Sendmail error (exited with code 75) - RHEL6

Hello list,

as debian users, I’m encountering the “sendmail” error, but with RHEL6

Here are the log from “/var/log/httpd/error_log” :
[Tue Sep 4 09:34:43 2012] [info]: <
rt-4.0.7-1945-1346751283-749.7324-10-0@myfirm.com> #7324/120687 - Scrip 10
On Resolve Notify Requestors
(/opt/rt4/sbin/…/lib/RT/Action/SendEmail.pm:301)
sendmail: fatal: edos@myfirm.com(48): No recipient addresses found in
message header
[Tue Sep 4 09:34:43 2012] [error]: <
rt-4.0.7-1945-1346751283-749.7324-10-0@myfirm.com>: /usr/sbin/sendmail -oi -t -f edos@myfirm.com exited with code 75
(/opt/rt4/sbin/…/lib/RT/Interface/Email.pm:465)
[Tue Sep 4 09:34:43 2012] [crit]: <
rt-4.0.7-1945-1346751283-749.7324-10-0@myfirm.com>: Could not send mail
with command /usr/sbin/sendmail -oi -t -f edos@myfirm.com: <
rt-4.0.7-1945-1346751283-749.7324-10-0@myfirm.com>: /usr/sbin/sendmail -oi -t -f edos@myfirm.com exited with code 75 at
/opt/rt4/sbin/…/lib/RT/Interface/Email.pm line 466.

Stack:
[/opt/rt4/sbin/…/lib/RT/Interface/Email.pm:466]
[/opt/rt4/sbin/…/lib/RT/Action/SendEmail.pm:308]
[/opt/rt4/sbin/…/lib/RT/Action/SendEmail.pm:128]
[/opt/rt4/sbin/…/lib/RT/ScripAction.pm:232]
[/opt/rt4/sbin/…/lib/RT/Scrip.pm:475]
[/opt/rt4/sbin/…/lib/RT/Scrips.pm:188]
[/opt/rt4/sbin/…/lib/RT/Transaction.pm:201]
[/opt/rt4/sbin/…/lib/RT/Record.pm:1461]
[/opt/rt4/sbin/…/lib/RT/Ticket.pm:3487]
[/opt/rt4/sbin/…/lib/RT/Ticket.pm:3188]
[/opt/rt4/sbin/…/lib/RT/Record.pm:932]
[/opt/rt4/sbin/…/lib/RT/Record.pm:911]
[/opt/rt4/sbin/…/lib/RT/Interface/Web.pm:2256]
[/opt/rt4/sbin/…/lib/RT/Interface/Web.pm:2368]
[/opt/rt4/share/html/Ticket/Modify.html:81]
[/opt/rt4/share/html/Ticket/autohandler:19]
[/opt/rt4/sbin/…/lib/RT/Interface/Web.pm:576]
[/opt/rt4/sbin/…/lib/RT/Interface/Web.pm:326]
[/opt/rt4/share/html/autohandler:53]
(/opt/rt4/sbin/…/lib/RT/Interface/Email.pm:470)

I have to restart apache each time I have this error, in other words, 4-5
time yesterday.

This problem appeared yesterday, the last package update was done on July
20th, and the last RT update, 4.0.7, was done 6 days ago.

Here are my system information :
[root@dgilx202 ~]# uname -a
Linux dgilx202.res.myfirm.org 2.6.32-220.7.1.el6.x86_64 #1 SMP Fri Feb 10
15:22:22 EST 2012 x86_64 x86_64 x86_64 GNU/Linux
[root@dgilx202 ~]# cat /etc/redhat-release
Red Hat Enterprise Linux Server release 6.2 (Santiago)
[root@dgilx202 ~]# cat /etc/httpd/conf.d/rt.conf
<VirtualHost :80>
servername rt.myfirm.com
RewriteEngine On
RewriteCond %{HTTPS} off
RewriteRule (.
) https://%{HTTP_HOST}%{REQUEST_URI}

<VirtualHost *:443>
### Optional apache logs for RT
# ErrorLog /opt/rt4/var/log/apache2.error
# TransferLog /opt/rt4/var/log/apache2.access
# LogLevel debug
servername rt.myfirm.com
AddDefaultCharset UTF-8

    SSLEngine on
    SSLCertificateFile /etc/ssl/certs/rt.myfirm.com.cer
    SSLCertificateKeyFile /etc/ssl/keys/rt.myfirm.com.key

    DocumentRoot "/opt/rt4/share/html"
    <Location />
            Order allow,deny
            Allow from all

