[mythtv] Fwd: Re: Backend crashes almost daily

[Dave]

Flo, thank you for your help and perseverence. Please do provide any 
additional info and i will try and help track this down.

As mentioned, i'm still out of code reach currently and only really able 
to debug this around 25th October. Keep trying to narrow the prob down 
yourself if you can.

Luckily i'm also in europe (well not really, but i can get to the 
hotbird sats) so if you find a 13E sat channel that triggers this 
problem, please shout at me and i will get to the bottom of this.

Cheers
D
-------- Original Message --------
Subject: Re: [mythtv]  Backend crashes almost daily
Date: Tue, 9 Oct 2007 23:13:58 +0200
From: flo <flo at flobittner.de>
To: Dave <dave at 0bits.com>
CC: mythtv-dev at mythtv.org
References: <47081A14.6010006 at 0bits.com>

I'm not sure about anything :-)

It's only a backend, all frontends are seperate machines. I did NOT 
restart the whole machine, only gdb with mythbackend.

After the last crash i did a sql query like this: "update channel set 
useonairguide=1 where mplexid in (select distinct mplexid from channel 
where visible=1)", in words: i set useonairguide=1 for all channels on 
any transponder i use.  Before that, i set useonairguide=1 only on 
channels i used. That was the only change i made to the database. I 
thought the crawler perhaps doesn't like it if only some channels on a 
transponder accept eit data, so i changed it back.

And remember, i applied your fix in pespacket.h:

     {
         VERBOSE(VB_RECORD, "AddTSPacket: Out of sync!!! "
                 "Need to wait for next payloadStart" +
                 QString(" PID: 0x%1, continuity counter: %2 (expected 
%3).")
                 .arg(packet->PID(),0,16).arg(cc).arg(ccExp));
-->        return false;
     }

It's running for more then 48 hours now, EPG data is fully there.

I will report what happens the next days. If there's a crash i let you 
know If there's no crash until the weekend, i execute my old query 
(update channel set useonairguide=0 where visible=0) and test it.
After that i take out your fix and test it again. If the crash then 
doesn't come back i call ghostbusters to clean up my  home :-)
I hope the result of this tests gives us an idea what happens on my machine!

Thanks for your patience,
Flo


> Are you shure this is related ?
>
> Crashes in qt is frontend specific (i.e. the GUI) and should not happen
> in the backend ? I'm not sure this is the same issue. I'm away from my
> machine at the moment (am in las vegas at the mo) so can't check the
> code to verify, but this looks like a error in the signal monitor code
> as oposed to the previous eit scanner.
>
> Did you do a full restart of the front/back end ?
>
> Cheers
> D