Am 31.07.2010 04:46, schrieb masarati@aero.polimi.it:
Every other write gets chained just fine when a slave is in this condition. It's only the PASSMOD operations that are stuck.
One quick question: can you tell the parameters of the offending PASSMOD operations? I mean: old/new password, password generated automatically or provided, operation performed for self or by a privileged identity, and so?
Thanks, p.
The PASSMODs are done by Linux 'passwd', using pam_ldap (mostly on Debian Lenny, if that matters). The new passwords are provided by the user and the operation is perfomed as "self". I don't know if pam_ldap provides the old password as a parameter to the PASSMOD operation.
Binding is done with 'simple bind'. The whole SASL/Kerberos stuff is configured and working, but not yet deployed.
I could add the client config (PAM, ldap.conf and so on) to the server related stuff in the linked directory, if you need it.
Regards, Christian Manal