            SetHandler modperl
            PerlResponseHandler Plack::Handler::Apache2
            PerlSetVar psgi_app /opt/rt4/sbin/rt-server
    </Location>
    <Perl>
            use Plack::Handler::Apache2;
            Plack::Handler::Apache2->preload("/opt/rt4/sbin/rt-server");
    </Perl>
[root@dgilx202 ~]# rpm -qa | grep mod_perl mod_perl-2.0.4-10.el6.x86_64 [root@dgilx202 ~]#

Email::Address version :
[root@dgilx202 ~]# cpan -D Email::Address
CPAN: Storable loaded ok (v2.20)
Reading '/root/.cpan/Metadata’
Database was generated on Tue, 04 Sep 2012 10:39:03 GMT
Email::Address
(no description)
R/RJ/RJBS/Email-Address-1.896.tar.gz
/usr/local/share/perl5/Email/Address.pm
Installed: 1.895
CPAN: 1.896 Not up to date
Ricardo SIGNES (RJBS)
rjbs@cpan.org

[root@dgilx202 ~]#

The only changes in RT I can notice are :

  • more users ;
  • news scrips and template.

FI, I restarted apache 3 times during the writing of this email…

Do you have any idea why this is happening ?
Raphaël Berlamont

[Tue Sep 4 09:34:43 2012] [info]:
rt-4.0.7-1945-1346751283-749.7324-10-0@myfirm.com #7324/120687 - Scrip 10
On Resolve Notify Requestors
(/opt/rt4/sbin/…/lib/RT/Action/SendEmail.pm:301)
sendmail: fatal: edos@myfirm.com(48): No recipient addresses found in
message header

This isn’t an RT error, it’s sendmail telling you that it thinks it’s
the local handler for myfirm.com, but can’t

http://www.linkedin.com/in/paultomblin
http://careers.stackoverflow.com/ptomblin

[Tue Sep 4 09:34:43 2012] [info]:
rt-4.0.7-1945-1346751283-749.7324-10-0@myfirm.com #7324/120687 -
Scrip 10
On Resolve Notify Requestors
(/opt/rt4/sbin/…/lib/RT/Action/SendEmail.pm:301)
sendmail: fatal: edos@myfirm.com(48): No recipient addresses found in
message header

This isn’t an RT error, it’s sendmail telling you that it thinks it’s
the local handler for myfirm.com, but can’t

Hello Paul, thank you for your answer.

OK for this first error, but the second one disappears as soon as I restart
apache, and RT work flawlessly for minutes/hours (depending on load maybe) :
"[Tue Sep 4 09:34:43 2012] [error]: <
rt-4.0.7-1945-1346751283-749.7324-10-0@myfirm.com>: /usr/sbin/sendmail -oi -t -f edos@myfirm.com exited with code 75
(/opt/rt4/sbin/…/lib/RT/Interface/Email.pm:465)
[Tue Sep 4 09:34:43 2012] [crit]: <
rt-4.0.7-1945-1346751283-749.7324-10-0@myfirm.com>: Could not send mail
with command /usr/sbin/sendmail -oi -t -f edos@myfirm.com: <
rt-4.0.7-1945-1346751283-749.7324-10-0@myfirm.com>: /usr/sbin/sendmail -oi -t -f edos@myfirm.com exited with code 75 at
/opt/rt4/sbin/…/lib/RT/Interface/Email.pm line 466.

I repeat : a simple apache restart makes all the message working again.
Users resend their answers, without modifying any information (IE : they
are not adding any recipient), and it works.

Raphaël Berlamont

OK for this first error, but the second one disappears as soon as I restart
apache, and RT work flawlessly for minutes/hours (depending on load maybe) :

"[Tue Sep 4 09:34:43 2012] [error]:
rt-4.0.7-1945-1346751283-749.7324-10-0@myfirm.com: /usr/sbin/sendmail -oi -t -f edos@myfirm.com exited with code 75

Error code 75 is “EX_TEMPFAIL”, which is sendmail telling you that a
temporary error occurred. The message is queued, but it can’t be
delivered immediately, but it will retry when it can.

http://www.diablotin.com/librairie/networking/sendmail/ch36_05.htm

I can’t think of any reason why sendmail thinks it needs to tell the
mail user agent (MUA) this. Mail Transfer Agents (MTA) like sendmail
should do queuing and retrying automatically. If anybody needs to be
told that it’s temporary failing, it’s the sysadmin, not the mail
sender.

