Hi List, I have cfg LDAP multimaster replication, one of the hosts will be offline for some days, do I need to disable the replication agreement completely at this point? (what will be the minimum cfg) What are the steps to resync the master after is been brought online ? Thank you Isabella
On Fri, 2015-04-24 at 12:50 -0700, ghiureai wrote:
Hi List, I have cfg LDAP multimaster replication, one of the hosts will be offline for some days, do I need to disable the replication agreement completely at this point? (what will be the minimum cfg) What are the steps to resync the master after is been brought online ? Thank you Isabella
I think that removing the replication agreement is the best course of action.
Say you have A <-> B, and you plan to disable B.
You would disable the replication agreement of B -> A first, then A -> B
Then when you reactivate B, you would add the replication agreement for A -> B, with the attribute nsds5beginreplicarefresh: start
Then once complete, you would re-add B -> A.
Sincerely,
On 04/27/2015 12:59 AM, William wrote:
On Fri, 2015-04-24 at 12:50 -0700, ghiureai wrote:
Hi List, I have cfg LDAP multimaster replication, one of the hosts will be offline for some days, do I need to disable the replication agreement completely at this point? (what will be the minimum cfg) What are the steps to resync the master after is been brought online ? Thank you Isabella
I think that removing the replication agreement is the best course of action.
Say you have A <-> B, and you plan to disable B.
You would disable the replication agreement of B -> A first, then A -> B
Then when you reactivate B, you would add the replication agreement for A -> B, with the attribute nsds5beginreplicarefresh: start
Then once complete, you would re-add B -> A.
you can avoid the reinitialization if you ensure that the changelog is not purged too much, if the server is down for "somedays" set nsds5replicapurgedelay: somedays+1, when the other server is coming online replication should catch up (disabling and reenabling should be enough)
Sincerely,
389-users@lists.fedoraproject.org