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

Re: Re[4]: openldap-server-2.2.29: multimaster support





--On Saturday, November 19, 2005 12:22 PM +0200 "Sergey A. Kobzar" <ksa@uaic.net> wrote:

Hello Quanah,

Saturday, November 19, 2005, 5:58:18 AM, you wrote:



--On Friday, November 18, 2005 10:46 PM +0200 "Sergey A. Kobzar"
<ksa@uaic.net> wrote:

Hello John,

Friday, November 18, 2005, 9:23:17 PM, you wrote:

Does the FAQ entry about multi-master replication* need to be updated?
What about the draft "LDAP Multi-master Replication Considered
Harmful"?

In fact I'd like to know if multi-master replication with syncrepl can
be considered as sure or if it is harmful too.

I only use sycnrepl to achieve high availability (by combining with
heartbeat), I haven't tested it as a true multi-master environment.  It
just seems likely to me that it might work, but I don't have the time
personally to mess with it.  Please do feel free to try it out. :)
Personally, I agree with Zeilenga's draft.

I can't think of a situation in which multi-master replication would
actually make any sense anyway.  (The closest scenario I can think of
is a load-balanced configuration, but even then, you can't rely on each
side of the cluster to be up to date at any given point in time since
replication is asynchronous.)

I have situation:

Two offices are connected by low-speed channel. On both offices I have
users, that want change their password without my assistance (by
Samba, for example).

If I'll setup Samba from office with slave use master LDAP server,
speed will be slow...

I think that having two masters syncrepl from each other is still
sufficient here.  Since the password changes are going to go to the local
master, there shouldn't be much risk of data collisions between the two
sites.

One user can present in one of offices. Users don't have duplicates! ;)

I think that is the point I was making? :)

--Quanah

--
Quanah Gibson-Mount
Principal Software Developer
ITSS/Shared Services
Stanford University
GnuPG Public Key: http://www.stanford.edu/~quanah/pgp.html