[mythtv] RFE: Allowing backend to trigger S3 sleep while frontend is connected and idle

Michael T. Dean mtdean at thirdcontact.com
Sat Oct 1 20:08:04 UTC 2011


On 09/26/2011 04:48 PM, Dan Wilga wrote:
> On 9/26/11 4:32 PM, Kenni Lund wrote:
>> 2011/9/26 Warpme:
>>> Hi,
>>> I'm glad S3 aspect pops-up as I want to add another aspect relevant to S3:
>>> sleep-resume on fe side.
>>> Currently, basing on BE logs, I get impression BE threats frontend sent to
>>> S3 as fe which lost connection to BE.
>>> I'm not verified it, but in past, long enough sleep period caused also lost
>>> of comm. mechanism on BE->FE direction.
>> A couple of months ago I finally got S3 configured on my main
>> frontend. I experienced the same thing - when waking up the frontend
>> after a few hours of the frontend being suspended, the connection
>> between the frontend and the backend was borked somehow - not enough
>> for it to give an error message in the frontend, but enough to give
>> funny behaviour in the frontend (missing recordings, etc). I didn't
>> have time to look into it (or even create a ticket or check for old
>> tickets) at that time, so a quick workaround to keep a high WAF, was
>> to restart the frontend every time the system was resumed. I haven't
>> had time to look any further into it yet, but the problem was
>> consistent on my 0.24-fixes system (up to date as of one month ago).
> Ditto. If somebody were to figure out the reason for this and provide a
> patch, I would be an eager tester. I currently use this feature on three
> different frontends, and having to restart MFE every time introduces
> enough of a delay to be distracting.

Since mythfrontend doesn't support resume from suspend, you should be 
shutting down mythfrontend in your suspend script and restarting 
mythfrontend in your resume script, at least until you or someone writes 
code to re-connect all MythTV protocol and database connections and 
reconfigure the context, properly, after a suspend (or, really, in the 
more general case when required).

Shutdown/restart is the 5s-of-effort fix.  Redesigning, rewriting, and 
testing the code, er, isn't.

Mike


More information about the mythtv-dev mailing list