some 
flexibility that QMT does not presently have.

Would someone care to take a closer look at it, and see what it would take?

Opinions?

Thanks.

[qmailtoaster-devel] DNS servers not responding

Bharath Chari — 2012-04-05T14:19:45

Hi Dan,

I just noticed that dns2.it4soho.com and dns3 aren't serving requests 
for qmailtoaster.com.

Bharath

[qmailtoaster-devel] vpopmail+dovecot lda

Bharath Chari — 2012-04-04T06:17:42

Hi,

In order that dovecot lda + sieve is used instead of vdeliver, the 
.qmail-default file needs to be changed to :
|/var/qmail/bin/preline -f /usr/local/dovecot/libexec/dovecot/deliver -d 
$EXT< at >$USER

However, the current version of vpopmail hard-codes vdeliver into the 
.qmail-default file. I will be patching vpopmail.c to include those 
changes.

This version will be generally available far in the future, but wanted 
to document this right away. A shell script will have to be written to 
convert existing .qmail-default files as part of QTP.

I hope to have a set of packages that will be available for testing in 
the near future.

Eric - Talk about full circle :) - 
http://comments.gmane.org/gmane.mail.vpopmail/20710

Bharath

[qmailtoaster-devel] qmail-dk

Eric Shubert — 2012-04-02T03:44:02

I've had about enough of this module. It has bugs on the incoming and 
outgoing side of things. Every time I upgrade, I need to disable the thing.

Does anyone have a problem if we change the stock qmail-toaster to link 
qmail-queue to qmail-queue.orig instead of qmail-dk? This would 
effectively turn off DK by default, but we can leave it there for the 
time being.

Eventually, I'd like to have DKIM baked in to the stock QMT, but that's 
not a priority yet.

[qmailtoaster-devel] rpm caveat

Eric Shubert — 2012-03-28T15:13:22

In my testing of clamav, I discovered a behavior of rpm that I didn't 
expect, regarding files coded as %config.

Here is a fairly nice write-up about rpm config files:
http://www-uxsup.csx.cam.ac.uk/~jw35/docs/rpm_config.html
It says:
"In summary: if a file is not marked as a config file, or if a file has 
not been altered since installation, then it will be sliently {sic} 
replaced by the version from an update RPM. If a config file has been 
edited on disk, but is not actually different from one RPM to another 
then the edited version will be silently left in place. It is only when 
a config file has been edited and is different from one RPM to the next 
that what happens depens {sic} on the (noreplace) option. If absent, the 
new file will be installed, and the the old edited version will be 
renamed with a .rpmsave suffix. If present, the edited version will be 
left in place, and the new version will be installed with a .rpmnew 
suffix. I don't know what happens if RPM needs to create an .rpmsave or 
.rpmnew file and one already exists - at least in some cases it seems 
that the new file isn't written under these circumstances."

Reading between the lines, and what I didn't expect, is that if a 
%config file is no different between the installed version and the 
version being updated, then nothing happens to this file. I mean 
absolutely nothing. *Even if it has been deleted, it is not added.* This 
happens regardless of whether (noreplace) is specified or not.

What this means is that if a %config file happens to get deleted (or 
severely mangled), doing a --force (implies --replacefiles) update of 
the existing package will not reinstall it. Neither will installing an 
older version if the file is not different between the two. The only way 
to get it back (that I've found) is to remove the package and 
(re)install it.

Personally, I think this sux. I would think that --replacefiles would at 
least add it back if it was missing, but it doesn't. This is probably 
something that seasoned RPM users are familiar with, but it was news to 
me. Nice thing to know.

FWIW, you can find out which files are coded %config by doing
# rpm -qlc <package>

Just thought I'd share this, as it drove me nuts for a day or so.

[qmailtoaster-devel] Dovecot LDA

Bharath Chari — 2012-03-28T07:05:33

Hi,

I've been experimenting with replacing maildrop with Dovecot-LDA. I am 
using dovecot for both IMAP and POP3, and have removed courier-imap, 
courier-authlib, maildrop, and pop3-toaster completely from my dev 
server, and it seems to be chugging along without any problems.

Anyone else care to share experiences - good or bad?

Bharath

[qmailtoaster-devel] RPM-ifying Configurations slideshow

Eric Shubert — 2012-03-27T03:29:36

For anyone who'll be working on spec files, you'll want to look through 
this. It has some nice solutions that can be used with QMT.

http://www.redhat.com/promo/summit/2010/presentations/summit/opensource-for-it-leaders/thurs/pwaterma-2-rpm/RPM-ifying-System-Configurations.pdf

[qmailtoaster-devel] Dovecot with vpopmail 5.4.33

Bharath Chari — 2012-03-25T16:44:43

Hi,

Has anyone tried my test version of vpopmail 5.4.33? I'm testing with 
dovecot from qtp now, and am unable to login (authentication failed). I 
just wanted to see if anyone else has the same issue - I'll start 
debugging after some feedback.

Bharath

[qmailtoaster-devel] vpopmail mysql limits

Bharath Chari — 2012-03-22T15:48:03

Hi,

I was wondering if there is any merit in moving the limits to mysql from 
the .qmailadmin-limits file in the 1.5.0 release by enabling the 
--enable-mysql-limits flag. Of course existing limits would need to be 
migrated to mysql via a script.

Comments/Thoughts?

Bharath