[mythtv] Multiple tuner conflict resolution bug or feature?

Bruce Markey bjm at lvcm.com
Sat Jan 17 22:31:26 EST 2004


Robert Kulagowski wrote:
> One thing that isn't handled right now is the following situation:
> 
> master and slave, each with two encoders, master having #1 and 2, slave 
> having 3 and 4.
> 
> at time "t", three programs begin recording, using encoders 1, 2 and 3. 
>  However, encoder 3 is a PVR-250, and the firmware locks up, leading to 
> no recording starting.  A keen eye notices this, CTRL-C's the slave 
> mythbackend, modprobe -r ivtv, modprobe ivtv then restarts the 
> mythbackend within the time period of recording #3.  The master backend 
> sees the slave going away and coming back.
> 
> Recording #3 does _not_ restart once the slave mythbackend re-registers 
> with the master.  Also, the master backend gets confused as to which 
> encoders are still available, leading to Bad Things Happening.
> 
> However, in the situation where there is only one or two recordings for 
> time "t" (and are therefore allocated to encoders 1 and 2 on the 
> master), if the masterbackend has the same firmware lockup and goes 
> through the same restart sequence detailed above, jobs #1 and #2 _will_ 
> get restarted correctly.

I feel your pain. Put another way, when the master restarts
it tries to restart all current recordings, however when a
slave reconnects there is no attempt to restart recordings.

Another interesting variation is if there is a recording on
the slave that continues unperturbed while the master restarts,
the master reports that it tried to start the recording for
the slave but the tuner was already busy recording. This is
fine because it was recording the right thing all along.

I think that when a slave reconnects, the master should check
to see if there are any current recordings assigned to cards
on that slave and attempt to restart them.

The (ugly) workaround for now is after restarting the slave,
restart the master. This will get all the recordings going
again on the right cards but if there were any recordings in
progress on the master, they would be split in two parts.

--  bjm


More information about the mythtv-dev mailing list