Re: SyncREPL Testing

["Jong-Hyuk" <jongchoi@OpenLDAP.org>]

> Ok, it appears I have SyncREPL working.  Wonderful!
>
> But it has raised a couple questions...
>
> 1.) What does the "rid" actually mean?  Is this value some kind of
'magic'?  A
> value of 50 works,  what makes this value special?

No. there's nothing special with the number.
The original problem was due to the mismatch between the rid specified in
the slapd.conf and the rid stored when consumer.ldif was added.
slapadd -r -i <rid list> will generate special subentries whose RDNs are
suffixed by the rids in the <rid list>. The subentries contain the
synchronization cookies for each replication specification by which the
provider server decides the sychronization status of the consumer. With "-i
100" option, the consumer database now contains the subentry for rid=100 as
specified in the slapd.conf.
rid is a positive value having up to three decimal digits. slapd interprets
it correctly, but slapadd -i <rid> didn't. slapadd -i <rid> was only fine
with rids having up to two decimal digits. This was fixed in HEAD.

> 2.) If I have multiple consumers should the rids all be the same or
unique, or
> does it matter?  I've read the SyncRepl section of the Admin guide but it
> doesn't really explain what the RID means.  Do I need to go read the
syncrepl draft?

The meaning of rid is local to the consumer server. It distinguishes between
more than one replications in the consumer server. The consumer will
retrieve the required information from the subetnry whose RDN is suffixed by
the rid given by the syncrepl specification in slapd.conf.

- Jong-Hyuk