replication tasks, configure freenas to run more than one at a time?

[Rich B]

It looks like replication tasks occur only one at at time. Is that true? If so, is there a way to configure freenas to execute two or three replication tasks at at time?

Some of my replication tasks are long running and therefore block all other replication tasks for a day or so at at time. I was hoping that I could configure replication to allow my long running tasks to proceed normally, while also allowing other smaller tasks to proceed unimpeded.

 

[cyberjock]

No. Replication takes your primary pool and bounces it off your secondary and basically 'catches up' anything that doesn't match. If you have no older stuff, you can't replicate the newer stuff until the older is there. So no, this isn't possible because of how replication works.

 

[panz]

Some of my replication tasks are long running and therefore block all other replication tasks for a day or so at at time.

Are you using v. 9.2.1.3? If so, replication has a bug and it is very slow. Replace the autorepl.py as described here

http://forums.freenas.org/index.php?threads/replication-task-speed-at-3-50m.19782/#post-111147

 

[cyberjock]

Are you using v. 9.2.1.3? If so, replication has a bug and it is very slow. Replace the autorepl.py as described here

http://forums.freenas.org/index.php?threads/replication-task-speed-at-3-50m.19782/#post-111147

Or upgrade to the 9.2.1.4-beta that should have all of the nasty 9.2.1.x bugs fixed.

 

[mpfusion]

This is true for a single dataset. But what prevents two datasets from being replicated in parallel, if they don't depend on each other? Take the following layout, for instance:

Code:

tank/datasetA
     datasetA@auto-20140401.2000-1m
     datasetA@auto-20140402.2000-1m
     datasetA@auto-20140403.2000-1m
 
tank/datasetB
     datasetB@auto-20140401.2000-1m
     datasetB@auto-20140402.2000-1m
     datasetB@auto-20140403.2000-1m

 

[panz]

Replicating two datasets at the same time will not increase your bandwidth; so, what's the reason to do that?

 

[mpfusion]

In cases where bandwidth is not the limiting factor. On some datasets replication almost maxes out my network, but on others bandwith is not the issue. See this graph, for instance:

This is the graph from my backup system replicating eleven datasets. Replication started at around 10am. As you see it's still replicating and doesn't max out the network at all. I could imagine if freenas could replicate those datasets in parallel, replication would finish much quicker.

 

[Rich B]

Running FreeNAS-9.2.1.2-RELEASE-x64 (002022c)

I guess I'm not sure that there's actually anything wrong. Replication seems to be working as it should, if it's single "connectioned" as cyberjock points out.

@cyberjock
It certainly makes sense that a single dataset's replication tasks would run sequentially, but the circumstance that @mpfusion describes is what I have - multiple independently replicated PUSH side datasets (see screenshots). I should have specified that in my original post. Is there still no way to run tasks from different datasets simultaneously.

"PUSH" Datasets:

"PUSH" Replication Tasks: