[mythtv-users] TFW, Error: Write() -- IOBOUND ..... and mythbox crashed

[Ma Begaj]

Thanks ...

These both crashes occured when I am recording 3 three shows and
watching/fast-forwarding a recording at the same time.

I turned off AAM (Acousting Management) of the both disks to try to
make them faster with:
hdparm -M 254 /dev/hd[ab]

Disks are louder and that is not optimal, but that might help.

iostat -xk gave me most of the time something like this (AAM turned off):
Device:  %util
hda  11.35
hdb  4.45

avg-cpu:  %user   %nice %system %iowait  %steal   %idle
          40.10    0.00    4.51    3.76    0.00   51.63

Utilization never goes above 15%.


I tried to fast-forward to recording which I am watching (during
running 3 recordings at the same time) and fast-forwarding was not
working so nice as usual (not fluent) and I got "Audio buffer
overflow" errors in mythfrontend logs:

NVP::AddAudioData():p1: Audio buffer overflow, audio data lost!


Disk is doing pretty much ... 3 recordings, transcoding,
fast-forwarding in a video... is that maybe just too much for these
IDE disks? I hope not.

But it is not crashing at the moment ... maybe turning off AAM helped.

M.
2007/10/18, George Mari <george_mythusers at mari1938.org>:
> Ma Begaj wrote:
> > Hi,
> >
> > I had yesterday three recordings running at the same time (1x pvr150,
> > 1x pvr500). At the same time, I was watching a recording. And suddenly
> > my mythbox crashed and I had to reset it.
> >
> > I checked syslog and found nothing, but I found something in
> > mythbackend log file:
> >
> > 2007-10-17 23:32:05.673 TFW, Error: Write() -- IOBOUND begin cnt(2048) free(835)
> > 2007-10-17 23:32:10.724 TFW, Error: Write() -- IOBOUND end
> > 2007-10-17 23:32:10.731 TFW, Error: Write() -- IOBOUND begin cnt(2048)
> > free(1095)
> > 2007-10-17 23:32:10.735 TFW, Error: Write() -- IOBOUND end
> > 2007-10-17 23:47:10.183 TFW, Error: Write() -- IOBOUND begin cnt(2048)
> > free(1503)
> >
> > I checked the older mythbackend logs and I found more of these
> > messages in the last few days, usually when I had 2-3 recordings (SD)
> > running and watching at the same time.
> >
> > I have two IDE hard disks (2x 160GB WD) and 1GB single channel RAM.
> > One of these disks has a system and home partitions and a partition
> > for myth recordings. 2nd disk is only one partition for recordings.
> > Everything is ReiserFS. I was thinking to change it to XFS these days.
> >
> >
> > I run "hdparm -tT" and they are more or less the same for both disks:
> > root at mythbox:/tmp# hdparm -Tt /dev/hd[ab]
> >
> > /dev/hda:
> >  Timing cached reads:   1332 MB in  2.00 seconds = 666.17 MB/sec
> >  Timing buffered disk reads:  136 MB in  3.03 seconds =  44.81 MB/sec
> >
> > /dev/hdb:
> >  Timing cached reads:   1090 MB in  2.00 seconds = 545.26 MB/sec
> >  Timing buffered disk reads:  172 MB in  3.02 seconds =  56.97 MB/sec
> >
> > not pretty fast? this
> > http://readlist.com/lists/mythtv.org/mythtv-users/7/39724.html has the
> > same problem, and the disks have two or almost three times higher
> > "Timing cached reads".
> >
> > any ideas? I could buy new disk(s), but I think that these disks
> > should be handle 2-3 recordings and 1 reading at the same time.
> >
>
> You may be running into the limit of how many I/O operations per second
> your single disk supports, as opposed to the bandwidth limit.
>
> If you have iostat, (part of the sysstat RPM on my FC6 installation) you
> can run iostat -xk 2 in a shell while you are recording 3 streams and
> watching another to get a picture of your I/O system.  The last column
> gives you % utilization per device, and if it's at or above 100%, well,
> there's your answer.
>
>
> _______________________________________________
> mythtv-users mailing list
> mythtv-users at mythtv.org
> http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
>


-- 
Glauben heißt nicht wissen wollen, was wahr ist ... Nietzsche