Help Required Glitches in overlapped recordings on the same channel

Hi,

Whenever I set two programs to record back to back, the second always has a brief glitch (video and audio) at the point where the first recording stops. There's never any problem with the first recording. Sometimes it's obvious, other times barely noticeable.

Visualised:


But recording from different channels on the same transponder is fine.

Anyone else have this with Octagon or other E2 boxes?

Octagon SF228
firmware tested:
Octagon-5.3-sf228-20161227_usb
openvix-5.1.028.release-sf228_usb
openatv-6.1-sf228-20180605_usb

same issue with both tuners.

intracube said:

Hi,

Whenever I set two programs to record back to back, the second always has a brief glitch (video and audio) at the point where the first recording stops. There's never any problem with the first recording. Sometimes it's obvious, other times barely noticeable.

Visualised:


But recording from different channels on the same transponder is fine.

Anyone else have this with Octagon or other E2 boxes?

Octagon SF228
firmware tested:
Octagon-5.3-sf228-20161227_usb
openvix-5.1.028.release-sf228_usb
openatv-6.1-sf228-20180605_usb

same issue with both tuners.

Click to expand...

You have got it set to over run by 2 mins at the start and 5 mins at the end.

You will need to adjust the timer settings.

Chris1979 said:

You have got it set to over run by 2 mins at the start and 5 mins at the end.

You will need to adjust the timer settings.

Click to expand...

I need the margins though.

So is this a known issue with E2 boxes?

intracube said:

I need the margins though.

So is this a known issue with E2 boxes?

Click to expand...

No it’s not an issue it’s just set like that as default.

You misunderstand, the problem isn't the overlap/margins, but with the glitch on the video/audio streams (pixelation/blocks and a chirp on the audio) which coincides with the end of the first timer recording. Sorry, I should've been more descriptive on the fault.

Here's what it looks like:
(bottom of the screen, about 6 seconds in)

The audio could be worked on with some filtering/processing, the video is however beyond redemption.

Channel Hopper said:

The audio could be worked on with some filtering/processing, the video is however beyond redemption.

Click to expand...

The recording isn't important, but I'd like to try and figure this out so it doesn't glitch on something that matters.

Right now I'm guessing this is probably a bug that only shows up on the SF228.

E2 debug logging is enabled. No warnings or errors listed there.

I'll keep digging around.

Edit: and in case it wasn't clear, this happens 100% consistently. At exactly the same point in overlap recordings, every time.

Of course it will, the receiver is doing what it can toe simultaneously work with two individual , separate pieces of content, and store the on a single device.

What are you recording to btw ? Is it via a humble USB cable or SATA lead ?

Have you tried recording two radio stations only with a similar overlap ?

Channel Hopper said:

Of course it will, the receiver is doing what it can toe simultaneously work with two individual , separate pieces of content, and store the on a single device.

What are you recording to btw ? Is it via a humble USB cable or SATA lead ?

Click to expand...

A Samsung USB SSD.

It doesn't seem to be CPU or I/O related, since stress-testing by doing two simultaneous downloads while recording a live channel doesn't cause any glitches in the live recording.

During the above test:
enigma2 process maxes out at 100% CPU on one of the two cores
each download is around 4.7MB/sec (37Mbps), so nearly 80Mbps combined
in addition, another ~8Mbps for the live recording

So if the USB and SSD can cope with all that but still fails on a radio recording, it looks like a software issue. Another factor that pushes it in this direction is the glitch happens in a completely predictable way. If it was random glitches, then it could point to CPU or data throughput bottlenecks.

Or it could be some oddity with the hardware tuner/demuxer side of things. But if anything, recording from a single channel, eg BBC One, should be less demanding on the hardware than two different channels on the same multiplex - since the latter will have to demux twice the number of elementary streams (video, audio, subtitle, teletext, etc).

Channel Hopper said:

Have you tried recording two radio stations only with a similar overlap ?

Click to expand...

Yep, same issue. Sometimes it's a loud squeek, sometimes an almost silent glitch - it just depends which bits are being corrupted or missing. But Ffmpeg always reports a discontinuity in the stream.

What is the eventual recording stored as ?

Channel Hopper said:

What is the eventual recording stored as ?

Click to expand...

On the SF228 HDD? The usual .ts format.

Edit:
If you mean the output of ffmpeg, in actuality I'm using 'ffplay' - which just decodes and plays the files. No re-encoding to another format.
The reported errors are on the decode.

and in case I didn't already mention, the glitches are visible on the SF228, not only on my PC.

whose .ts format though ?

Chinese ?

Channel Hopper said:

whose .ts format though ?

Chinese ?

Click to expand...

I don't know what you're asking. The files are standard MPEG-TS format.

There is no standard mpeg-ts format.

The TS format is whatever format the SF228/Enigma2/OpenVIX 5.1 writes out to disk. Codec details will be different depending on the channel.

TS is transport stream, which can contain more than one er...stream. The codec depends on what is transmitted by the transponder since you're not transcoding it.

I would send a message to Octagon on their Farcebook page - they seem pretty responsive. It's certainly not an E2 bug as I don't have this on my Mutant.

Captain Jack said:

TS is transport stream, which can contain more than one er...stream. The codec depends on what is transmitted by the transponder since you're not transcoding it.

Click to expand...

Yep, each of the video/audio/sub/teletext are 'elementary streams' within the TS. But since this happens no matter the codec - (SD/MPEG2), (HD/AVC H.264) and radio (MP2 audio) it all sort of points to a lower level issue.

Captain Jack said:

I would send a message to Octagon on their Farcebook page - they seem pretty responsive. It's certainly not an E2 bug as I don't have this on my Mutant.

Click to expand...

Heh, I don't do Facebook but I'll give their forums a go.

I've also been testing a Vuplus Duo2 and that doesn't show any signs of this either.