Mailgate errors RT 3

I have RT 3.0.6 running on a Debian system. Everything was working fine but I made some changes to the RT.SiteConfig and installed apache-ssl to use a virtual host with HTTPS. I’m not exactly sure at what point the mailgate stopped working but I can create, correspond, reply, etc from the web interface but I cannot use the rt-mailgate command any longer. Since the --url option does not support HTTPS I followed the advice of another posting to configure apache to respond to localhost. I keep recieving these errors:

(Deferred: prog mailer (/usr/lib/sm.bin/smrsh) exited with EX)
"|/usr/bin/rt-mailgate --debug --queue --url http://localhost"

Part of the error message recieved from deferred message:

Can't use string ("SessionType") as a HASH ref while "strict refs" in use at /usr/share/perl5/RT/Interface/Email.pm line 406

There is quite a bit more in the deferred email notification but this looks like the start of the issue.  Has anyone seen this problem before.

Thanks in advance for any help.

Frank Balonis

Here is an updated output of my error message:

error: Can’t use string (“SessionType”) as a HASH ref while “strict refs” in use at /usr/share/perl5/RT/Interface/Email.pm line 406.
context:

402:
403: sub Gateway {
404: my $argsref = shift;
405:
406: my %args = %$argsref;
407:
408: # Set some reasonable defaults
409: $args{‘action’} = ‘correspond’ unless ( $args{‘action’} );
410: $args{‘queue’} = ‘1’ unless ( $args{‘queue’} );

code stack: /usr/share/perl5/RT/Interface/Email.pm:406
/usr/share/request-tracker3/html/REST/1.0/NoAuth/mail-gateway:32

Thanks for any help.

Frank Balonis

I have RT 3.0.6 running on a Debian system. Everything was working fine but I made some changes to the RT.SiteConfig and installed apache-ssl to use a virtual host with HTTPS. I’m not exactly sure at what point the mailgate stopped working but I can create, correspond, reply, etc from the web interface but I cannot use the rt-mailgate command any longer. Since the --url option does not support HTTPS I followed the advice of another posting to configure apache to respond to localhost. I keep recieving these errors:

(Deferred: prog mailer (/usr/lib/sm.bin/smrsh) exited with EX)
"|/usr/bin/rt-mailgate --debug --queue --url http://localhost"

I think the issue here is that you have lost the name of the queue
that should come after the --queue option, e.g. General.

As an aside, I think rt-mailgate will work fine with an https server
if you install the libio-socket-ssl-perl package so that libwww-perl
(which provides LWP::UserAgent) can work via SSL.

Stephen Quinney

I think the issue here is that you have lost the name of the queue
that should come after the --queue option, e.g. General.

I verified that I’m have General in the commandline. What I didn’t recognize yesterday is that I was recieving a full HTML encoded error. The error looked like this when displayed in a browser:

error: Can’t use string (“SessionType”) as a HASH ref while “strict refs” in use at /usr/share/perl5/RT/Interface/Email.pm line 406.
context:

402:
403: sub Gateway {
404: my $argsref = shift;
405:
406: my %args = %$argsref;
407:
408: # Set some reasonable defaults
409: $args{‘action’} = ‘correspond’ unless ( $args{‘action’} );
410: $args{‘queue’} = ‘1’ unless ( $args{‘queue’} );

code stack: /usr/share/perl5/RT/Interface/Email.pm:406
/usr/share/request-tracker3/html/REST/1.0/NoAuth/mail-gateway:32
raw error

As an aside, I think rt-mailgate will work fine with an https server
if you install the libio-socket-ssl-perl package so that libwww-perl
(which provides LWP::UserAgent) can work via SSL.

You were right I installed libio-socket-ssl-perl and I no longer get the HTTPS not support error now I’m back to usering HTTPS but recieving the error above.

Thanks,

Frank Balonis

I also have found that if I remove “use strict;” in the Email.pm file that the email goes through and the ticket is created but there is no information in the ticket at all. And no matter who the requestor is it is created by rt_system.

Frank

The best hardware available to me is a celeron 733 server with 160Mb RAM and
slow IDE disks. My questions is how can I make the most of this hardware.
I want to run RT3 w/ RTFM and this server will be used for nothing else.

So… which do you think would be best:

Apache … Apache2?
modperl … modperl2 … fastcgi?
mysql … postgresql?

Any other optimization tips? I have previously run RT2 and it was decent
performancewise with Apache and regular CGI with mysql 3. I have never used
postgresql so I don’t know if it’s any faster/better. I have never used
fastcgi or modperl2 either.

Thanks for any advice!

-Michael

The best hardware available to me is a celeron 733 server with 160Mb RAM and
160? I am running RT3 with RTFM on a 733 with 256 MbRAM. If you’ve got less,
performance will take a huge hit no matter what you end up with below.

