[mythtv-users] Scheduling new recordings requires mythtv restart

[Joel]

Michael T. Dean wrote:
> Your mythfilldatabase run is failing to finish.  This could be due to 
> having a broken video sources or input connections configuration--i.e. 
> you have a video source (usually an unimportant one) whose grabber is 
> not working properly or you have an extra input connected (i.e. for use 
> with a VCR or whatever) whose video source /should/ be set to "No 
> Grabber", but isn't.  However, I'm pretty sure we've fixed all the cases 
> where such broken configurations prevent mythfilldatabase from doing the 
> part of its job it /can/ do.

I have 3 sources (confirmed by checking videosource table), Digital
antenna, Digital cable, and Analog cable.  My listings include data from
all three sources, so I believe my sources are configured properly.

> 
> Therefore, I would guess it's happening because you have too little 
> memory/swap (are you sure swap is enabled?) and while mythfilldatabase 
> runs (and consumes a /large/ amount of memory, but likely when it's 
> doing post-insert processing), the OOM killer has to kill it off to 
> allow other programs to run.  Then again, it may just happen because you 
> have bad hardware that causes memory corruption or something, which 
> causes mythfilldatabase to segfault.

I have 4 GB of RAM and 2 GB of swap.  Only 6M of swap is even in use, so
there's plenty of room to spare.

> 
> Make sure you specify a location for "mythfilldatabase Log Path" in 
> mythfrontend's main General settings and then check the log after it 
> runs automatically.  Do not run mythfilldatabase manually, as doing so 
> may run it in such a way that it does not fail.  Feel free to post the 
> log file after the run if you need help interpreting the problem.  If 
> you see no problems and/or the log file just stops without saying, 
> "mythfilldatabase run complete", look in your system log files to see if 
> the OOM killer is activating or some other problem is occurring during 
> the run.

I have been running mythfilldatabase out of cron.daily, rather than
having the backend call it.  If it's better to have the backend run it
for some reason, I can switch to that method.  I get emailed if there is
any output on stderr, and I have only occasionally seen mythfilldatabase
segfault, though it did it 2 days in a row earlier this month.  For now,
I've added -v most to mythfilldatabase and redirected to a log file
which I'll monitor more closely.  I ran it once now, but I can't confirm
whether it worked or failed since the data was up to date, and I
recently restarted mythbackend.  I do see some error's in mythsocket as
well as the message "Error rescheduling id -1 in
ScheduledRecording::signalChange"  I'm not sure how to interpret these
errors though, as it doesn't seem to match any of your suggested
problem.  You may look to see if you notice anything odd here:
http://ebel.mybox.org/mythfilldatabase.log  Or you can wait until
tomorrow and I'll collect a log run in a normal environment for which I
can confirm success or failure.

Thank you for considering my problem
Joel