On Sun, Sep 22, 2013 at 12:35 AM, Gareth Collins <[email protected] > wrote:
> Hello, > > I see that ApacheDS 2.0 supports Master-Master replication which is great > (based on RFC 4533)! I understand that Master-Master replicas will have > eventual consistency? > yes > > Are there any potential "gotchas" setting up Master-Master replication. > e.g.: > > (1) Anything special you need to do in the case of a network partition? > > if a peer is unavailable then the server will keep checking for its availability in regular intervals and meanwhile stores all the modifications in a log which will be sent when the peer come back online. > (2) Are there any scenarios where Apache DS Master-Master instances > can get into a state where they cannot synchronize (i.e. a conflict) which > would require manual intervention? If yes, how would a user know that > a conflict exists? > > not any that I have run into, but I will depend on the contextCSN value stored in ou=system of each server to detect if they are in sync i.e when two servers A and B are in M-M mode then they are in sync if there is an entry Eb in B with the entryCSN attribute value matching with the contextCSN value of A and vice versa. Any info here, would be a bug help! > > thanks in advance, > Gareth > -- Kiran Ayyagari http://keydap.com
