nscd daemon stops on 64-bit amd works fine 32-bit

[TheArnold]

I was told to report this as a bug here is the link with all the info
https://answers.launchpad.net/ubuntu/+question/8812
With upgrade from edgy to feisty or fresh install of feisty nscd daemon amd 64-bit version. 32 bit server version works fine. Ncsd
starts for a minute to fifteen minutes than stops. if you delete the nscd's database it will run longer,but ncsd will stop running with no error message.

Thanks
-Glenn

[Steffen Neumann (sneumann)]

I can confirm that nscd (combined with LDAP ?) is completely unusable under 64bit.
I am running 2.7-5ubuntu2 in a Hardy alpha-6. The LDAP is living on a Novell eDirectory,
and has some ~200 users and a number of other objects.

Yours,
Steffen

26191: Reloading "sshd" in password cache!
26191: Reloading "root" in password cache!
26191: Reloading "root" in group cache!
nscd: getentry.c:36: ldap_first_entry: Assertion Chain != ((void *)0)'
failed.
Aborted (core dumped)

Another try from within gdb:
5680: remove GETPWBYUID entry "0"
5680: remove GETPWBYNAME entry "sshd"
5680: remove GETPWBYUID entry "111"
5680: remove GETPWBYNAME entry "root"
5680: Reloading "44000" in password cache!

Program received signal SIGPIPE, Broken pipe.
[Switching to LWP 5683]
0x00007f376051f54b in ?? ()

[Steffen Neumann (sneumann)]

Might be this one:

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=362966

Yours,
Steffen

[TheArnold (garnold-unrealsolutions)]

I will give this a try and let you know. Fortunately, I upgrade my main
server which had ldap on the server and improved the performance of ldap
that nscd has not been a big issue. But, it would be nice to get nscd
working properly though.

-Glenn

-----Original Message-----
From: Steffen Neumann [mailto:<email address hidden>]
Sent: Tuesday, March 25, 2008 8:05 AM
To: Glenn Arnold
Subject: [Bug 122815] Re: nscd daemon stops on 64-bit amd works fine
32-bit

Might be this one:

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=362966

Yours,
Steffen

--
nscd daemon stops on 64-bit amd works fine 32-bit
https://bugs.launchpad.net/bugs/122815
You received this bug notification because you are a direct subscriber
of the bug.

[TheArnold (garnold-unrealsolutions)]

Steffen,

Thanks for the info. I changed my Administrator uid from 0 to 300 in ldap and that has appeared to fix the problem on my secondary ldap server. Nscd has been running for eight hours now. But, I will admit the server is not get hit too hard this week because of spring break at my school customer. But, I had to delete the nscd database and start fresh. ON the server with the main ldap server it ran for a while and I got the following error.

1249: Reloading "0" in password cache!
1249: Reloading "1000" in password cache!
1249: Reloading "8278" in password cache!
1249: Reloading "hsfnp01" in hosts cache!
1249: Reloading "0" in group cache!
1249: Reloading "15570" in group cache!
1249: Reloading "1" in password cache!
1249: Reloading "2005" in group cache!
1249: Reloading "110" in password cache!
1249: Reloading "garnold" in group cache!
1249: Reloading "102" in password cache!
1249: Reloading "111" in password cache!
1249: Reloading "103" in password cache!
1249: Reloading "104" in password cache!
1249: Reloading "1023" in group cache!
1249: Reloading "7163" in password cache!
1249: Reloading "15572" in group cache!
1249: Reloading "113" in password cache!
1249: Reloading "15573" in group cache!
1249: Reloading "1679" in password cache!
1249: Reloading "root" in group cache!
1249: Reloading "114" in password cache!
1249: Reloading "106" in password cache!
1249: Reloading "1004" in group cache!
1249: Reloading "DU012PC5736$" in password cache!
1249: Reloading "15574" in group cache!
nscd: cache.c:143: cache_add: Assertion `newp->key + newp->len <= table->head->first_free' failed.
Aborted (core dumped)

I did delete the nscd database and start nscd of fresh. Any ideas?

Thanks
-Glenn
-----Original Message-----
From: Steffen Neumann [mailto:<email address hidden>]
Sent: Tue 3/25/2008 8:05 AM
To: Glenn Arnold
Cc:
Subject: [Bug 122815] Re: nscd daemon stops on 64-bit amd works fine 32-bit

Might be this one:

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=362966

Yours,
Steffen

--
nscd daemon stops on 64-bit amd works fine 32-bit
https://bugs.launchpad.net/bugs/122815
You received this bug notification because you are a direct subscriber
of the bug.

[Steffen Neumann (sneumann)]

Just an update, we have also had many assertion failed related to LDAP in other applications,
so we backported (simple compile+install) 2.4.11 of libldap:

libldap-2.4-2 2.4.11-0ubuntu3

nscd is still crashing:

22940: handle_request: request received (Version = 2) from PID 24545
22940: GETFDPW
22940: provide access to FD 4, for passwd
22940: handle_request: request received (Version = 2) from PID 24545
22940: GETPWBYUID (45678)
22940: Haven't found "45678" in password cache!
nscd: getentry.c:36: ldap_first_entry: Assertion `chain != ((void *)0)' failed.

Any ideas ?

Yours,
Steffen

[Steffen Neumann (sneumann)]

Another update:
We brought back the getentry.c:ldap_first_entry()
in an ancient the version from openldap 2.1

http://www.openldap.org/devel/cvsweb.cgi/~checkout~/libraries/libldap/getentry.c?rev=1.20.2.1&cvsroot=OpenLDAP-src&hideattic=1&sortbydate=0

It is working fine for a few days, although I have no clue
what we break when returning NULLs.

Yours,
Steffen

[Steffen Neumann (sneumann)]

N.B. our patch is in libldap-2.4-2 2.4.11-0ubuntu3,
so it might be an openldap bug after all.

Steffen

[Steffen Neumann (sneumann)]

Hi again,

We're running lucid since a few month,
and with nscd 2.11.1-0ubuntu7.2
we have not been bitten by this anymore.

Closing bug.

Yours,
Steffen

[Steffen Neumann (sneumann)]

Somewhere between 2.7-5ubuntu2 and 2.11.1-0ubuntu7.2
there has been a fix.