[mythtv-users] All's well with DataDirect again
Michael T. Dean
mtdean at thirdcontact.com
Sun Apr 8 18:58:24 UTC 2007
On 04/08/2007 08:50 AM, Michael T. Dean wrote:
> Oh, and neither box had been refreshing more than tomorrow and +13 (or
> so) on any previous runs ("Data is already present for ..., skipping".
> I can't (yet) explain that (as the data shouldn't have changed if mfdb
> didn't change it--and no one has touched the DB manually)... (But my
> beach volleyball game beckons.)
Figured it out. All comparisons are made to the previous day. Since,
while DD was having issues, the amount of data available was continually
dropping, mfdb was not doing the refreshes.
However, as soon as DD fixed the issues, my box downloaded Sunday's
data, so Monday's data--which was now being compared to a /full/ set of
listings for Sunday--was suddently sparse in comparison. Basically, so
much new data came in on Sunday, that it caused a domino effect of
refreshes (Sunday causes a refresh of Monday, which causes a refresh of
Tuesday, ...).
So, manually running mfdb with --refresh-all before the scheduled run
would likely have been identical to running without --refresh-all. That
means, also, that those who ran with --refresh-all (or any other
arguments) before the scheduled runs didn't "waste" bandwidth/DD
processing time/... (the end result was the same). Therefore, the only
"waste" would have come from users running --refresh-all after their
normal run picked up everything for them, anyway (which probably didn't
happen because of people checking their logs and noticing that every day
was updated).
In other words, it seems Myth was better than we planned at fixing the
issue for us. Good work with the crystal ball all who contributed to
the "to refresh or not to refresh" code.
Anyway, I just had to figure it out...
Mike
More information about the mythtv-users
mailing list