[mythtv-users] HELP! - More failures while recording - IOBOUND - blocking in ThreadedFileWriter::Write()

[Mark Knecht]

HELP!

WAF & KAF have gone to ZERO!

For the last two days I've been failing to record properly when
recording two shows. The mythbackend log file is filling with messages
such as this:

2005-09-23 08:29:41.168 IOBOUND - blocking in ThreadedFileWriter::Write()
2005-09-23 08:29:41.190 IOBOUND - blocking in ThreadedFileWriter::Write()
2005-09-23 08:29:41.202 IOBOUND - blocking in ThreadedFileWriter::Write()
2005-09-23 08:29:41.248 IOBOUND - blocking in ThreadedFileWriter::Write()
2005-09-23 08:29:41.266 IOBOUND - blocking in ThreadedFileWriter::Write()
2005-09-23 08:29:41.278 IOBOUND - blocking in ThreadedFileWriter::Write()
2005-09-23 08:29:41.301 IOBOUND - blocking in ThreadedFileWriter::Write()
2005-09-23 08:29:41.330 IOBOUND - blocking in ThreadedFileWriter::Write()
2005-09-23 08:29:41.367 IOBOUND - blocking in ThreadedFileWriter::Write()
2005-09-23 08:29:44.473 IOBOUND - blocking in ThreadedFileWriter::Write()
2005-09-23 08:29:44.666 IOBOUND - blocking in ThreadedFileWriter::Write()

Any ideas? This is all a new problem since Tuesday.

3GHz P4, 512MB, 2.6.12-gentoo-r6 for the last two months,
PVR-150/PVR-250, ivtv-0.3.7 & ivtv-0.3.9 - both drivers fail the same
way.

I have lots of disk space. The video storage drive is mounted over NFS
so it's possible there's some sort of NFS permissions thing going on,
but the problem only occurs when I'm recording two shows. If I record
a single show the receording is good and I do not get these messages.

Local DMA is enabled. Video storage is over the network to NFS so I
could use a program for testing NFS read and write speeds, etc.

I've found a few threads talking about changing MythTV code to make
buffers larger, etc., but why now when it's worked great for the last
2-3 months?

As best I can tell I did some updates to the machine on Sept. 13th, 7
days before the problem started. Thse updates where:

xorg-x11
python
udev
libpcap
eix
k3b

After that there were no more updates until Sept. 22nd, yesterday,
which was two days after the problem started.

thanks,
Mark