I'm also seeing new odd behaviour after upgrading libnss-ldap.

As long as nscd is not running, slapd start blocks for about 5-10 minutes. If I start nscd before slapd it works, but I might end up with a invalid cache for some period of time if other lookups have happened and failed due to unavailable ldap server in between.

Reordering the init sequence is probably not a solution due to the inconsistent cache problem mentioned, while current order blocks boot for ages. The strace of slapd seems similar to the udev case.

nscd is supposed to be entirely optional, just behaving as a pluggable cache, so its odd that nss blocks without it if rubbed in the wrong way (like udev, and slapd seems to do). It seems nss is trying to connect to nscd, and failing, thus going the long retry route. Earlier libnss-ldap did not do this, it just bypassed nscd immidiatly if it was unavailable.

I tried both files ldap and compat ldap in passwd, group and shadow entries in nsswitch.conf, but all combinations block as long as ldap is in there.

All system accounts are satisfied with files/compat on this system.


--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Reply via email to