| Howard Chu 2007-11-15, 1:11 am |
| Emmanuel Lecharny wrote:
> Howard Chu wrote:
> Here you are :
> http://cwiki.apache.org/confluence/...xSRVx11/Mitosis
>
> These pages are really very rough.
Thanks. I see you've read through all the previous failures on this path...
[vbcol=seagreen]
> The pb is that if you have a lot of replication, it may not be a viable
> approach. Also if you want to implement the 'paging control', as it's up
> to the client to request the next results, you have no good solutions...
> At some point, I think we should consider that a LDAP server is not like
> a RDBMS system with a full transaction system, and you must assume that
> dirty read is a standard mode. That does not mean we shouldn't consider
> blocking incoming requests until a sync-up is done when a server is
> restarted.
In OpenLDAP I've recommended that a replica chain all incoming requests to a
provider while it's in startup mode. There can be transient inconsistencies at
later points in its runtime, and we live with those. But if you're bringing up
a brand new server, it's best to pretend it doesn't even exist until it's at
least 80-90% sync'd. Obviously you can manage this outside of LDAP (e.g., if
you have a DNS pool or hardware load-balancer directing clients to servers,
don't add the new replica to the pool until it's ready) for a new server.
> This is not a simple problem...
Heh heh. So very true.
--
-- Howard Chu
Chief Architect, Symas Corp. http://www.symas.com
Director, Highland Sun http://highlandsun.com/hyc/
Chief Architect, OpenLDAP http://www.openldap.org/project/
|