RE [Samba] panic actions on samba 3.2.2 (also)

[Stéphane PURNELLE]

Jeremy,

On the RedHat Network, I found a bug fixe : 

RHBA-2008:0611 - Bug Fix Advisory 
Détails
Paquetages
Systèmes affectés
Synopsis
nss_ldap bug fix update

Publié :
28/07/08
Mis à jour :
28/07/08

Sujet
An updated nss_ldap package that fixes a bug is now available.
Description
The nss_ldap package contains the nss_ldap and pam_ldap modules. The
nss_ldap module allows applications to use a directory server as a primary
source for information about users and groups.

By default, the nss_ldap module keeps connections to a server open between
uses. In order to avoid problems that would occur if two client processes
attempted to issue queries over the same connection to a server, the 
module
installs a fork handler to allow it to close the client-side of any open
connection for a newly-created child process. This method frequently
caused a SIGPIPE signal to be received by the child process, which in many
cases, caused the child process to exit.

This issue may have caused commands, such as "su" to fail, broken pipe
errors, cron jobs not running, scripts not running as expected, and remote
session hangs.

Users of nss_ldap are advised to upgrade to this updated package, which
resolves this issue.


is this bug fixe can correct my problem ?

thanks

        Stéphane Purnelle

-----------------------------------
Stéphane PURNELLE                         stephane.purnelle at corman.be
Service Informatique       Corman S.A.           Tel : 00 32 087/342467

samba-bounces+stephane.purnelle=corman.be at lists.samba.org a écrit sur 
30/08/2008 19:49:42 :

> Thanks Jeremy.
> 
> but how I use valgrind if the panic action reproduct ?
> 
> If think that's it's particular purpose because it's maybe due to 
failover 
> ldap configuration.
> 
> 127.0.01 is the ldap master
> xx.xxx.x.xx is a ldap slave.
> 
> both smb.conf and ldap.conf contain theses address.
> 
> if ldap slave is not available (network load or slave server restart) 
then 
> it cause nss_ldap problem.
> Starting nscd can be workaround ?
> 
> thanks in advance
> 
>         Stéphane Purnelle
> 
> -----------------------------------
> Stéphane PURNELLE                         stephane.purnelle at corman.be
> Service Informatique       Corman S.A.           Tel : 00 32 087/342467
> 
> Jeremy Allison <jra at samba.org> a écrit sur 29/08/2008 23:32:55 :
> 
> > On Fri, Aug 29, 2008 at 04:57:32PM +0200, Stéphane PURNELLE wrote:
> > > I think that's it's a bug in nss_ldap !
> > > 
> > > in ldap.conf I put local ldap server (127.0.0.1) and an adress for a 

> slave 
> > > ldap server (xx.xxx.x.x).
> > > Each time I have a panic action, I have 
> > > 
> > > Aug 29 15:11:24 coradm01-node1 smbd: nss_ldap: reconnected to LDAP 
> server 
> > > ldap://xx.xxx.x.xx
> > > or 
> > > Aug 29 16:33:22 coradm01-node1 smbd: nss_ldap: reconnected to LDAP 
> server 
> > > ldap://127.0.0.1/ after 1 attempt
> > > 
> > > Coud anyone confirm that
> > 
> > Yes, looking at your backtraces it looks deep inside nss_ldap. Of
> > course that's not 100% - you should probably run under valgrind
> > to be sure.
> > 
> > Jeremy.
> -- 
> To unsubscribe from this list go to the following URL and read the
> instructions:  https://lists.samba.org/mailman/listinfo/samba