Yahoo Groups archive

Milter-greylist

Index last updated: 2026-04-28 23:32 UTC

Thread

memory footprint problem in 1.5.3->1.6rc1 update

memory footprint problem in 1.5.3->1.6rc1 update

2004-11-30 by Matthieu Herrb

Hi,

I've tried to upgrade our mail servers (sparc running Solaris 2.8) from 
1.5.3 to 1.6rc1. I've noticed almost immediatly that the memory usage 
climbed from ca 15MB to more than 220MB for a greylist.db that is around 
100k entries. This makes the new version almost unusable on our current 
machines (Netra X1 with 512MB of RAM).

Any ideas of what could cause a so big difference ?
-- 
Matthieu Herrb

Re: [milter-greylist] memory footprint problem in 1.5.3->1.6rc1 update

2004-11-30 by manu@netbsd.org

Matthieu Herrb <matthieu.herrb@...> wrote:

> I've tried to upgrade our mail servers (sparc running Solaris 2.8) from
> 1.5.3 to 1.6rc1. 

Heh, the release candodate trick worked one more time to get people
testing the beast :)

> I've noticed almost immediatly that the memory usage 
> climbed from ca 15MB to more than 220MB for a greylist.db that is around
> 100k entries. This makes the new version almost unusable on our current
> machines (Netra X1 with 512MB of RAM).
> 
> Any ideas of what could cause a so big difference ?

Does it remains steady at 220 MB? I'd like to be sure it's not a memory
leak.

I recall some token ordering in the parser could eat huge amount of
memory. Maybe that's the problem.

-- 
Emmanuel Dreyfus
Il y a 10 sortes de personnes dans le monde: ceux qui comprennent 
le binaire et ceux qui ne le comprennent pas.
manu@...

Re: [milter-greylist] memory footprint problem in 1.5.3->1.6rc1 update

2004-12-01 by Matthieu Herrb

manu@... wrote:
> Matthieu Herrb <matthieu.herrb@...> wrote:
> 
> Does it remains steady at 220 MB? I'd like to be sure it's not a memory
> leak.
> 
Yes, once started it remains steady.

> I recall some token ordering in the parser could eat huge amount of
> memory. Maybe that's the problem.
> 
I'll try to do some off-line debugging, using our greylist.db as input.

-- 
					Matthieu

Re: [milter-greylist] memory footprint problem in 1.5.3->1.6rc1 update

2004-12-01 by Matthieu Herrb

Matthieu Herrb wrote:

> 
> I'll try to do some off-line debugging, using our greylist.db as input.
> 

Ok, I found the problem. Milter-greylist is innocent. It's a Solaris 
bug. Solaris 8 needs the 108827-40 patch (which has since then been 
obsoleted by 108993-39). There seem to be severe leaks in getaddrinfo() 
and friends in non-patched libpthread/libsocket/libnsl.

-- 
Matthieu Herrb

Re: [milter-greylist] memory footprint problem in 1.5.3->1.6rc1 update

2004-12-01 by manu@netbsd.org

Matthieu Herrb <matthieu.herrb@...> wrote:

> Ok, I found the problem. Milter-greylist is innocent. It's a Solaris 
> bug. Solaris 8 needs the 108827-40 patch (which has since then been 
> obsoleted by 108993-39). 

Well done!

> There seem to be severe leaks in getaddrinfo() 
> and friends in non-patched libpthread/libsocket/libnsl.

In an ideal world, we could check for that bug and avoid using
getaddrinfo() if it is broken. Are you able to write a test? 

-- 
Emmanuel Dreyfus
Il y a 10 sortes de personnes dans le monde: ceux qui comprennent 
le binaire et ceux qui ne le comprennent pas.
manu@...

Re: [milter-greylist] memory footprint problem in 1.5.3->1.6rc1 update

2004-12-01 by S L

On Wed, Dec 01, 2004 at 01:49:57PM +0100, Matthieu Herrb wrote:
> Matthieu Herrb wrote:
> 
> >
> >I'll try to do some off-line debugging, using our greylist.db as input.
> >
> 
> Ok, I found the problem. Milter-greylist is innocent. It's a Solaris 
> bug. Solaris 8 needs the 108827-40 patch (which has since then been 
> obsoleted by 108993-39). There seem to be severe leaks in getaddrinfo() 
> and friends in non-patched libpthread/libsocket/libnsl.
> 
> -- 
> Matthieu Herrb


Cool !
Clearly milter-greylist 1.6rc1 memory footprint is not far
from the one of version 1.4 on my linux boxes.
I wish I could say the same for SpamAssassin through Milters :-)  
Gotta get rid of SpamAssassin and try j-chkmail instead  ... 

SL/

Re: [milter-greylist] memory footprint problem in 1.5.3->1.6rc1 update

2004-12-02 by Christian Pelissier

>Matthieu Herrb <matthieu.herrb@...> wrote:
>
>> Ok, I found the problem. Milter-greylist is innocent. It's a Solaris 
>> bug. Solaris 8 needs the 108827-40 patch (which has since then been 
>> obsoleted by 108993-39). 
>
>Well done!
>
>> There seem to be severe leaks in getaddrinfo() 
>> and friends in non-patched libpthread/libsocket/libnsl.
>
>In an ideal world, we could check for that bug and avoid using
>getaddrinfo() if it is broken. Are you able to write a test? 
>

108993-39 SunOS 5.8: LDAP2 client, libc, libthread and libnsl libraries patch

This patch correct LDAP, PAM and a lot of libraries  and is available as 108994 
for X86.


It seems that this old Solaris 8 bug doesn't affect Solaris 9 : Solaris 9 LDAP 
patch patchs only ldap. 

--
Christian Pélissier
Office National d'Études et de Recherches Aérospatiales BP 72 92322 Chatillon
Tel: 33 1 46 73 44 19, Fax: 33 1 46 73 41 50

Re: [milter-greylist] memory footprint problem in 1.5.3->1.6rc1 update

2004-12-06 by Matthias Scheler

On Wed, Dec 01, 2004 at 07:24:59PM +0100, Emmanuel Dreyfus wrote:
> > There seem to be severe leaks in getaddrinfo() 
> > and friends in non-patched libpthread/libsocket/libnsl.
> In an ideal world, we could check for that bug and avoid using
> getaddrinfo() if it is broken. Are you able to write a test? 

I think it would be better to abort "configure" and ask the admin
to install the patch that Matthieu mentioned.

	Kind regards

-- 
Matthias Scheler                                  http://scheler.de/~matthias/

Move to quarantaine

This moves the raw source file on disk only. The archive index is not changed automatically, so you still need to run a manual refresh afterward.