[mythtv-users] MythTV 0.22 Upgrade mythwelcome problem

[Jason But]

On 04/10/10 03:19, nospam312 wrote:
>> I have recently upgraded my MythTV FE/BE from 0.21 to 0.22. After the
>> upgrade everything is working properly except for one thing. My FE/BE is no
>> longer automatically waking up to record a program. Strangely, it still
>> wakes up automatically for the daily startup time setting in mythwelcome.
>> However it fails to start for any scheduled recording
>> This was all working perfectly prior to the 0.22 upgrade
> 
> Also worth checking out http://www.mythtv.org/wiki/ACPI_Wakeup
> 
> I updated to a new Mythbuntu and I have the problem where the system
> is not waking up - but it does not wake up even using the debug
> commands.
> 
> Unfortunately it looks like the new Mythbuntu v10.04/Ubuntu may still
> not have a high enough kernel to fix the problem being experienced
> with these motherboards.
> 
> Quoted From URL above:
> "Note that kernel version 2.6.26, included in the current Debian
> stable distribution, contains a bug that makes ACPI wakeup fail on at
> least some motherboards. If this affects you, either upgrade to
> 2.6.33-rc4 or patch 2.6.26 with this patch"
> _______________________________________________
> mythtv-users mailing list
> mythtv-users at mythtv.org
> http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
> 

OK, long response follows in case anybody else is searching the lists with a
similar problem.

I checked my clock settings, I run all my systems on UTC time and everything
was correct. I run my own NTP server and my myth box was good so no problem
there.

Then I looked at adding extra debugging to the mythshutdown calls so I added
a (-v most) to the commands in my myth setup and tried again, I also turned
off the regular wakeups for testing purposes.

Now I saw something strange, if I scheduled a recording for the same day as
the day of my test I got

2010-04-10 17:35:24.107 recording scheduled at: 2010-04-10T00:00:00
2010-04-10 17:35:24.117 Scheduled recording time has already passed.
Schedule deleted

If I scheduled a recording for the next day I got

2010-04-10 20:26:42.523 recording scheduled at: 2010-04-11T00:00:00
2010-04-10 20:26:42.531 will wake up at next scheduled program

So it appeared that mythshutdown was changing all the time fields from the
correct time to 00:00:00 while getting the date fields correct.

Then I looked at the mythshutdown source code and noticed nothing
particularly strange. I then checked the contents of the database where the
time is stored and saw that the correct time/date was there (not 00:00:00)

So it appeared that the time was reset to 00:00:00 _after_ mythshutdown
extracted the value from the DB

Finally I did some more googling and eventually stumbled on something, the
wakeup-time format in mythtv-setup needed to be "yyyy-MM-ddThh:mm:ss" but I
had "yyyy-MM-ddThh:mm". I appended the ":ss" and lo and behold, all works
well in the world.

This really should be documented better, it appears that "yyyy-MM-ddThh:mm"
was fine for v0.21 but v0.22 needs the ":ss" or it won't work.

I mentioned earlier that I had not upgraded much but one thing that also
upgraded was the QT libs from v3 to v4. I looked again at the mythshutdown
source and noticed that all time conversions (string<->bin) were managed by
the QT libs. I assume that the QT upgrade suddenly treated time strings that
were missing a "seconds" field to be reset to 00:00:00. A bit of a nasty
gotcha here

So to summarise. People upgrading fro mythtv 0.21 to 0.22 AND using ACPI
auto-wakeup to record stuff NEED to check the "Wakeup time format" field in
the mythtv backend configuration and ensure that seconds are appended to the
end.

On a side issue, I have a recent (2years old) MB on this system and am
running 2.6.30 kernel. I have been unable to get the RTC wakeup to work and
had to resort to the /proc/acpi/alarm as the only means of waking up my
system. I know that /proc/acpi/alram is supposed to be deprecated but
hopefully it will stay in there

Jason