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

Re: slurpd's disaster recovery capability

I think it would be easier for you to use a transactional backend on
the master (bdb) and use syncrepl for more reliable slave-updates. 
That way, if the master powers down, updates can be initiated by the
slaves when the master comes back up.

I find the integrity of slurpd-written replicas to be about 98% when
the master server is fully powered up, so I don't like to think about
power outages.  :)

On 6/28/05, Sung Ho Jee <jee.sung@ansaldo-signal.com.au> wrote:
> Hi,
> I am investigating whether openLDAP is suitable for our production use.
> My question is, if the machine running the master slapd instance is powered
> down in the middle of replication by slurpd to various slave slapd
> instances, would slurpd attempt to complete the replication upon restart?
> i.e. Does the replication implementation guarantee that all slaves will
> come into sync with the master in this case?