[mythtv] Scheduler needs table keys?

f-myth-users at media.mit.edu f-myth-users at media.mit.edu
Sun Mar 4 06:27:35 UTC 2007


    > Date: Sun, 4 Mar 2007 00:23:03 -0500
    > From: Chris Pinkham <cpinkham at bc2va.org>

    > * On Sat Mar 03, 2007 at 11:26:46PM -0500, f-myth-users at media.mit.edu wrote:
    > > The thing that weirds me out a little about this is that those IOBOUND
    > > errors were really common around scheduler runs.  I suppose I could be

    > I can see how that could happen.  The scheduler inadvertently blocked the
    > recorded table.  The recorder got hung up waiting to update
    > recorded.filesize.  The ivtv buffers filled up.  Now, all the sudden the
    > scheduler query finishes, the recorder updates recorded.filesize and
    > starts grabbing all that queued up data which in turn filters it's way
    > over to TFW which potentially kicks out a few IOBOUND errors because
    > of the surge of data it just received.

In fact, I think this is likely correct---I just spot-checked some
mythtv logs which had IOBOUNDs in them.  In the ~10 cases I examined,
the IOBOUND was followed within 500ms by 
   |-- 11018 results in 24.92 sec. Processing...
or a similar message, whereas I'd expect a more-random distribution of
those messages in the 10-20 seconds before the "Processing..." message
if this wasn't a sudden buffer-release at the end.

The cases of corruption that failed to show IOBOUND must have been
when ivtv data got dropped, but for some reason the disk subsystem
happened to be fast enough that TFW didn't have to wait too long
when it emptied the full ivtv buffer(s).


More information about the mythtv-dev mailing list