[mythtv-users] strange real-time commercial flagging behavior

Peter Lee peteratcmu at gmail.com
Fri Jun 3 02:58:08 UTC 2005


I've been noticing some strange behavior with the (semi)real-time
commercial flagging.  I am running current cvs on both a
frontend/master-backend and on a slave backend.  However, only
recently (in the past week) have I turned on the real-time flagging. 
I noticed an issue right away, but it has taken me some time to narrow
down what has been going on.

Basically, on some recordings (about 1 in 10, I would say), the
real-time flagging for a program appears to "finish" well before the
program has actually finished recording.  At least this is what the
log says.  For example, here are some log entries from today, to
illustrate the phenomenon:

617  	scheduler  	2005-06-02 19:00:00  	localhost  	Finished recording
 	Finished recording Malcolm in the Middle on channel: 1015
615 	commflag 	2005-06-02 18:47:06 	mythslave 	Commercial Flagging
Finished 	Finished Commercial Flagging for "Malcolm in the Middle"
recorded from channel 1015 at Thu Jun 2 18:30:00 2005. Found 0
commercial break(s).
612 	commflag 	2005-06-02 18:30:32 	mythslave 	Commercial Flagging
Starting 	Starting Commercial Flagging for "Malcolm in the Middle"
recorded from channel 1015 at Thu Jun 2 18:30:00 2005.
611 	scheduler 	2005-06-02 18:30:02 	localhost 	Schedule Change
	Started recording "Malcolm in the Middle" on channel: 1015 on cardid:
1, sourceid 1

Like I said earlier, this doesn't happen all the time, only about once
in every 10 or so programs recorded.  When this happens, typically the
backend (master, in this case)  stops recording.  (The process doesn't
die, but sometimes shows "subsys locked" and fails to start any new
recordings.)  Furthermore, the recording itself is truncated, roughly
at around the spot that the commercial flagging "finishes".

In case it matters, the master server (on "localhost") is a much
faster machine than the slave server (on "mythslave").  Master is
AthlonXP and runs on FC2 and slave is on 933MHz PIII and runs on
Debian unstable.

I guess it would be easy enough to turn off the real-time flagging,
but this looks like a possible bug to me and so I'm happy to leave it
on and gather more info, if requested.


More information about the mythtv-users mailing list