New Site - HockeyStreams.com - Development Thread (1 Viewer)

[geoffstewart]

OK, will do... see if I can capture for a game this weekend...

On another topic, how are you able to test these URLs? You have a test harness you use? I'm surprised the tokens and such work for you after any length of time since these are live hockey games I'm working with.

The site has archived games too and I assume the same issue exists for that too.

Anyway, I'll post longer logs of firefox viewing of a game.

g

 

[georgius]

I just take your URL and put it into Firefox. I don't know how, but it works

 

[geoffstewart]

I captured 3 minutes of a live game just now... I don't see any HTTP 503...
I tried a few of the URLs in firefox, too. I can create a 503 by getting a fragment that doesn't exist yet... so, if the live stream is at 500 and I ask for 750, I get a HTTP 503.

Is it possible that OV is asking for fragments that do not yet exist on the server? Or is that the type of info that's encoded in the bootstrap responses?

Thanks.

Geoff

 

[georgius]

You're correct, bootstrap response have information about segments and fragments. But in your dump I found interesting information. It start on fragment 1429 and end on fragment 1476. But fragments 1450, 1454, 1462, 1465 and 1470 weren't requested. I think that these fragment are causing HTTP 503 response. But for identifying why aren't these fragments requested I need complete dump (best with Fiddler, because response and request can be found very easy, but it can be also in Wireshark format (pcap format)).

 

[geoffstewart]

OK,

I'll try to get onto the windows server tonight to get a fiddler capture.

Cheers.

Geoff

 

[geoffstewart]

OK,

Here's just the headers output from Fiddler.... the full capture is too big to upload.

I notice this capture is skipping fragments too... weird.

 

[georgius]

I hope that you still have full capture. I will need bootstrap responses for analysis.

 

[geoffstewart]

I do... it's 127 MB... I can try clipping out a few of the packets to reduce the size...

 

[geoffstewart]

Here's a part of the saz (zipped up) that shows a few skipped fragments... let me know if you want a different set.

 

[georgius]

I identified problem:

First fragment: 446, first fragment timestamp: 1780000, fragment duration: 0, discontinuity indicator: 1

I never saw this case, fragment with zero duration which should be skipped. I'll try to fix it and let you know when it will be finished.