I have a two-machine system; a PVR-350 is on machine A, and some 250's are on machine B. I've just installed both from R5A22.

When I tried a test installation a couple days ago, in a one-machine configuration, running SnapperDragon's PVR scripts worked perfectly: I wound up with a machine that could display video and audio from the TV out of the 350.

But I just tried running the script on this two-machine configuration, and while I see video, I have no audio. I'm looking for places to check or ways to debug this.

What I'm doing: I'm trying "Watch TV", which is using one of the PVR-250's in machine B and sending the video to the PVR-350 in machine A to be watched.

I know that audio is physically connected; for example, if I boot machine A, the TV goes "thump" at one point during boot. Should I have the "internal audio mixer" selected in setup? It started out that way; I turned it off & nothing changed.

I tried connecting the TV's audio-in directly to the motherboard's soundcard out. No audio in live TV---but I wouldn't have expected that to work, anyway, since I believe that if the 350 is displaying video, audio will be routed only to its audio outputs.

What should the audio output device be in the setup program? Right now, it's /dev/dsp, which seems a little suspicious, because I think I -should- be using the audio output of the 350. Certainly none of the other devices seem to have the PVR's name in 'em, but I don't even know if this field is just ignored after running the PVR350 script.

Unfortunately, since audio worked last time I tried this (in the single-machine configuration), I didn't even think to try seeing if audio worked -to my onboard motherboard's audio- in the two-machine state before running the PVR script, so I don't have a baseline of knowing if audio was working in the first place.

I tried looking through all the logs in /var/log for anything suspicious, but didn't see anything. I also tried powercycling machine A (the one with the 350), just in case the script and the 2-3 boots it requires left the hardware in some weird state; that didn't change anything.

Searching for "no audio" in the forums doesn't work well. Searching for ``mythtv "no audio" pvr-350'' yields 1330 hits; the first 20 or haven't been helpful.

[Would it be a useful debugging datapoint to boot machine A as a frontend from the CD and see if I can get audio out of the motherboard's soundcard when trying to watch live TV using a tuner in the backend? This is such a different audio path that I wonder if it would tell us anything---unless it also fails, in which case maybe it would implicate the backend as not -sending- any audio, but I'm not even sure if that's possible.]

Ideas?

Thanks!

Aha! Problem resolved!

I'd noticed the following error message when I ran the pvr350 script (and pm'ed SnapperDragon asking if I should worry, but probably after he went to sleep :)



So I went and took a look at what the script was doing, and it looks like it was trying to send this command to the database:



It's not clear to me why it failed in the two-machine case, but it looks like that's where the bug is (and hopefully can be corrected in a newer version of the script!), 'cause i tried this on machine B (the backend, which has the SQL database---and -not- on machine A, which has the 350 and is where I ran the pvr350 script):



...and then, when I booted machine A, Watch TV had audio. (I might have been able to just restart mythfrontend, but it was surer and faster to boot, since I'd been poking at other things at the same time.)

It's -really- unclear to me how the -video- wound up on the 350's TVout if the database wasn't updated; is it the case that the ivitv & XF86Config-4 mods take care of the video, and the OutputEnable stuff in the database is audio-specific? Where is this sort of thing documented, short of reading all the code? :)

Anyway, in case anybody else is getting screwed by PVR-350's and multiple-machine configurations, I hope this is helpful. And if SnapperDragon can figure out why the script couldn't update the backend database, hopefully we can avoid this in the future.