http://www.linkedin.com/in/paultomblin
http://careers.stackoverflow.com/ptomblin

I can’t think of any reason why sendmail thinks it needs to tell the
mail user agent (MUA) this. Mail Transfer Agents (MTA) like sendmail
should do queuing and retrying automatically. If anybody needs to be
told that it’s temporary failing, it’s the sysadmin, not the mail
sender

I agree with your conclusion, but that doesn’t explain why a restart of
apache make the thing working again… A limit somewhere ? Are there any
tunning to do for heavy load installation ? (even if I don’t really think
that our system is under real heavy load).
Raphaël Berlamont

This problem appeared yesterday, the last package update was done on
July 20th, and the last RT update, 4.0.7, was done 6 days ago.

What version of RT did you upgrade from?

If you go to the system configuration page, what’s the value of MailCommand?

The only changes in RT I can notice are :

  • more users ;
  • news scrips and template.

Do you mean new templates provided by the RT upgrade, or new templates
you’ve made? If the latter, please send the templates along.

This problem appeared yesterday, the last package update was done on
July 20th, and the last RT update, 4.0.7, was done 6 days ago.

What version of RT did you upgrade from?

4.0.6 (was 4.0.5 a week earlier)

If you go to the system configuration page, what’s the value of
MailCommand?

MailCommand ‘sendmailpipe’

The only changes in RT I can notice are :

  • more users ;
  • news scrips and template.

Do you mean new templates provided by the RT upgrade, or new templates
you’ve made? If the latter, please send the templates along.

New template i’ve set up, very basic though (sorry, French here ;)) :
Bonjour,

Votre demande n°{$Ticket->id} ({ $Ticket->SubjectTag }) vient d’être prise
en charge par { $Ticket->OwnerObj->Name }.

Vous pouvez suivre l’avancement de votre demande en vous connectant sur
{$RT::WebURL}Ticket/Display.html?id={$Ticket->id}

Il n’est pas nécessaire de répondre à ce message.
L’équipe de test
rt-test@myfirm.fr

Raphaël Berlamont

I can’t think of any reason why sendmail thinks it needs to tell the
mail user agent (MUA) this. Mail Transfer Agents (MTA) like sendmail
should do queuing and retrying automatically. If anybody needs to be
told that it’s temporary failing, it’s the sysadmin, not the mail
sender

I agree with your conclusion, but that doesn’t explain why a restart of
apache make the thing working again… A limit somewhere ? Are there any
tunning to do for heavy load installation ? (even if I don’t really think
that our system is under real heavy load).

What’s your load average? I seem to recall that by default sendmail
would stop delivery if the load average is above 4.0.

You can probably find more information in the mail logs.

http://www.linkedin.com/in/paultomblin
http://careers.stackoverflow.com/ptomblin

I agree with your conclusion, but that doesn’t explain why a restart of
apache make the thing working again… A limit somewhere ? Are there any
tunning to do for heavy load installation ? (even if I don’t really think
that our system is under real heavy load).

What’s your load average? I seem to recall that by default sendmail
would stop delivery if the load average is above 4.0.

We’re far from this :

[root@dgilx202 ~]# uptime
18:56:25 up 22:57, 1 user, load average: 0.08, 0.05, 0.04
[root@dgilx202 ~]#

You can probably find more information in the mail logs.

Sadly, logs don’t say anything more. When it start to stop working (until I
restart Apache), I have this king message every time RT tries to send a
mail :
Sep 3 13:04:28 dgilx202 postfix/sendmail[16133]: fatal:
digital-service-desk@myfirm.com(48): No recipient addresses found in
message header
Sep 3 13:04:28 dgilx202 postfix/sendmail[16135]: fatal:
digital-service-desk@myfirm.com(48): No recipient addresses found in
message header

Here you can fin a pastebin of ll those messages :
[root@dgilx202 ~]# grep "No recipient addresses found in message header" /var/lo - Pastebin.com .

This kind of message appeared for the first time yesterday @ 12:54:39 as
mentioned in the above pastebin, I’ve checked in older logs, no occurrence
appeared before.

Raphaël Berlamont

Here you can fin a pastebin of ll those messages :
[root@dgilx202 ~]# grep "No recipient addresses found in message header" /var/lo - Pastebin.com

.

What happens if you just try to send an email to one of those
addresses from the command line on that server? I.e.
Mail -s “test message” corp_admin_dba@myfirm.fr
test me
.

http://www.linkedin.com/in/paultomblin
http://careers.stackoverflow.com/ptomblin

Unrelated notes about your template below.

