Email randomly 0 bytes on internal messages

Hello.

I use Fedora 4/5 with fresh instalation and latest VirtualminPro.

Instalation with Postfix + Dovecot

Test:

1.- use mail -s "Here subject" virtual.user@domain< any.txt
virtual.user get a messge with 0 bytes.
2. Same probe how root
User get message correct.
3.- Same probe how root
virtual.user get a messge with 0 bytes.

We have a lot of probes, send from external accounts to this server, and not fail.

But in send from machine, (shell with mail, php script) users get a message 0 bytes.

We are lost.

Hey Marco,

Any clues in the /var/log/maillog?

Any. Meesages normal.

Just example:

Meesages send and show this on account:

21 None 0 kB
20 root 24/05/2006 08:22 2 kB aca va el subject 20:43
19 None 0 kB

Two last send form shell. With 3 second of diference.

19 send by oscommerce order.

maillog
May 24 08:22:25 ns57 postfix/smtpd[[16767]]: connect from ns57.hostinglmi.net[[127.0.0.1]]
May 24 08:22:25 ns57 postfix/smtpd[[16767]]: disconnect from ns57.hostinglmi.net[[127.0.0.1]]
May 24 08:22:28 ns57 postfix/pickup[[24211]]: 5BA85838140: uid=0 from=<root>
May 24 08:22:28 ns57 postfix/cleanup[[24280]]: 5BA85838140: message-id=<20060524062228.5BA85838140@ns57.hostinglmi.net>
May 24 08:22:28 ns57 postfix/qmgr[[5249]]: 5BA85838140: from=<root@ns57.hostinglmi.net>, size=1738, nrcpt=1 (queue active )
May 24 08:22:32 ns57 postfix/local[[24282]]: 5BA85838140: to=<pedidos.dibenede@ns57.hostinglmi.net>, orig_to=<pedidos@regalosdibenedetto.com>, relay=local, delay=4, status=sent (delivered to command: /usr/bin/procmail-wrapper -o -a $DOMAIN -d $LOGNAME)
May 24 08:22:32 ns57 postfix/qmgr[[5249]]: 5BA85838140: removed
May 24 08:22:34 ns57 imap-login: Aborted login [[::ffff:127.0.0.1]]
May 24 08:22:37 ns57 pop3-login: Disconnected [[::ffff:127.0.0.1]]
May 24 08:23:28 ns57 postfix/smtpd[[16767]]: connect from ns57.hostinglmi.net[[127.0.0.1]]
May 24 08:23:28 ns57 postfix/smtpd[[16767]]: disconnect from ns57.hostinglmi.net[[127.0.0.1]]
May 24 08:23:35 ns57 imap-login: Aborted login [[::ffff:127.0.0.1]]
May 24 08:23:39 ns57 pop3-login: Disconnected [[::ffff:127.0.0.1]]
May 24 08:23:53 ns57 postfix/anvil[[17104]]: statistics: max connection rate 1/60s for (smtp:202.152.232.200) at May 24 08:13:53
May 24 08:23:53 ns57 postfix/anvil[[17104]]: statistics: max connection count 1 for (smtp:202.152.232.200) at May 24 08:13:53
May 24 08:23:53 ns57 postfix/anvil[[17104]]: statistics: max cache size 1 at May 24 08:13:53
May 24 08:24:25 ns57 postfix/pickup[[24211]]: 75958838140: uid=0 from=<root>
May 24 08:24:25 ns57 postfix/cleanup[[24406]]: 75958838140: message-id=<20060524062425.75958838140@ns57.hostinglmi.net>
May 24 08:24:25 ns57 postfix/qmgr[[5249]]: 75958838140: from=<root@ns57.hostinglmi.net>, size=1738, nrcpt=1 (queue active)
May 24 08:24:29 ns57 postfix/local[[24407]]: 75958838140: to=<pedidos.dibenede@ns57.hostinglmi.net>, orig_to=<pedidos@regalosdibenedetto.com>, relay=local, delay=4, status=sent (delivered to command: /usr/bin/procmail-wrapper -o -a $DOMAIN -d $LOGNAME)
May 24 08:24:29 ns57 postfix/qmgr[[5249]]: 75958838140: removed

I’ve make a several probes.

Just install a new virtual server, but not Migrate. Fresh virtual serer.

Put a same template. (Ulimited)

Work fine.

Also inspect postfix configuration and virtual and virtual.db hash.

Any suspect for me.

Hey Marco,

This sounds like it might be a similar issue to one we had in the past with procmail losing messages that had to have attachments decompressed for virus scanning…the quota would prevent it from uncompressing and it was then dropped by clamav. But, obviously, a plaintext message isn’t going to be subject to that particular problem…maybe something similar though.

So, you’re saying in the last message that a fresh install does not exhibit the problem while an older install does? That could be due to procmail rules being different. Take a look at the system-wide /etc/procmailrc and the domain-specific /etc/webmin/virtual-server/spam/SERVERID/virtualmin.cf (where SERVERID is the numeric ID of the domain–you can find this out by grepping for the domain name in the domains subdirectory of /etc/webmin/virtual-server). Is there anything different about the ones on the working system and the older system?

/etc/procmailrc
>>
:0wi
VIRTUALMIN=|/etc/webmin/virtual-server/lookup-domain.pl $LOGNAME
:0

  • ?/usr/bin/test "$VIRTUALMIN" != ""
    {
    INCLUDERC=/etc/webmin/virtual-server/procmail/$VIRTUALMIN
    }
    ORGMAIL=$HOME/Maildir/
    DEFAULT=$HOME/Maildir/
    DROPPRIVS=yes
    :0
    $DEFAULT

/etc/webmin/virtual-server/spam/SERVERID/virtualmin.cf

For two accoutns it’s the same: 0 bytes a user and group it’s account.
18647210 0 -rwxr-xr-x 1 specka specka 0 may 24 08:39 virtualmin.cf (Good account)
18647270 0 -rwxr-xr-x 1 dibenede dibenede 0 may 20 06:48 virtualmin.cf (Problem account)