I've noticed that in bulk-replication mode, strict journaling slows things down 
quite a bit. Switching to fast journaling helps quite a bit. That seems odd to 
me: it makes sense to be strict about user transactions on the master forests, 
but it seems like the bulk replication could use the fast mode until it catches 
up. Anyway if you are using strict journals, you might try fast until the 
replicas have caught up. Then you could go back to strict.

If that isn't it, try to find out what the bottleneck is. My guess is either 
the disk or the network. I've seen replication fill up gigabit ethernet, for 
example, and replication can put heavy demands on the disks. If that's it, you 
would probably need to increase the performance capacity of the limiting 
subsystem.

-- Mike

On 2 Apr 2013, at 18:35 , Danny Sinang <[email protected]> wrote:

> Hello,
> 
> Is there an ML config setting I can tweak to speed up the rebuilding / 
> catch-up of forest replicas.
> 
> Regards,
> Danny
> _______________________________________________
> General mailing list
> [email protected]
> http://developer.marklogic.com/mailman/listinfo/general

_______________________________________________
General mailing list
[email protected]
http://developer.marklogic.com/mailman/listinfo/general

Reply via email to