On Monday 24 March 2008 19:40:01 Naufal Sheikh wrote:
> Hi,
>
> Ok, I am using openldap 2.2.20 on both machines. My production server is
> Solaris 8, while my backup machine is redhat linux 8. I am not really using
> some kind of sophisticated replication scheme, but simply this is what I
> have done.
>
> I have added replog attribute in the slapd.conf of my backup machine. I
> switch off my production for maintainance, and swithc the backup on. AS it
> has replog enabled it starts creating logs of the events, After
> maintainance activity I ftp the replog to production and use ldapmodify to
> apply those logs on production.
This is really not a good idea. You may rather want to consider one of these
options:
1)Running read-only on the slave during maintenance on the master
2)Use cluster software to run an HA master
3)Upgrade to 2.4 and run mirrormode or multi-master
> ldapmodify script which I am using is :
>
> LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/main/soft/openldap/openldap-2.2.20
> /lib:/main/soft/openssl/openssl-0.9.7e/lib:/main/soft/berkeleydb/db-4.2.52
> /lib
> export LD_LIBRARY_PATH
>
> /main/soft/openldap/openldap-2.2.20/bin/ldapmodify \
> -d 7 \
> -v -x -W -D "cn=nsadmin" -h hostname -p 389 -f /main/backup replog.
>
If you are processing a replication log, you should use the identity that is
listed in the updatedn parameter on the target. It is the only identity
allowed to write to operational attributes.
Regards,
Buchan