There are 2 types of replication for clustered servers.
The first is cluster replication and is event driven. Thus if a db on one server receives an update, the cluster should push this to the other server(s) in the cluster. Monitor the statistics (e.g. replica.cluster.secondsonqueue). If this number is very high (> 60) or has peaks, you should increase the number of cluster replicator tasks on the servers.
The second type of replication is scheduled replication. This is a “backup” for the cluster replication. Any changes that are not replicated via the cluster replication should be transferred via scheduled replication. I would schedule this at least once per day (and via program documents “On startup”).
See the redbook below for more details (an oldie but a goodie).
i don’t think this will solve the problem, because the replication between servers is working well . but the issue in some mail boxes that does not replicate or late in replicate.
I am not quite sure why you think that this will not solve the problem.
If cluster replication is failing, it is probably because the cluster replicator task cannot cope with the amount of changes that it has to replicate to the cluster server during “peak” periods.
As I mentioned in my previous response, the way to deal with this is to add an extra cluster replicator task.
At any rate, because you have differences, you should also consider defining scheduled replication between your cluster servers.
the value of replica.cluster.secondsonqueue is 61 i think as you told me , i must change this value.so what is the best number to assign to cluster replicators task ( 0 , 1 ,or any other number)