I am trying to track down the cause of some errors that have been occurring on a number of our servers using LDAP. We have noticed that when a certain LDAP group exceeds 65 users we begin seeing glibc errors for users in the group. Users that are not in the group do not exhibit this behavior.
We have seen this issue on machines running Centos 4.5 x86 & x86_64 with glibc-2.3.4-2.36 and RH4 x86_64 running glibc-2.3.4-2.25. We are running Fedora Directory Server 1.0.4 on Centos 4.5. We have added a 3rd FDS slave and turned up debugging but have not seen anything that appears to be relevant in the logs on the 3rd slave.
This problem is not a 65 user limit as we have other groups with well over 65 members that do not display this behavior. We also created a new group with identical users and it did not display this problem.
1) With 66 users in the massweb group
[root@megalon ~]# getent group massweb *** glibc detected *** free(): invalid next size (normal): 0x09a225b0 *** Aborted
2) After removing any user from the massweb group to reduce the total members to 65
[root@megalon ~]# getent group massweb massweb:x:3016:afaxon,ashairza,ccrump,jhorowit,mmarum,morendai,usmall,adroffne,afaxon,amurphre,aoliver,ayellipe,bdonohue,bdunn,beckert,blarsen,bwphilli,catanis,ccrump,chaynes,clyons,cwong,dbarber,dmuse,dnicol,edougher,egallant,ekubosia,evizcain,feeddrop,gtulonen,halokush,hbraverm,jbartus,jbrown,jhobbs,jking,jlederma,jpecora,jthayer,kbalbedi,kbeam,kginn,mbuonfig,mcrawfor,mlarsen,mlong,mwalsh,osedano,proche,prondeau,pweinber,rhorriga,rkersey,rskutins,sasaro,sbrodeur,sculver,smoriart,support,swestenh,tbaltimo,torgelfi,usmall,vvalenti
Has anyone encountered a similar problem? Any suggestions would be most welcome.
Best regards
Bjorn Oglefjorn wrote:
I am trying to track down the cause of some errors that have been occurring on a number of our servers using LDAP. We have noticed that when a certain LDAP group exceeds 65 users we begin seeing glibc errors for users in the group. Users that are not in the group do not exhibit this behavior.
We have seen this issue on machines running Centos 4.5 x86 & x86_64 with glibc-2.3.4-2.36 and RH4 x86_64 running glibc-2.3.4-2.25. We are running Fedora Directory Server 1.0.4 on Centos 4.5. We have added a 3rd FDS slave and turned up debugging but have not seen anything that appears to be relevant in the logs on the 3rd slave.
This problem is not a 65 user limit as we have other groups with well over 65 members that do not display this behavior. We also created a new group with identical users and it did not display this problem.
This sounds like a problem with pam_ldap or nss_ldap. You might want to report this on a CentOS list. If you are a Red Hat customer, you can report it through your usual support channel. I don't think this is a directory server problem.
- With 66 users in the massweb group
[root@megalon ~]# getent group massweb *** glibc detected *** free(): invalid next size (normal): 0x09a225b0 *** Aborted
- After removing any user from the massweb group to reduce the total
members to 65
[root@megalon ~]# getent group massweb massweb:x:3016:afaxon,ashairza,ccrump,jhorowit,mmarum,morendai,usmall,adroffne,afaxon,amurphre,aoliver,ayellipe,bdonohue,bdunn,beckert,blarsen,bwphilli,catanis,ccrump,chaynes,clyons,cwong,dbarber,dmuse,dnicol,edougher,egallant,ekubosia,evizcain,feeddrop,gtulonen,halokush,hbraverm,jbartus,jbrown,jhobbs,jking,jlederma,jpecora,jthayer,kbalbedi,kbeam,kginn,mbuonfig,mcrawfor,mlarsen,mlong,mwalsh,osedano,proche,prondeau,pweinber,rhorriga,rkersey,rskutins,sasaro,sbrodeur,sculver,smoriart,support,swestenh,tbaltimo,torgelfi,usmall,vvalenti
Has anyone encountered a similar problem? Any suggestions would be most welcome.
Best regards
-- Fedora-directory-users mailing list Fedora-directory-users@redhat.com https://www.redhat.com/mailman/listinfo/fedora-directory-users
Bjorn Oglefjorn wrote:
- With 66 users in the massweb group
[root@megalon ~]# getent group massweb *** glibc detected *** free(): invalid next size (normal): 0x09a225b0 *** Aborted
...
Has anyone encountered a similar problem? Any suggestions would be most welcome.
Sounds kind of like a bug in glibc that was fixed in 2.3.4-2.13. You're probably best off reporting this to RH via bugzilla. I might be able to help you track down the precise cause of the problem, if you'd like, and help Red Hat get the issue resolved more quickly.
* confirm the version of glibc on that system * install "ltrace" if necessary * stop the "nscd" service temporarily * run "ltrace -s 512 -S getent group massweb > /tmp/ltrace.getent 2>&1"
Email me /tmp/ltrace.getent off-list. If there is any private info in /etc/ldap.conf, you'll want to clear it out of the ltrace file before sending it.
2007/9/27, Bjorn Oglefjorn sys.mailing@gmail.com:
I am trying to track down the cause of some errors that have been occurring on a number of our servers using LDAP. We have noticed that when a certain LDAP group exceeds 65 users we begin seeing glibc errors for users in the group. Users that are not in the group do not exhibit this behavior.
We have seen this issue on machines running Centos 4.5 x86 & x86_64 with glibc-2.3.4-2.36 and RH4 x86_64 running glibc-2.3.4-2.25. We are running Fedora Directory Server 1.0.4 on Centos 4.5. We have added a 3rd FDS slave and turned up debugging but have not seen anything that appears to be relevant in the logs on the 3rd slave.
This problem is not a 65 user limit as we have other groups with well over 65 members that do not display this behavior. We also created a new group with identical users and it did not display this problem.
- With 66 users in the massweb group
[root@megalon ~]# getent group massweb *** glibc detected *** free(): invalid next size (normal): 0x09a225b0 *** Aborted
You should disable the GLIBC internal sanity via MALLOC_CHECK_ environment variable with a zero(0) value. The RHEL4 Release Notes show how do that, check glibc section for more information: http://www.red-hat.com/docs/manuals/enterprise/RHEL-4-Manual/release-notes/w...
389-users@lists.fedoraproject.org