[mythtv-users] Mysteriously missing SD data

f-myth-users at media.mit.edu f-myth-users at media.mit.edu
Wed Nov 7 19:04:03 UTC 2007


Did anyone else notice that they lost all their SD data on Sunday?
My grabber ran a little after noon GMT (I'm in the US, but this way
others know what time I'm talking about irrespective of timezones),
and I noticed a few hours later that everything* in my entire program
grid was NO DATA -except- for the Paid Programming slots (go figure!).

*[Yes, this is immediately after the (new!) timezone switchover in most
parts of the US.  Suspicious.  Though I have two video sources and I
think this only bashed one of 'em, e.g., I -think- the channels on my
other source were okay, though now I'm not 100% sure.]

Naturally, I missed a few recordings, although the failure was subtle,
too---recordings scheduled for after around 4-5pm local time were
-still- listed by MythWeb's status page as scheduled, even though
there was no data at all for the entire day in the scheduling grid.
(I would have expected the entire schedule to have vanished.)  If I
hadn't remembered that I had a bunch of recordings scheduled in the
noon-5pm range, I wouldn't even have noticed that anything was amiss
---nasty!

But recordings that should have taken place before noon didn't
happen, and I wasn't going to wait around some more to see if others
that were still listed were also affected.

I tried simply rerunning the grabber/mfdb script and everything came
back.  Unfortunately, I don't have -quite- enough debugging/logging
info to really track this down, although nothing abnormal showed up in
the logging of the process that I already have in place, either for
the run in the morning that failed, or the run shortly before noon
that restored everything.  (And I'm arranging to save more history
for the next time.)

Yes, my database is uncorrupted, and in fact I checked it during
debugging this and no tables had any problems, as usual.

What really caught my eye was this post, which seems to be talking
about the next day, but a similar failure mode:

    > Date: Mon, 5 Nov 2007 23:29:19 -0500
    > From: "Maaltan Naatlam" <maaltan at gmail.com>
    > Subject: [mythtv-users] mythtv 18.1 -> 20.2 upgrade

    > [ . . . lots deleted . . . ]

    > 2: Mythfilldatabase wiped out all program data last night (nov 5th ~5 am)
    > I was debugging until about 8AM or so, it magically worked again. An issue
    > at schedules direct maybe?

Hmm.

This is the first time I've seen this behavior, but I'd really like to
have some reassurance it won't strike again at an even more inopportune
time.

P.S.  I still have the XML data for the video sources, but
unfortunately -after- it was massaged by a script, e.g., for
the source I noticed affected, I no longer have exactly the
bits downloaded from SD.  The ouput data from the script,
though, looks just like the data from any other day's run,
at least by eyeballing it---it's not as if it magically got
zero bytes of data from SD on that run or something.


More information about the mythtv-users mailing list