Tvheadend

Hi Hans-Gerhard,

the current implementation of unicable is not fully stable.
I actual looking for the problem. But i do not find it, since one month or so.

Can you test this branch please: https://github.com/InuSasha/tvheadend/tree/fixes/stablize_en50494
There is a little improvment, i hope it helps a little bit.

And can you post same debug log (diseqc, linuxdvb and en50494), please.
Add "--logfile tvh_debug.log --debug diseqc,en50494,linuxdvb" to you tvheadend commandline

@All: everyone who can help me to stablize unicable is willcome. some ideas for looking for, can be enough.

regards,
Sascha

Hello Sasha,

I did a bit testing of your branch and the actual unstable Version 1992.

The log is attached.

What I discovered is, that adapter0 always stops at 32 "Uncorrected Blocks" [WFE: Status-->Stream].
This is independent from the SCR/Unicable frequency (I didn't include this in the log, but I tested it...) and for about 95% safety (I didn't test it again, but I'm pretty sure to remember it right ...) independent from the cabling, as I rotated them over the 3 connected devices (2x Cine S2, 1x Technisat-Receiver @ SCR:3/Freq. 1632).
If I switch off adapter0 (which is configured on SCR:0 / Freq.: 1284) it seems, that all muxes are correctly discovered/scanned by adapter1 (this test is acutally still running).

So my assumption is, that while rotating over the muxes each mux sometimes is scanned by adapter1 and found good and then scanned by adapter0 and found bad. There must be a bug that only affects adapter0 - maybe in a loop or similar. If someone has more than 2 adapters he could approve this ...

BTW: I'm using a Technirouter 5/2x4G, which seems to work right, as 2 EN50494-capable receivers work as expected.

I hope that helps (me too .... ;-)

HG

Next three discoveries:

1. Mux scan with only adapter1 active has now finished. There're only 5 muxes left which failed discovery (see tvh-3.jpg)

2. I found some duplicates in the muxes! They all had the same Name(=Freq.) and Transport Stream ID. One of them had more services discovered than the other. Always the one with the smaller number of Services failed the scan. Actually I deleted them via the WFE.

3. I also have some duplicates in the service list (see attached file). There's always a copy of the service on mux 12603.75 - which is wrong! - with identical Service IDs. (see tvh-2.jpg).
I' going to delete them, too.

There must also be a wrong assignment during scan - maybe that's another symptom of the above mentioned "adapter0-problem".

HG

Finally (for today ..):

According to http://www.satindex.de/frequenz/12604/ there should be 11 Streams on Mux 12603.75

After I deleted the "stowaways" and re-scanned the mux they all came back! So there's also a problem with adapter1 !!

HG

So after one day of letting the system run for itself the duplicate muxes are back (examples: see attached screenshot). Unfortunately I switched off logging so in the next 24h I try to confirm this with enabled logging.

Interesting: all the duplicate muxes have either an even TS-Id & a vertical polarisation (1070/1080/1084/1086/1088/1092/1094/1102) or the opposite (1119+horizontal) - but this could also be a coincidence.

HG

So, here we are with the log of the suddenly appearing double-muxes.

Fortunately this can be completely reproduced - even the doubled muxes are identical ...

@Sascha/all: are there additional logs I could provide?

Hans-Gerhard

Sorry,
i had updated my system, and actually same problems.
So i can't look into logs, now. I hope next weekend, i have time for it.

Sascha

No Problem.

I changed the cabling and the SCR-Freq. again with quite the same result as described above:

• adapter1 works as expected (I'm just waiting for the double-muxes...)
• Sat-Receiver (Technisat) works as expected
• adapter0 has the only change: it doesn't stop at 32 Uncorrected Blocks - sometimes there're >40000(!).

• Either I have a poor SNR - then it stops at 32 Uncorrected Blocks
• Or I have a better SNR than adapter1 - and Uncorrected Blocks tend to infinity ...

BTW: Shall I make an official bug out of this? Which module(s) are affected? If someone (Sascha?) tells me I also could have a look. Although I'm not a C-programmer - only VBA (pls don't blame me for that :-) - I could "read" the source code and try to understand. I once helped a colleague in his debugging efforts...

Hans-Gerhard

@Sascha: This is actually also checked under Bug #2474 !

I don't know, if it the same.
But we will see.

I have looked in your log,
but i have not seen new perception.

Grüße,
Sascha