XMailforum is a readonly knowledge archive now.

Registering as a new user or answering posts is not possible anymore.

Might the force be with you, to find here what you are looking for.

2019-09-20 - hschneider, Admin

Cookie Disclaimer: This forum uses only essential, anonymous session cookies (xmailforum*), nothing to be scared of.

XMail Forum [Powered by Invision Power Board]
Printable Version of Topic
Click here to view this topic in its original format
XMail Forum > XMail Server > Xmail Mem Usage & Crashes On Linux Kernel >2.4.1!


Posted by: hschneider Jan 10 2004, 05:15 PM
QUOTE

It has been reported that on some Linux systems the XMail memory size
(RSS) grows in an almost uncontrolled way. It has been verified that this
is a problem with never glibc (maybe 2.3.2). The problem has been verified
on RH9.0. If you have those kind of problems you can use another Linux box
(using older glibc) and using -static inside LD_FLAGS in Makefile.lnx.
Then use the generated image on the newer system.

- Davide

Posted by: hschneider Jan 13 2004, 11:09 AM
This phenomena seems to dissappear, when you:
- build XMail _without_ -static (this dements the mail above)
- additionally put 'export LD_ASSUME_KERNEL=2.4.1' in your XMail startup script.

Looks like the problem here is the same like with XQM Agent: The new threading scheme introduced with RH9 causes unpredictable behavior of multi-threaded applications. Therefore the LD_ASSUME_KERNEL env var exists, which will force the kernel to emulate the old threadiing scheme.

Source (read from bottom to top):

QUOTE

Fine. I experienced a similar thing on RH9 with my XQM Agent:=20 mem was exhausted and the application crashed when huge amounts of data = were scanned ...=20 Switching back to the 2.4.1 thread scheme was the cure.

I think putting LD_ASSUME_KERNEL into your profile is a good idea.

--Harald


> -----Urspr=FCngliche Nachricht-----
> Von: xmail-bounce@xmailserver.org=20 
>[mailto:xmail-bounce@xmailserver.org] Im Auftrag von Pascal de R.
> Gesendet: Dienstag, 13. Januar 2004 08:51
> An: Harald Schneider
> Betreff: [xmail] Re: AW: Re: AW: Re: AW: Re: XMail growing=20  memory
>image ... =20
>=20
>=20
>   Memory is ok :-)
>  =20
>   Running normally around 66 Mo and without crash till yesterday...we
>   have to wait more to be sure about crash!
>=20
>=20
> mardi 13 janvier 2004 at 08:48:16, you said :
>=20
> Harald> Did you check mem usage now? Maybe this prob is also solved.
>=20
> Harald> --Harald
>=20
>=20
> >> -----Urspr=3DFCngliche Nachricht-----
> >> Von: xmail-bounce@xmailserver.org=3D20 =20
> >>[mailto:xmail-bounce@xmailserver.org] Im Auftrag von Pascal de R.
> >> Gesendet: Dienstag, 13. Januar 2004 08:38
> >> An: Davide Libenzi
> >> Betreff: [xmail] Re: AW: Re: AW: Re: XMail growing memory=20
> image ...=20
> >>=3D20 =3D20
> >>=3D20
> >> mardi 13 janvier 2004 at 08:30:34, you said :
> >>=3D20
> >> Davide> On Tue, 13 Jan 2004, Pascal  de R. wrote:
> >>=3D20
> >> >> XMail has running all night without crash....
> >>=3D20
> >> Davide> I lost the track. So, now you have RH9 + -static +=3D20=20
> >> Davide> LD_ASSUME_KERNEL, don't you?
> >>=3D20
> >> No I have :
> >>=3D20
> >> Compil on RH9 without -static but with LD_ASSUME_KERNEL =3D20
> >>=3D20
> >> >> Davide; do you the coredump file of previous crash or can=3D20
> >> i destroy=3D20
> >> >> it ?
> >>=3D20
> >> Davide> You can nuke it.
> >>=3D20
> >>=3D20
> >>=3D20
> >>=3D20
> >> Davide> - Davide


Posted by: hschneider Jan 14 2004, 01:09 PM
QUOTE

Dear Davide,

  Just to information you about XMAIL :

  Running without -static but with LD_ASSUME_KERNEL in the starting
  script.

  Running from Jan 12 :

      - no crash
      - Using never over 85 Mb
     
  Seems very good :-)

Posted by: hschneider Jan 19 2004, 11:17 AM
QUOTE

Dear Veeresh,

lundi 19 janvier 2004 at 07:43:49, you said :

Veeresh> Any link, pointer where I can understand what's happening with
Veeresh> XMail if its on Red Hat 9?


