Project

General

Profile

Continuity Errors cause stuttering stream

Added by red nose over 9 years ago

I am running tvheadend 3.4.patch1.01 on an openwrt-router connected with a dvb-c-stick. I get a picture but the stream is too slow. CPU is barely used during playback but sometimes swap-usage goes up. However, I get countless of continuity-errors during playback. No matter if I listen to radio or try to watch TV. I tried some advices from other postst, e.g. [https://tvheadend.org/boards/14/topics/8626?r=8751 [here]], reading turning off "full mux reception" would do the trick. Though it did reduce the stuttering, the streams is still buggy and not fluent.

[https://tvheadend.org/boards/5/topics/8502?r=8790 [This threat]] recommends turning off "coll 'n quiet". Since I'm running tvheadend on a router, this is not an option for me. However, DVB-C doesn't need CPU at all since the mpeg-stream just gets streamed over network. All work is done by the stick. Monitoring the cpu-load clearly shows: During playback the cpu is not stressed at all. It goes up a few percent - that's all.

Another threat was reading something about antennas and disturbing influences. Since DVB-C ist working with a cable plugged into the stick, I think this hint is for DVB-T-users.

There were several severe bugs in previous versions of tvheadend for openwrt. E.g. 3.2 missing parts of webfrontend, 3.4 (the official version from open-wrt-repository) doesn't offer a dvb-tab in the webfrontend at all and some other version was not be able to scan for channels. So I really tried 3 different versions until I compiled version 3.4patch1.0.1 manually. My question now is: Is there still a bug in my manually compiled tvheadend-version? I found [https://tvheadend.org/issues/1537 [this]] but don't know if it is related to my problem. The memory-usage was always high when trying to watch TV. For some reason it was not anymore, the last time I tried watching TV. The stream was lagging nevertheless.
[https://tvheadend.org/issues/1713 [This]] threat is similar but I don't know which version they used to solve the problem. They just don't write it.


    (1-1/1)