[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: LDAP slurp problem

Pierangelo Masarati wrote:

It might have something to do with the fact that the replog is unique; where did you put the "replogfile" directive? I also note you're using a very old release, this might be part of the problem. I don't recall what issues were present in that release. I suggest you put the "replogfile" directive in the global scope (i.e. before any backend/database directive). With 2.2 the setup you're describing works with no problem.

As far as I understand, either you use a single "replogfile" directive in the global section or separate "replogfile" directives in each database section. In any case, you need to run as many slurpds as the "replogfile" directives are, specifying the replogfile with the "-r" slurpd command line option. In any case, you have the problem that the "host:port" line is the same for all the replicas, so each slurpd will try to apply the change. Binding will succeed for all, since the replication identity is the same for all; but only the right one will be able to apply the change, the others will return with LDAP_REFERRAL (if "updateref" is set) or with something like LDAP_UNWILLING_TO_PERFORM because there's no referral knowledge. In your case, I suggest you use a single "replogfile" in the global section of slapd.conf and try to make the host names different (e.g. using aliases, or the IP number or so). Even some letters in upper/lowercase, e.g. "SERVER1" vs. "server1" should suffice, since slurpd (at least in 2.2) uses case-sensitive match to determine if an entry is associated to a "host:port" key.


   SysNet - via Dossi,8 27100 Pavia Tel: +390382573859 Fax: +390382476497