Re: Some records missing in replication

[John Morrissey <jwm@horde.net>]

On Tue, Oct 29, 2002 at 08:19:30AM +0100, Roland Rosenfeld wrote:
% I'm running OpenLDAP 2.0.25 and have to notice, that some changes 
% aren't replicated from the master to the slaves.  The master servers
% two slaves and the problems are synchronous to both slaves.  One of
% the slaves runs on the same machine as the master (for backup
% purpose), so this cannot be a network problem but has to be a problem
% with OpenLDAP.  I notice an average of 1-2 failed replications per
% day, which is IMHO too much.
% 
% Any idea where I could start searching for the reason of this problem?
% I don't have any idea, what's going wrong.  The problem was there with
% 2.0.23 before, the upgrade to 2.0.25 doesn't change anything.

FWIW, I've seen something similar, although not to the degree that you're
seeing it. Every once in a while (every few weeks), I'll notice that a new
entry created on the master wasn't propagated to any slaves. Nothing ever
seems out of the ordinary; if the entry was ever in the replog, it had been
cycled out of it by the time I looked. Nothing in the replication reject
logs. Just never got replicated at all.

% Maybe this has to do with the fact, that there are two slurpd.replog
% (and slurpd.replog.lock) files on the machine, one (empty) in
% var/openldap-slurp and another one in var/openldap-slurp/replica (this
% one keeps the data).  All four files have the same time stamp.
% In the config I only mention var/openldap-slurp/slurpd.replog, but
% slapd and slurpd seem to use var/openldap-slurp/replica/slurpd.replog
% instead (no idea what this means).

This is normal. slapd writes a master replog, then slurpd moves replog
entries out of the master replog and into the replog in replica/.

john
-- 
John Morrissey          _o            /\         ----  __o
jwm@horde.net        _-< \_          /  \       ----  <  \,
www.horde.net/    __(_)/_(_)________/    \_______(_) /_(_)__