Situation: an incomplete recording of a programme on BBC Three. The recording was started just before the changeover between CBBC and BBC3, and I use a recording margin of three minutes; the recording was started at 18:57.
I should mention that these two channels share the same VPID and APIDs, and here is a short interval in which nothing is being broadcast on those PIDs. I also have the channel update set for names only (altering this setting is not an option).
This usually does not cause a problem. This time, however...
The "video data stream broken" timeout was reached, resulting in an emergency exit. Of course, vdr can't be restarted immediately, and runvdr (mine) decided to sleep for five minutes because of too many restart attempts in too short a time... whoops, there go the first few minutes of the programme which I wanted to record...
Maybe I should put in a short delay in runvdr.c to make this less likely. Or I could increase MAXBROKENTIMEOUT. Or perhaps VDR should be fixed so as not to assume that no data == broken stream if it can receive *any* other data from the same multiplex, or maybe channels could be individually protected against updates...?
On Thursday 15 Jun 2006 00:03, Darren Salt wrote:
Situation: an incomplete recording of a programme on BBC Three. The recording was started just before the changeover between CBBC and BBC3, and I use a recording margin of three minutes; the recording was started at 18:57.
I should mention that these two channels share the same VPID and APIDs, and here is a short interval in which nothing is being broadcast on those PIDs. I also have the channel update set for names only (altering this setting is not an option).
This usually does not cause a problem. This time, however...
The "video data stream broken" timeout was reached, resulting in an emergency exit. Of course, vdr can't be restarted immediately, and runvdr (mine) decided to sleep for five minutes because of too many restart attempts in too short a time... whoops, there go the first few minutes of the programme which I wanted to record...
Maybe I should put in a short delay in runvdr.c to make this less likely. Or I could increase MAXBROKENTIMEOUT. Or perhaps VDR should be fixed so as not to assume that no data == broken stream if it can receive *any* other data from the same multiplex, or maybe channels could be individually protected against updates...?
I've had exactly the same problem with BBC-3 and BBC-4. My short-term fix is to manually remove the pre-padding of the recording, since the changeover time between channels is normally spot-on. However a proper fix would be appreciated.
Dave
vdr-bounces@linuxtv.org wrote:
Situation: an incomplete recording of a programme on BBC Three. The recording was started just before the changeover between CBBC and BBC3, and I use a recording margin of three minutes; the recording was started at 18:57. The "video data stream broken" timeout was reached, resulting in an emergency exit. Of course, vdr can't be restarted immediately, and runvdr (mine) decided to sleep for five minutes because of too many restart attempts in too short a time... whoops, there go the first few minutes of the programme which I wanted to record...
Maybe I should put in a short delay in runvdr.c to make this less likely. Or I could increase MAXBROKENTIMEOUT. Or perhaps VDR should be fixed so as not to assume that no data == broken stream if it can receive *any* other data from the same multiplex, or maybe channels could be individually protected against updates...?
I don't know how long the gap is, but have you tried to increase
#define MAXNONUSEFULDATA (10*1024*1024) in remux.c ?
This defines how much data with garbage is accepted before VDR tries to restart...
Just my 2 cents
/hgm.bg