retry code doesn't quite work, but might be close...

[johnsonx42]

I see you did add some code that tries to recover from an NDigitalHost / graph stop hang, but it doesn't quite work. (also, kudos on the reduction of log spam during EPG update... one NRecord log file now covers nearly 4 days!)

in the attached logs, if you look at the end of NDigitalHost-1524.log, you'll see that I watched livetv twice on channel 32.2 "MeTV" (the corresponding NRecord activity is near the end of NRecord.log.2). It worked ok the first time, but the second time I just got a blank screen so I exited; this resulted in "2013-04-10 13:05:39.235 [DEBUG] [21] Async stop taking way too long". It tried to exit, but hung like it always does.

Then at 15:30 it got the request from NRecord (NRecord.log.1) to record Zula Patrol, and got stuck trying to start the graph. At this point, NRecord sat there doing nothing all day long, missing every recording even though most or all were on other tuners (1524 is my lowest priority tuner, mostly just handles a few channels I get only by antenna), until 21:10 when I realized it had crashed; from past experience I knew all I could was reboot, so I did.

Suddenly on the way down, NRecord woke up, noticed the failure, and started trying to recover from it 6 hours too late. It actually managed to record 5 seconds or so of whatever was on that channel as the system was shutting down.

I'm assuming this retry code is new, since I've never seen NRecord do that before.

I'm hoping maybe seeing how it didn't work will provide clues to make it work right.