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

Re: new admin guide draft

--On Wednesday, September 17, 2003 9:34 AM +0200 Michael Ströder <michael@stroeder.com> wrote:

Luke Howard wrote:
I find that "ldapsearch -h ldap-dev0 cn=ldapsync" returns nothing!
Since  it is a part of the database, shouldn't I be able to see this?

It appears that you need to read this, eg:

$ ldapsearch -h ldap-dev0 -b "cn=ldapsync,dc=stanford,dc=edu" -s base

> sed s/ProviderSubentry/ConsumerSubentry/ dbfile > db_file.new
 > You don't need to do this -- the consumer will suck the database from
 > provider when it starts up.

Just a short note without having used this new replication engine so far:

With another LDAP server product it is sometimes faster to dump the whole
database at once (in a special replica LDIF format), copy it to the
replica system and load the database from LDIF than using the function
"Initialize Consumer" which pushes the whole database via LDAP.

Therefore I could imagine that it'd be nice to do a slapcat (replica
mode) on provider, scp/ftp/whatever to replica system and slapadd on
replica system.

This is exactly my issue -- It currently takes 4.5-6.5 hours (on identical systems, which still puzzles me a little) to load our DB via slapadd. Our DB is >450MB when exported via slapcat. I'm concerned about how long it would take to have my DB transported and indexed over the network. Especially considering that we have 9 replica's and 1 provider. In the case of some type of disaster, I'd not want to have all 9 feeding off the provider.


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