[mythtv-users] Recordings off 1 Hour

[Roger Heflin]

Grant Swenson wrote:
> Weird, I just saw this two. My master backend was fine, but my slave
> backend was an hour behind. No idea why. I fixed it with an "rdate -s"
> to an rdate server.
> Output from that rpm command is the same on both: tzdata-2007k-2.fc8
> (Yep, F8, upgraded from RH9...so I had FC6 on there at one time.)
> For the record, NTP was not running, but starting it or running it
> manually didn't seem to help (I think the time difference was too much
> for NTP to correct).
> I lost a recording tonight...but it wasn't a big deal. (It's actually
> pretty interesting when one system starts the recording, and the other
> ends it...one system thought the recording was 3 minutes long, but it
> was actually 63 minutes.) It would be an interesting feature request
> to let MythTV handle backends in multiple timezones.


When a machine is shutdown properly it syncs the current time with the bios 
clock.   If a machine crashes (after Daylight savings time has happened-without 
a proper shutdown this of course does not happen) then when the machine comes 
back up it is an hour off and too far for ntp to sync it and it will stay off by 
an hour, and you get the sort of behavior  you were seeing.

It is one disadvantage to having slave backends, I started using a cron job to 
sync things with rdate (to the master backend) as ntp is somewhat fragile and 
takes careful monitoring (kind of painful in a home environment, not so bad for 
enterprise environments).

                        Roger