Apache … Apache2?
modperl … modperl2 … fastcgi?
mysql … postgresql?
I’m running Apache1, Mod_perl1 with MySQL. Nothing else runs on the
box, except my Ximian Evolution (psuedo Outlook) and Netscrape…errr
Netscape. It is fast, but I reboot every night to free up memory (cron
job * 3 * * 1-5 reboot and * 4 * * 1-5 /usr/local/apache/bin/apachectl
start …cuz I’m too lazy to use rc3.d)

John

-----Original Message-----
From: Michael Marziani [mailto:marziani@oasis.com]
Sent: Thursday, December 04, 2003 12:21 PM
To: rt-users@lists.fsck.com
Subject: [rt-users] Max performance suggestions on slow hardware?

The best hardware available to me is a celeron 733 server
with 160Mb RAM and slow IDE disks. My questions is how can I
make the most of this hardware. I want to run RT3 w/ RTFM and
this server will be used for nothing else.

So… which do you think would be best:

Apache … Apache2?
modperl … modperl2 … fastcgi?
mysql … postgresql?

Any other optimization tips? I have previously run RT2 and
it was decent performancewise with Apache and regular CGI
with mysql 3. I have never used postgresql so I don’t know
if it’s any faster/better. I have never used fastcgi or
modperl2 either.

Thanks for any advice!

-Michael

Hi there Michael,

I've discovered in my brief introduction to RT/IR thus far that is

definitely a RAM intensive process. I have an odd little frankenbox I was
testing it on that is a P4 2.4 GHz with a mere 256 megs of ram. In
forwarding 300 email to it in relatively quick succession, both apache-ssl
and mysql keeled over horribly due to a lack of memory. The processor
itself rarely came under any significant load at all even with multiple
rt-mailgate/apache-ssl/mysql child processes.

I'm currently rebuilding on the same hardware with a new swap

partition of 1GB and using Gentoo (www.gentoo.org) to tweak for performance.
Rather lengthy install from stage 1, but I’m hoping it’s worth it. I’ll
post back to the list once the build is complete and let you know how it
went.

Cheers,

Tremaine

Hi there Michael,

I’ve discovered in my brief introduction to RT/IR thus far that is
definitely a RAM intensive process. I have an odd little frankenbox I was
testing it on that is a P4 2.4 GHz with a mere 256 megs of ram. In
forwarding 300 email to it in relatively quick succession, both apache-ssl
and mysql keeled over horribly due to a lack of memory. The processor
itself rarely came under any significant load at all even with multiple
rt-mailgate/apache-ssl/mysql child processes.

IT would probably pay to reduce the number of concurrent apache
processes in low memory situations. Also, it’s worth making sure your
MTA knows not to fork off delivery processes if load is skyrocketing.

http://www.bestpractical.com/rt – Trouble Ticketing. Free.

I’m receiving the same error message after an upgrade.

The prevous version was rt 3.0.4, with perl 5.8.0. The packages are from
the time rt 3.0.4 was released.
I’ve upgraded to rt 3.0.7_01 and also upgraded perl to 5.8.2. All
required perl packages are at the latest version.

I’ve still seeking for a solution.

What I didn’t recognize yesterday is that I was recieving a full HTML
encoded error. The error looked like this when displayed in a browser:

error: Can’t use string (“SessionType”) as a HASH ref while “strict
refs” in use at /usr/share/perl5/RT/Interface/Email.pm line 406.

Just complementing, returning back to the old rt 3.0.4 install (keeping
new perl and perl packages and new mysql indexes) did work.

Mauro Carvalho Chehab wrote:

Also pay close attention to mysql’s memory usage. On linux mysql spawns a
child mysqld process for every connection. A good place to look is in the
<mysql_home>/support_files and there are several example cnf files. If you
have a box with little memory, consider the my-small.cnf

Also lowering the connection timeouts can help as well.

A.J.From: rt-users-bounces@lists.fsck.com
[mailto:rt-users-bounces@lists.fsck.com] On Behalf Of Jesse Vincent
Sent: Thursday, December 04, 2003 2:58 PM
To: Tremaine Lea
Cc: 'rt-users@lists.fsck.com’
Subject: Re: [rt-users] Max performance suggestions on slow hardware?

Hi there Michael,

I’ve discovered in my brief introduction to RT/IR thus far that is
definitely a RAM intensive process. I have an odd little frankenbox I was
testing it on that is a P4 2.4 GHz with a mere 256 megs of ram. In
forwarding 300 email to it in relatively quick succession, both apache-ssl
and mysql keeled over horribly due to a lack of memory. The processor
itself rarely came under any significant load at all even with multiple
rt-mailgate/apache-ssl/mysql child processes.

IT would probably pay to reduce the number of concurrent apache
processes in low memory situations. Also, it’s worth making sure your
MTA knows not to fork off delivery processes if load is skyrocketing.

http://www.bestpractical.com/rt – Trouble Ticketing. Free.
rt-users mailing list
rt-users@lists.fsck.com
http://lists.fsck.com/mailman/listinfo/rt-users

Have you read the FAQ? The RT FAQ Manager lives at http://fsck.com/rtfm

Thanks for the advice. I thought I had done that in my troubleshooting but looking back at the history I had downgraded just about everything but RT. Downgraded the package and it works just fine now. Thanks again.

Frank