Hmm. Can you wrap /usr/sbin/sendmail with a tiny shell script that
tee’s the input somewhere else and then passes it through to the real
sendmail?

mv -v /usr/sbin/sendmail{,.real}
cat /usr/sbin/sendmail
#!/bin/bash
exec tee -a /tmp/sendmail-stdin | /usr/sbin/sendmail.real “$@”
^D
chmod a+rx /usr/sbin/sendmailOn 09/04/2012 09:30 AM, Raphaël Berlamont wrote:

New template i’ve set up, very basic though (sorry, French here ;)) :

Subject: {$Ticket->Subject} : pris en charge par { $Ticket->OwnerObj->Name }
Bonjour,

The lack of a blank line between your header block (containing Subject:
…) and the body of your message “Bonjour,” is going to cause problems.
“Bonjour,” is going to be parsed as a header. You should put a blank
line between the two lines above.

What happens if you just try to send an email to one of those
addresses from the command line on that server? I.e.
Mail -s “test message” corp_admin_dba@myfirm.fr
test me
.

Hello Paul,

It works and open me a new ticket (no reply though, because root is not
correctly aliased).
Raphaël Berlamont

What happens if you just try to send an email to one of those
addresses from the command line on that server? I.e.
Mail -s “test message” corp_admin_dba@myfirm.fr
test me
.

Hello Paul,

It works and open me a new ticket (no reply though, because root is not
correctly aliased).

Wait a minute - are you saying that RT is sending outgoing mail to
addresses that RT will receive and create a ticket for? Won’t that
lead to a cascade of tickets?

http://www.linkedin.com/in/paultomblin
http://careers.stackoverflow.com/ptomblin

Wait a minute - are you saying that RT is sending outgoing mail to
addresses that RT will receive and create a ticket for? Won’t that
lead to a cascade of tickets?

Nope. Locally, addresses are like this : xxx@rt.myfirm.fr, translation is
done by compagny’s MTA.
So the mail you told me to send went out of the server, then came back,
translated.
Raphaël Berlamont

Unrelated notes about your template below.

Hmm. Can you wrap /usr/sbin/sendmail with a tiny shell script that
tee’s the input somewhere else and then passes it through to the real
sendmail?

mv -v /usr/sbin/sendmail{,.real}
cat /usr/sbin/sendmail
#!/bin/bash
exec tee -a /tmp/sendmail-stdin | /usr/sbin/sendmail.real “$@”
^D
chmod a+rx /usr/sbin/sendmail

Well! Very, very strange behaviour that I can’t explain : the
/tmp/sendmail-stdin filled by tee is fulfilled all the time, but when I
encounter the bug, nothing appear in the file!

For example, at the very same moment, in /var/log/maillog, I have the
following :
Sep 5 19:11:10 dgilx202 postfix/smtpd[7412]: connect from
unknown[10.223.2.22]
Sep 5 19:11:10 dgilx202 postfix/smtpd[7412]: 0F18F25E10:
client=unknown[10.223.2.22]
Sep 5 19:11:10 dgilx202 postfix/cleanup[7416]: 0F18F25E10: message-id=<
4D44666F4F0B2248A2C57DD0AF573F930ECD1D21@PLA122VS151.fr.myfirm.org>
Sep 5 19:11:10 dgilx202 postfix/qmgr[1793]: 0F18F25E10: from=<
raphael.berlamont@myfirm.fr>, size=1803, nrcpt=1 (queue active)
Sep 5 19:11:10 dgilx202 postfix/smtpd[7412]: disconnect from
unknown[10.223.2.22]
Sep 5 19:11:10 dgilx202 postfix/sendmail[7441]: fatal: rt-test@myfirm.fr(48):
No recipient addresses found in message header
Sep 5 19:11:10 dgilx202 postfix/local[7417]: 0F18F25E10: to=<
RT-Test@rt.myfirm.com>, relay=local, delay=0.6, delays=0.02/0/0/0.58,
dsn=2.0.0, status=sent (delivered to command: /opt/rt4/bin/rt-mailgate
–queue ‘Test Queue’ --action correspond --url https://rt.myfirm.com)
Sep 5 19:11:10 dgilx202 postfix/qmgr[1793]: 0F18F25E10: removed

But nothing appeared in “/tmp/sendmail-stdin”…

I’ve done the test 3 times, same result : logs appears in
“/var/log/maillog”, and nothing in “/tmp/sendmail-stdin”. I don’t even
understand how can this be possible…> On 09/04/2012 09:30 AM, Raphaël Berlamont wrote:

New template i’ve set up, very basic though (sorry, French here ;)) :

Subject: {$Ticket->Subject} : pris en charge par {
$Ticket->OwnerObj->Name }
Bonjour,

The lack of a blank line between your header block (containing Subject:
…) and the body of your message “Bonjour,” is going to cause problems.
“Bonjour,” is going to be parsed as a header. You should put a blank
line between the two lines above.

