[mythtv-users] 'mythtv is idle' and wont auto shutdown/suspend

Michael T. Dean mtdean at thirdcontact.com
Sun Aug 7 15:49:34 UTC 2011


On 08/07/2011 10:16 AM, Brian J. Murrell wrote:
> On 11-08-07 08:40 AM, Michael T. Dean wrote:
>> The timeout is for when a frontend crashes instead of shutting down
>> properly.  When shut down properly, the backend sees the frontend
>> disconnect, and knows that it has no clients connected, so is allowed to
>> shut down immediately.
> Right.  But that doesn't cover the case where the FE just goes to sleep
> due to idleness.
>
>> This sounds more like a system socket issue or
>> network issue than like a MythTV bug to me.
> I don't think it's even that complicated.  It's simply a case of a FE
> going to sleep, which, indeed would look to the backend just as if it
> crashed.

OK, let me put it this way.  We only support frontends going idle by way 
of disconnecting gracefully.  If you just rip the frontend out from 
under the backend, we never know--unless your system/network/... 
disconnects the connection the backend sees.  As long as the backend 
sees a connection from that frontend, it assumes that frontend is connected.

> I think this is simply a case of the timeout being too long and/or not
> enough is being done to determine real idleness.  A non-idle FE should
> be very easy to determine.  A FE where somebody has gotten up and walked
> away, less so, but a multi-hour timeout for that seems excessive.
>

Yes, because the frontend does not have a concept of idleness.  That's 
what mythwelcome is for.  Eventually, we may support the frontend's 
going idle, but we don't now.  I think Raymond Wagner plans to do some 
work on this with a protocol rewrite he's planning for MythTV apps.

So either fix your sleep script to properly exit (and the start script 
to properly restart) mythfrontend or start using mythwelcome or write 
the code that adds a concept of idleness to the frontend.  :)

Mike


More information about the mythtv-users mailing list