[mythtv-users] Prebufferring pauses, I/O service time spikes
Chase Douglas
chasedouglas.lists at gmail.com
Sat Jul 25 21:41:53 UTC 2009
Recently I upgrade my backend/frontend from an Athlon XP 2000+ system
(a little over 4 years old) to a new core i7 system. However, I am
using the same hard drives I had in the old machine in the new one.
Though I had two hard drives in the old machine, one of them had old
data on it and I never touched it. The only I/O performance issue I
had with it was while watching a recording or live tv while
mythfilldatabase was running since the db was on the same drive as the
recordings. Now that I've moved to the new system I've carved out
enough room from the second hard drive and put the database on there
to alleviate the I/O issue. However, now I get NVP buffering pauses
while just watching a recording. The system isn't recording anything
new or doing any other operations other than trying to play back the
recording. Here's what I see in the frontend log:
2009-07-25 17:15:04.327 NVP(3): prebuffering pause
2009-07-25 17:15:05.674 NVP(3): Prebuffer wait timed out 10 times.
Here's what a one-second interval iostat usage (iostat -x 1) prints
out around that time (sda1 is /, sda2 is an extended partition, sda5
is swap):
avg-cpu: %user %nice %system %iowait %steal %idle
6.53 0.00 0.85 0.12 0.00 92.62
Device: rrqm/s wrqm/s r/s w/s rMB/s wMB/s
avgrq-sz avgqu-sz await svctm %util
sda 0.00 5.00 14.00 1.00 1.75 0.02
242.13 0.04 2.93 1.60 2.40
sda1 0.00 5.00 14.00 1.00 1.75 0.02
242.13 0.04 2.93 1.60 2.40
sda2 0.00 0.00 0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00 0.00
sda5 0.00 0.00 0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00 0.00
avg-cpu: %user %nice %system %iowait %steal %idle
7.01 0.00 1.58 1.70 0.00 89.71
Device: rrqm/s wrqm/s r/s w/s rMB/s wMB/s
avgrq-sz avgqu-sz await svctm %util
sda 1.00 4.00 14.00 3.00 1.88 0.03
229.18 0.62 1.88 13.18 22.40
sda1 1.00 4.00 14.00 3.00 1.88 0.03
229.18 0.62 1.88 13.18 22.40
sda2 0.00 0.00 0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00 0.00
sda5 0.00 0.00 0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00 0.00
avg-cpu: %user %nice %system %iowait %steal %idle
6.65 0.00 1.43 20.08 0.00 72.10
Device: rrqm/s wrqm/s r/s w/s rMB/s wMB/s
avgrq-sz avgqu-sz await svctm %util
sda 2.00 3.00 2.00 2.00 0.25 0.01
136.00 4.54 532.00 250.00 100.00
sda1 2.00 3.00 2.00 2.00 0.25 0.01
136.00 3.94 532.00 250.00 100.00
sda2 0.00 0.00 0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00 0.00
sda5 0.00 0.00 0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00 0.00
avg-cpu: %user %nice %system %iowait %steal %idle
5.13 0.00 0.93 21.59 0.00 72.58
Device: rrqm/s wrqm/s r/s w/s rMB/s wMB/s
avgrq-sz avgqu-sz await svctm %util
sda 2.00 2.00 2.00 6.00 0.50 0.04
137.00 3.36 720.00 125.00 100.00
sda1 2.00 2.00 2.00 5.00 0.50 0.04
156.57 3.10 721.14 142.86 100.00
sda2 0.00 0.00 0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00 0.00
sda5 0.00 0.00 0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00 0.00
avg-cpu: %user %nice %system %iowait %steal %idle
3.06 0.00 2.33 15.91 0.00 78.82
Device: rrqm/s wrqm/s r/s w/s rMB/s wMB/s
avgrq-sz avgqu-sz await svctm %util
sda 1.00 0.00 1.00 3.00 0.25 0.02
136.00 3.18 743.00 250.00 100.00
sda1 1.00 0.00 1.00 2.00 0.25 0.02
181.33 2.59 742.67 333.33 100.00
sda2 0.00 0.00 0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00 0.00
sda5 0.00 0.00 0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00 0.00
avg-cpu: %user %nice %system %iowait %steal %idle
0.37 0.00 0.37 15.83 0.00 83.43
Device: rrqm/s wrqm/s r/s w/s rMB/s wMB/s
avgrq-sz avgqu-sz await svctm %util
sda 1.00 0.00 1.00 1.00 0.25 0.01
264.00 2.00 754.00 500.00 100.00
sda1 1.00 0.00 1.00 1.00 0.25 0.01
264.00 1.34 754.00 500.00 100.00
sda2 0.00 0.00 0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00 0.00
sda5 0.00 0.00 0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00 0.00
avg-cpu: %user %nice %system %iowait %steal %idle
5.03 0.00 0.49 21.47 0.00 73.01
Device: rrqm/s wrqm/s r/s w/s rMB/s wMB/s
avgrq-sz avgqu-sz await svctm %util
sda 2.00 4.00 13.00 6.00 1.88 0.04
205.89 2.15 90.95 52.63 100.00
sda1 2.00 4.00 13.00 5.00 1.88 0.04
217.33 2.05 54.22 55.56 100.00
sda2 0.00 0.00 0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00 0.00
sda5 0.00 0.00 0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00 0.00
avg-cpu: %user %nice %system %iowait %steal %idle
4.91 0.00 0.65 17.70 0.00 76.87
Device: rrqm/s wrqm/s r/s w/s rMB/s wMB/s
avgrq-sz avgqu-sz await svctm %util
sda 0.00 0.00 36.00 2.00 4.62 0.01
249.68 1.36 81.05 18.00 68.40
sda1 0.00 0.00 36.00 1.00 4.62 0.01
256.43 1.35 83.14 18.49 68.40
sda2 0.00 0.00 0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00 0.00
sda5 0.00 0.00 0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00 0.00
avg-cpu: %user %nice %system %iowait %steal %idle
6.72 0.00 0.98 0.12 0.00 92.30
Device: rrqm/s wrqm/s r/s w/s rMB/s wMB/s
avgrq-sz avgqu-sz await svctm %util
sda 0.00 0.00 18.00 0.00 2.25 0.00
256.00 0.04 2.44 1.56 2.80
sda1 0.00 0.00 18.00 0.00 2.25 0.00
256.00 0.04 2.44 1.56 2.80
sda2 0.00 0.00 0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00 0.00
sda5 0.00 0.00 0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00 0.00
The first and the last readings are typical during playback. I see
spikes of svctm at least once every 10 seconds, and I get the NVP
buffering issues when a bunch of them occur like what I pasted above.
So here are my questions: what are normal await spike levels? Do await
times of hundreds of seconds with minimal actual throughput suggest an
issue with the hard drive, the ata controller, or some other
subsystem? Could my hard drive be dying? I also switched from -fixes
to trunk during the system upgrade, but I doubt there would be any
sort of buffering or I/O changes that would elicit this behavior.
I would appreciate any insight others can give me. I can purchase a
new HD if I need to of course, but I want to be sure I need it before
I act.
Thank you,
Chase Douglas
More information about the mythtv-users
mailing list