Hello,
I have two LDAP-Master with refresh and persist configured. Both are located in different countries and the bandwidth between them is only 10Mbit/s. Both masters are used for read and writing entires (writing from the local country), the Setup is running well. But now, somehow one Server is crashing. My application is detecting this and is now writing to the other remote Server in the different country.
To recover the crashed server, I stop the slapd with the crashed database, and delete the whole database directory and start the slapd again. The slapd now is syncing with the other remote slapd (present and delete phase). The syncing will take about 1,5 hour.
But what will happen during this time. My application is detection that the local slapd is running again and will read an write entires to the this server which is in the syncing state. In this same time, the other server in the remote site is trying to establish the replication to the crashed server who is not fully synchronized, but who may have some new entires.
* Can it be, that the remote server now receive false information an gets delete entires during this state? * Do I have (manually) prevent for writing to the syncing server?
Can someone help me for this Setup - Especially the recovery procedure? ..
Thanx in advance
Jeff