Re: Read Performance during Replication on Slave

[Quanah Gibson-Mount <quanah@stanford.edu>]

--On Friday, July 15, 2005 10:38 PM +0200 Axel Urban <mailing@udslinux.de> 
wrote:

> Hi everybody,
>
> iÂm maintaining a openldap-infrastructure with 1 master and 7 slaves
> (slurpd-replication). The bdb-database have about 1 million entrys (2
> GB). Every morning we make a "provisioning" with new or deleted entrys
> (a perl skript, that adds one entry after the other).
> This takes about 3-5 hours (2000-5000 entrys) of replication. During
> this time, the read-operations on the slaves are very bad, so that our
> radius-servers have timeouts and users cannot authenticate.
>
> Is it possible to make a priority for read-operations? Or to configure a
> priority for special IP-adresses? Or have 2 slapd-instances, one for
> writing and one for reading?

What version of OpenLDAP are you using?  What version of BDB? Have you read 
through the BDB cache tuning suggestions on the OpenLDAP FAQ?  Have you 
tuned OpenLDAP with any cachesize and idlcachesize settings?

--Quanah

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

"These censorship operations against schools and libraries are stronger
than ever in the present religio-political climate. They often focus on
fantasy and sf books, which foster that deadly enemy to bigotry and blind
faith, the imagination." -- Ursula K. Le Guin