Bug#308825: gdm and libnss_ldap works for me

[Rob Caelers]

Martin Samuelsson wrote:
> For what it is worth, I think libnss_ldap and gdm works for me. I don't
> know if I am lucky or if the bug is fixed in upstream release 2.8.0.6?

No. The problem is not fixed in 2.8.0.6. I still need a patched gdm to
be able to log in.

> The last mail to this bug report is posted after that version entered
> unstable, but maybe the debugging was made against the version in
> testing? Could that please be clarified?

My analysis was against a 2.6 version of gdm from unstable. However,
2.8.0.6 (also from unstable) still contains the same code and same bug
as 2.6.

> It could of course be something in my configuration that makes it work
> for me while not working for others. (I for example don't use ldaps)

With TLS disabled, I still can't login with gdm. However, the following
messages seem to be gone when TLS is disabled:

Jan  1 09:52:01 ensor isdnlog: nss_ldap: reconnecting to LDAP server...
Jan  1 09:52:01 ensor isdnlog: nss_ldap: reconnected to LDAP server
after 1 attempt(s)

As far as I remember, this problem surfaced months after the last update
 of libnss-ldap or openldap, so perhaps something else is causing the
problem. Furthermore, I also have a computer that doesn't have this
problem. The only difference between the two computers is that the
'broken' one runs nscd. Well, at least for a very short time:

robc:~$ sudo /etc/init.d/nscd start
Starting Name Service Cache Daemon: nscd.
robc:~$ ps auxw | grep nscd
root      8519  0.4  0.4  11528  2288 ?        Ssl  19:44   0:00
/usr/sbin/nscd
robc      8529  0.0  0.1   1656   532 pts/0    S+   19:44   0:00 grep nscd
robc:~$ sudo strace -p 8519
Process 8519 attached - interrupt to quit
time(NULL)                              = 1136141081
epoll_wait(10, bff5d91c, 100, 29988)    = -1 EINTR (Interrupted system call)
+++ killed by SIGABRT +++

Needs further investigation.

Groetjes,
  Rob