[mythtv-users] Mythtranscode running on both FE/BE and FE only box !

[Michael T. Dean]

On 09/05/2007 08:21 PM, Steven Sartorius wrote:
> I have two Myth boxes -- one is a fast dual core FE/BE machine and  
> the other is an old, repurposed Shuttle FE only.  The FE/BE box is  
> the master backend where all recording takes places and the mysql  
> database lives.  The FE box mounts the recordings directory (which is  
> physically on the FE/BE) via NFS.  Two weeks ago I upgraded both  
> boxes to 20.2 (via svn update on the 20-fixes trunk) and noticed that  
> some of my mythtranscode jobs were taking a lot longer to run than  
> they used to.  After a little poking around I stumbled on the reason  
> -- some of the transcoding jobs are being run on the (slow) FE only  
> machine.  This is actually pretty cool -- it looks as if myth is  
> assigning the transcoding to whichever machine on the network is  
> least busy (not sure how that's determined).  If I had two machines  
> of equivalent speed I wouldn't care but in my case I'd like all the  
> transcoding to happen on the fast FE/BE box.  Anybody know of a way  
> to force this behavior?  Otherwise I'm at a bit of a loss...

In mythtv-setup (on each host) set, as appropriate:

Allow Commercial Detection jobs
Allow jobs of this type to run on this backend.

Allow Transcoding jobs
Allow jobs of this type to run on this backend.

(and similar settings for user jobs).

Run Jobs only on original recording backend
If set, jobs in the queue will be required  to run on the backend that
made the original recording.

Mike