Thank you, I corrected this.

Raphaël Berlamont

Unrelated notes about your template below.

Hmm. Can you wrap /usr/sbin/sendmail with a tiny shell script that
tee’s the input somewhere else and then passes it through to the real
sendmail?

mv -v /usr/sbin/sendmail{,.real}
cat /usr/sbin/sendmail
#!/bin/bash
exec tee -a /tmp/sendmail-stdin | /usr/sbin/sendmail.real “$@”
^D
chmod a+rx /usr/sbin/sendmail

I’ve done the test 3 times, same result : logs appears in
“/var/log/maillog”, and nothing in “/tmp/sendmail-stdin”. I don’t even
understand how can this be possible…

I confirm this behaviour. I slightly modified the script :
[root@dgilx202 ~]# cat /usr/sbin/sendmail
#!/bin/bash

TMP_FILE=“/tmp/sendmail-stdin”
echo “=====NEW SENDMAIL CALL=======” >> “${TMP_FILE}”
echo “date +%Y%m%d-%H%M%S” >> “${TMP_FILE}”
echo “========MESSAGE BEGIN========” >> “${TMP_FILE}”
exec tee -a “${TMP_FILE}” | /usr/sbin/sendmail.real “$@”
echo “========MESSAGE ENDED========” >> “${TMP_FILE}”
echo “date +%Y%m%d-%H%M%S” >> “${TMP_FILE}”
echo “======END SENDMAIL CALL======” >> “${TMP_FILE}”
[root@dgilx202 ~]#

And with a test, here is what I have in the log fie :
[root@dgilx202 ~]# tail -f /tmp/sendmail-stdin
=====NEW SENDMAIL CALL=======
20120906-154906
========MESSAGE BEGIN========
========MESSAGE ENDED========
20120906-154906
======END SENDMAIL CALL======
^C
[root@dgilx202 ~]#

And in the maillog, this what I have :
Sep 6 15:49:06 dgilx202 postfix/smtpd[19149]: connect from
unknown[10.223.2.22]
Sep 6 15:49:06 dgilx202 postfix/smtpd[19149]: 3B22B23538:
client=unknown[10.223.2.22]
Sep 6 15:49:06 dgilx202 postfix/cleanup[19154]: 3B22B23538: message-id=<
4D44666F4F0B2248A2C57DD0AF573F930ECD205E@PLA122VS151.fr.myfirm.org>
Sep 6 15:49:06 dgilx202 postfix/qmgr[1793]: 3B22B23538: from=<
raphael.berlamont@myfirm.fr>, size=1803, nrcpt=1 (queue active)
Sep 6 15:49:06 dgilx202 postfix/smtpd[19149]: disconnect from
unknown[10.223.2.22]
Sep 6 15:49:06 dgilx202 postfix/sendmail[19225]: fatal: rt-test@myfirm.fr(48):
No recipient addresses found in message header
Sep 6 15:49:06 dgilx202 postfix/local[19155]: 3B22B23538: to=<
RT-Test@rt.myfirm.com>, relay=local, delay=0.6, delays=0.02/0/0/0.58,
dsn=2.0.0, status=sent (delivered to command: /opt/rt4/bin/rt-mailgate
–queue ‘Test Queue’ --action correspond --url https://rt.myfirm.com)
Sep 6 15:49:06 dgilx202 postfix/qmgr[1793]: 3B22B23538: removed

So I can now tell that sendmail IS called, but nothing is piped in, or it
crashs before RT can pipe anything in it…

Idea ?

Raphaël Berlamont

So I can now tell that sendmail IS called, but nothing is piped in, or
it crashs before RT can pipe anything in it…

This is indeed strange. Can you show us the output of apachectl -V?

apachectl may be spelled apache2ctl or httpdctl on RHEL, not sure off
the top of my head.

Can you show us the output of apachectl -V?

Hello Thomas,

Here is what you asked :
[root@dgilx202 ~]# apachectl -V
Server version: Apache/2.2.15 (Unix)
Server built: Feb 7 2012 09:50:11
Server’s Module Magic Number: 20051115:24
Server loaded: APR 1.3.9, APR-Util 1.3.9
Compiled using: APR 1.3.9, APR-Util 1.3.9
Architecture: 64-bit
Server MPM: Prefork
threaded: no
forked: yes (variable process count)
Server compiled with…
-D APACHE_MPM_DIR=“server/mpm/prefork”
-D APR_HAS_SENDFILE
-D APR_HAS_MMAP
-D APR_HAVE_IPV6 (IPv4-mapped addresses enabled)
-D APR_USE_SYSVSEM_SERIALIZE
-D APR_USE_PTHREAD_SERIALIZE
-D SINGLE_LISTEN_UNSERIALIZED_ACCEPT
-D APR_HAS_OTHER_CHILD
-D AP_HAVE_RELIABLE_PIPED_LOGS
-D DYNAMIC_MODULE_LIMIT=128
-D HTTPD_ROOT="/etc/httpd"
-D SUEXEC_BIN="/usr/sbin/suexec"
-D DEFAULT_PIDLOG=“run/httpd.pid”
-D DEFAULT_SCOREBOARD=“logs/apache_runtime_status”
-D DEFAULT_LOCKFILE=“logs/accept.lock”
-D DEFAULT_ERRORLOG=“logs/error_log”
-D AP_TYPES_CONFIG_FILE=“conf/mime.types”
-D SERVER_CONFIG_FILE=“conf/httpd.conf”
[root@dgilx202 ~]#

Regards,
Raphaël Berlamont

Here is what you asked :

[root@dgilx202 ~]# apachectl -V
Server version: Apache/2.2.15 (Unix)
Server built: Feb 7 2012 09:50:11
Server’s Module Magic Number: 20051115:24
Server loaded: APR 1.3.9, APR-Util 1.3.9
Compiled using: APR 1.3.9, APR-Util 1.3.9
Architecture: 64-bit
Server MPM: Prefork
threaded: no
forked: yes (variable process count)
Server compiled with…
-D APACHE_MPM_DIR=“server/mpm/prefork”
-D APR_HAS_SENDFILE
-D APR_HAS_MMAP
-D APR_HAVE_IPV6 (IPv4-mapped addresses enabled)
-D APR_USE_SYSVSEM_SERIALIZE
-D APR_USE_PTHREAD_SERIALIZE
-D SINGLE_LISTEN_UNSERIALIZED_ACCEPT
-D APR_HAS_OTHER_CHILD
-D AP_HAVE_RELIABLE_PIPED_LOGS
-D DYNAMIC_MODULE_LIMIT=128
-D HTTPD_ROOT="/etc/httpd"
-D SUEXEC_BIN="/usr/sbin/suexec"
-D DEFAULT_PIDLOG=“run/httpd.pid”
-D DEFAULT_SCOREBOARD=“logs/apache_runtime_status”
-D DEFAULT_LOCKFILE=“logs/accept.lock”
-D DEFAULT_ERRORLOG=“logs/error_log”
-D AP_TYPES_CONFIG_FILE=“conf/mime.types”
-D SERVER_CONFIG_FILE=“conf/httpd.conf”
[root@dgilx202 ~]#

Issue still running, and I don’t have an idea where to look at…

To refresh your memory : sometime, RT can’t send any mail piped into
sendmail binary. To make it working again, I have to restart apache. It
work for a few minutes, then stop working again.

To make this bad behavior almost unnoticeable by the user, I’ve set a cron
that restart apache every five minutes, but of course, this is not a final
solution…

For now, I changed the “MailCommand” to “smtp”, and it works. I hope there
is no limitation using this method instead of “sendmailpipe”.

Good day!
Raphaël Berlamont

Issue still running, and I don’t have an idea where to look at…

To refresh your memory : sometime, RT can’t send any mail piped into
sendmail binary. To make it working again, I have to restart apache. It
work for a few minutes, then stop working again.

This is very strange, and it’s especially strange that you’re seeing
this under the prefork MPM. I expected you used a threaded MPM and were
running into weird threading bugs.

To make this bad behavior almost unnoticeable by the user, I’ve set a
cron that restart apache every five minutes, but of course, this is not
a final solution…

For now, I changed the “MailCommand” to “smtp”, and it works. I hope
there is no limitation using this method instead of “sendmailpipe”.

Setting MailCommand to ‘smtp’ is dangerous, and will likely be removed
in the next major version of RT (4.2). It can result in dropped mail if
the SMTP server isn’t responding since RT isn’t queueing mail like an
MTA does.