--- Begin Message ---
David,
Do you write the transaction log files synchronously to disk, i.e. you don't
use "set_flags DB_TXN_NOSYNC" in your DB_CONFIG? Do you also checkpoint the
database on a regular interval? What manual steps do you perform when slapd
is crashed? Slapd does not do automatic recovery in OpenLDAP 2.1.23.
It might help if you provide your slapd.conf and DB_CONFIG to this
mailinglist.
B.t.w. it is also useful to give some logging or gdb output to see why your
slapd is actually crashing......
Regards,
Erik
On Wednesday 29 October 2003 03:38, David Shirley wrote:
> hi all,
>
> recently we have begun to notice that our master server is not in sync with
> the slaves.
>
> more specifically it seems to be rolling back to old data eg:
>
> user changes password
> master pushes out password change to slave
> *master crashes*
> when master reboots master has old password.
>
> now my question is this caused by the caching of information in bdb or
> the caching of writes done by the disk?
>
> cheers
> dave
>
> /-----------------------------------
> David Shirley
> System's Administrator
> Computer Science - Curtin University
> (08) 9266 7817
> -----------------------------------/
--- End Message ---