[mythtv-users] Startup deadlock with slave backend ?

[Mark Cooke]

Hi all,

(Using CVS from a week ago.  Updating to current CVS when this evening's
recordings are done)

I have a problem restarting mythtv if a scheduled recording is supposed
to be in progress.  This isn't normally the case here, but we just had a
power-cut.

My master backend has a bttv, as tuner 2 (input connected to 'dummy').
My slave backend has a PVR250 (input to 'cable tv') and PVR350 (input to
aerial), fanless nfsroot from the master machine.

After a restart, the master backend comes up without a problem, and
returns status to mythweb, etc.

After starting the slave backend, when a recording is in-progress on
'Cable', the master backend deadlocks, and ceases to return status to
mythweb.  The frontend reports loss of communication to the backend.


Curiously, if I start the master backend, then 'strace -p' the master
backend process before starting the slave backend, the changed timing
caused by the strace seems to avoid the deadlock and the system starts
up normally, and the in-progress recording is stored to disk.

Anyone else seeing this ?  It seems to be a timing related issue with a
recording starting on a slave backend, so I'm guessing it is rare / non-
existant for the main developers.


There's nothing obvious in either the slave or master logs.  Ask if you
want copies to help diagnose this.


Cheers,

Mark

-- 
Mark Cooke <mpc at jts.homeip.net>