Vendredi 26 décembre 2003 at 10:11:51, Pascal said :

Pascal> First, Merry Christmas everybody :-)

Pascal> I have xmail 1.17 installed on RH9 managing around 500 domains.
Pascal> around 200 e-mail / hour

Pascal> When I start running xmail process, he needs 281MB, 24h after
Pascal> 360 and 3/4 days after 690 MB!. I hope it is a setup problem and
Pascal> I have to correct it because on this computer i'm running
Pascal> Apache, MySql and Bind, and 1GB memory seems to be short if i
Pascal> don't restart xmail every 2 days.

Pascal>   Best regards,
Pascal>   Pascal, pascal@dri.fr

Pascal> help: send the line "help" in the body of a message to
Pascal> ecartis@xmailserver.org


After many test I confirme LD_ASSUME_KERNEL=2.4.1 in the starting script solves the problem.

...


Posted by: atomant Jan 19 2004, 03:04 PM
It is important for all users running kernel > 2.4.1 which also includes RH8 users with updated kernel to the latest version.

Posted by: leyland Feb 5 2004, 12:37 PM
Same Here.. Today keeping Crash... ..

I tested 1.5 1.6 1.7 same problem

added LD_ASSUME_KERNEL=2.4.1, same problem. It will not fix. Before 5-FEB-2004 without this kind crash.

Config :-
RH 7.3
Glibc 2.2.5-44 RPM
GCC 2.96-113 RPM
CELRON 2G
512MB RAM
Anti Virus Filter
SpamAssAssin Filter

Every hour receive 200-500 Emails

Best Regards,
Leyland

Posted by: hschneider Feb 5 2004, 01:12 PM
What is the output of
uname -a

Replace your
export LD_ASSUME_KERNEL=2.4.1
by
export LD_ASSUME_KERNEL="2.4.1"

Where is your export statement located? In your XMail startup script ?
Sure that THIS script is executed?

Posted by: leyland Feb 5 2004, 03:01 PM
QUOTE (hschneider @ Feb 5 2004, 01:12 PM)
What is the output of
uname -a

Replace your
export LD_ASSUME_KERNEL=2.4.1
by
export LD_ASSUME_KERNEL="2.4.1"

Where is your export statement located? In your XMail startup script ?
Sure that THIS script is executed?

[root@sun root]# uname -a
Linux sun.os.st 2.4.22 #2 Sun Sep 21 13:50:10 HKT 2003 i686 unknown
[root@sun root]#

Posted by: hschneider Feb 5 2004, 03:13 PM
Fact is, that you need LD_ASSUME_KERNEL to be set. Please check all other things mentioned above.

Posted by: leylandhk Mar 9 2004, 02:40 AM
QUOTE (hschneider @ Feb 5 2004, 03:13 PM)
Fact is, that you need LD_ASSUME_KERNEL to be set. Please check all other things mentioned above.

I HAVE BEEN ADDED

#! /bin/sh
export LD_ASSUME_KERNEL="2.4.1"
/var/MailRoot/bin/XMail -Md -Ms /var/MailRoot -FI 127.0.0.1:0 -Pt 60 -St 60 -Sl -SI 0.0.0.0:8025 -SI 0.0.0.0:25 -PI 0.0.0.0:8110 -PI 0.0.0.0:110

Today it crashed again!!!!!

I still dunno what happen! sleep.gif;

Posted by: hschneider Mar 9 2004, 07:55 AM
1.17 has a small mem leak, which can cause coredumps under certain conditions. Please use the latest 1.18 pre release:

http://www.xmailserver.org/xmail-1.18.tar.gz

Posted by: vegetto34 Mar 16 2004, 11:26 AM
I have not noticed this on my RH 9 install... Maybe because I only have 2 users and not that much traffic? my server has been up 24/7 for the past month and i've been able to send and receive e-mails without a problem.

amd k6-2 550mhz w/ 128 megs ram. it would have crashed a lot since it doesn't have much room or CPU power to run it well. i'd love to put in a P4 system but for my low traffic it would be overkill... (although administrating it would be much faster)

Posted by: hschneider Mar 16 2004, 12:09 PM
In most cases 1.17 runs stable. But when it crashes, an update to 1.18 is the cure.

Powered by Invision Power Board (http://www.invisionboard.com)
© Invision Power Services (http://www.invisionpower.com)