TSWriter deadlock potential fix. (1 Viewer)

[SciDoctor]

Just had a failre during continued tesing as a subchannel is decomposed at the end of a schedule and this casued a BSOD reference the tuner driver.

Logs attached look for 'ch 4+1' and schedule '7th cavalry' as the offending schedule.

I have looked through the logs, I can't see anything, the failure occured at 15:43 when the schedule on 'ch4+1' '7th cavalry' was ending and being decomposed the BSOD occured to quickly for logs to be writen I suppose.

 

[lulsam]

I don't want to hijack this thread, but maybe this isssue is related to this one: my tv server (1.2.3) freezes as well with a very similar behaviour as the one described here. I managed to determine that the cause was the epg grabber but only if I selected a specific frequency (I had to test chananel by channel). In my case (Spanish DVB-T in Madrid) whatever channel selected from the 708 MHz mux drives to a TV Server freeze in not to much time. If this channels are unselected from the epg grabber but all others (more than 50) are selected TV server works correclty without any particular issue for several days.

So my question is: Could non existent epg or malformed data in a mux drives to a TV Server freeze like the described here? The new TSWritter and TSReader files could help to solve this issue as well if unrealted to the focus of this thread?

Thx in advance.

Try the files from the first page and let me know if they work.

Are they intended only for 1.3.0 Beta or could work with 1.2.3 (my current installation)? The question is because I have a TV Server with five MePo clients installed serving to my family so I cannot afford to upgrade to a beta just for testing (they will kill me).

If these files just work with 1.3.0 Beta I will contact with some other spanish users and ask for help testing results with epg enable in the problematic mux to check if the issue is out / improves.

Let me know

 

[ramu]

Hi,

here are new log-files. In between I tested V7 and V2 togehter. Tvserver crashed several times...

 

[DJBlu]

I'm beginning to think this may be deeper than I first thought.

I am going to have to take a look into the TV Library code also as I believe the subchannel dictionary isn't thread safe.

 

[DJBlu]

@SciDoctor @ramu

Uploaded v8 to first page.

Please can you give it a whirl. Thanks.

 

[radical]

I've made a Stresstest with TsWriter V8 and Filewriter v2.
I saw many SQL Errors, but MySQL is working flawlessly with Live TV. I've Seen no problems there.
So i think that something other is wrong, that hasen't anything to do with this bug and your work.

 

[Vasilich]

@radical
"DB could not be reached" - any firewall or antivirus is active? or something in background takes all CPU so DB server can't answer?
you should clear your DB issues first and then continue testing

 

[radical]

Thank you for the hint.
But I'm a little confused. No problems with regular TV-Service. Just on stresstesting.
I am sure that the database is working without any error. It is allready testet.
No Antivirus running.
Firewall is only from Windows and how there can be a firewall conflicts on a local maschine? Maybe someone could point me where i should search?

 

[DJBlu]

This isn't the thread to ask I'm affraid.

Please report it in the 1.3.0 Beta bug forum.

 

[SciDoctor]

Testing has been good the most stable so far. But a problem has occured logs attached.

Normal use had been followed by stress test this morning and one oddity is that the tuner signal dies on the substreams of the last channel left on the mux after other substreams have ended.

'Masters snooker' only recorded for 40mins of the 2hr length coinciding with the ending of another substream

'Poirot' the same.

If I know use setuptv to start a substream on these tuners the tvservice 'No PMT found' showing the drivers have crashed.