Occasional bad fragments from f4fpackager?

Report · Dec 05, 2012

Hello

I am experiencing occasional odd behavior on packaged content. Sometimes, when playing a stream, the player completely stalls on a fragment. Manually skipping forward a minute solves the stall and the player continues playing the stream.

I see this behavior in several different players (pure OSMF Strobe, as well as mpx thePlatform's player) across multiple OS and Flash 10.x and 11.x runtimes.

If I take the same input .mp4 files (I use multiple bitrates) and repackage, using the exact same f4fpackager command, I often get an output that is playable. I have tried with v4.5.1 and v5 of f4fpackager and see the same behavior. There are no errors or warnings thrown during the packaging process.

This problem of stalls is a major problem for a VOD service as I now have to manually look through each bitrate and check that it is playable without stalls.

I saved two .f4f files: one that fails and one that works. Not knowing exactly the format of f4f (I cannot find a parser) I post my findings here. Perhaps it triggers someone's knowledge. I can only hope. Let me know if more info is needed.

The files are of identical size:

$ ls -l correct_Seg1.f4f failed_Seg1.f4f

-rw-r--r-- 1 hans hans 91378076 Dec 5 09:20 correct_Seg1.f4f

-rw-r--r-- 1 hans hans 91378076 Dec 4 18:07 failed_Seg1.f4f

But they don't contain the same content:

$ diff correct_Seg1.f4f failed_Seg1.f4f

Binary files correct_Seg1.f4f and failed_Seg1.f4f differ

$ cmp -l -b correct_Seg1.f4f failed_Seg1.f4f

55 346 M-f 266 M-6

56 130 X 7 ^G

63 346 M-f 266 M-6

64 130 X 7 ^G

183 346 M-f 266 M-6

184 130 X 7 ^G

191 346 M-f 266 M-6

192 130 X 7 ^G

295 346 M-f 266 M-6

296 130 X 7 ^G

303 346 M-f 266 M-6

304 130 X 7 ^G

40757 346 M-f 266 M-6

40758 130 X 7 ^G

40765 346 M-f 266 M-6

40766 130 X 7 ^G

40869 346 M-f 266 M-6

40870 130 X 7 ^G

40877 346 M-f 266 M-6

40878 130 X 7 ^G

81280 346 M-f 266 M-6

81281 130 X 7 ^G

81288 346 M-f 266 M-6

81289 130 X 7 ^G

81416 346 M-f 266 M-6

81417 130 X 7 ^G

81424 346 M-f 266 M-6

81425 130 X 7 ^G

A more visual diff using emacs hexl-mode:

Correct file, notice the "X" (hex cce3 e658)

00000000: 0000 001c 6674 7970 6634 7620 0000 0001 ....ftypf4v ....

00000010: 6973 6f6d 6d70 3432 6d34 7620 0001 e007 isommp42m4v ....

00000020: 6d6f 6f76 0000 0078 6d76 6864 0100 0000 moov...xmvhd....

00000030: 0000 0000 cce3 e658 0000 0000 cce3 e658 .......X.......X

00000040: 0000 03e8 0000 0000 0002 45eb 0001 0000 ..........E.....

00000050: 0100 0000 0000 0000 0000 0000 0001 0000 ................

00000060: 0000 0000 0000 0000 0000 0000 0001 0000 ................

00000070: 0000 0000 0000 0000 0000 0000 4000 0000 ............@...

00000080: 0000 0000 0000 0000 0000 0000 0000 0000 ................

00000090: 0000 0000 0000 0000 0000 0004 0000 9e7e ...............~

000000a0: 7472 616b 0000 0068 746b 6864 0100 000f trak...htkhd....

000000b0: 0000 0000 cce3 e658 0000 0000 cce3 e658 .......X.......X

000000c0: 0000 0001 0000 0000 0000 0000 0002 45b9 ..............E.

000000d0: 0000 0000 0000 0000 0000 0000 0100 0000 ................

000000e0: 0001 0000 0000 0000 0000 0000 0000 0000 ................

000000f0: 0001 0000 0000 0000 0000 0000 0000 0000 ................

00000100: 4000 0000 0400 0000 0240 0000 0000 01a2 @........@......

00000110: 6d64 6961 0000 002c 6d64 6864 0100 0000 mdia...,mdhd....

00000120: 0000 0000 cce3 e658 0000 0000 cce3 e658 .......X.......X

00000130: 0000 03e8 0000 0000 0002 45b9 15c7 0000 ..........E.....

00000140: 0000 002c 6864 6c72 0000 0000 0000 0000 ...,hdlr........

00000150: 7669 6465 0000 0000 0000 0000 0000 0000 vide............

...

Incorrect failed file, notice absence of "X" (hex cce3 b607)

00000000: 0000 001c 6674 7970 6634 7620 0000 0001 ....ftypf4v ....

00000010: 6973 6f6d 6d70 3432 6d34 7620 0001 e007 isommp42m4v ....

00000020: 6d6f 6f76 0000 0078 6d76 6864 0100 0000 moov...xmvhd....

00000030: 0000 0000 cce3 b607 0000 0000 cce3 b607 ................

00000040: 0000 03e8 0000 0000 0002 45eb 0001 0000 ..........E.....

00000050: 0100 0000 0000 0000 0000 0000 0001 0000 ................

00000060: 0000 0000 0000 0000 0000 0000 0001 0000 ................

00000070: 0000 0000 0000 0000 0000 0000 4000 0000 ............@...

00000080: 0000 0000 0000 0000 0000 0000 0000 0000 ................

00000090: 0000 0000 0000 0000 0000 0004 0000 9e7e ...............~

000000a0: 7472 616b 0000 0068 746b 6864 0100 000f trak...htkhd....

000000b0: 0000 0000 cce3 b607 0000 0000 cce3 b607 ................

000000c0: 0000 0001 0000 0000 0000 0000 0002 45b9 ..............E.

000000d0: 0000 0000 0000 0000 0000 0000 0100 0000 ................

000000e0: 0001 0000 0000 0000 0000 0000 0000 0000 ................

000000f0: 0001 0000 0000 0000 0000 0000 0000 0000 ................

00000100: 4000 0000 0400 0000 0240 0000 0000 01a2 @........@......

00000110: 6d64 6961 0000 002c 6d64 6864 0100 0000 mdia...,mdhd....

00000120: 0000 0000 cce3 b607 0000 0000 cce3 b607 ................

00000130: 0000 03e8 0000 0000 0002 45b9 15c7 0000 ..........E.....

00000140: 0000 002c 6864 6c72 0000 0000 0000 0000 ...,hdlr........

00000150: 7669 6465 0000 0000 0000 0000 0000 0000 vide............

...

Any insights most welcome.

Thanks!

Hans

Report · Dec 05, 2012

How are you generating VOD MP4 files at first place? Which encoder do you use? Are you saying with same mp4 file - f4fpackager gives different results at different times?

Report · Dec 06, 2012

Hello SE_0208, thanks for responding.

The mp4s are created using ffmpeg and play nicely on their own.

And, yes, the exact same mp4 input, as well as f4fpackager executable and command, yield different results!

-Hans

Report · Dec 06, 2012

Is there a way you can host this file from where i can download it - say some ftp location or public webserver. Because its little strange that you will get different results when using same file and same executable.