Hi all,
I've been running Dapper on my laptop for over month now, but just
in the last day or so I've been noticing really slow disk access.
For instance, one of the test suite programs for libsndfile is taking
over a minute to run on the laptop (1.1GHz CPU) and runs in under 6
seconds on my
On Thu, Jun 22, 2006 at 10:50:39PM +1000, Erik de Castro Lopo wrote:
Hi all,
I've been running Dapper on my laptop for over month now, but just
in the last day or so I've been noticing really slow disk access.
For instance, one of the test suite programs for libsndfile is taking
over a
Andrew Bennetts wrote:
I've found my system (or just X?)
I'm not noticing X, or general CPU performance, only disk access.
seems very sluggish in general for me with
2.6.15-25-686, compared to 2.6.15-23-686.
Problem disappears with 2.6.15-23-686. Looks like 2.6.15-25-686
is borked.
Erik
--
Erik de Castro Lopo wrote:
Problem disappears with 2.6.15-23-686. Looks like 2.6.15-25-686
is borked.
Ooops, I spoke too soon. First run on my test suite after booting
to 2.6.15-23-686 ran fine. Second run was slow again.
I do notice that when things are running slow, top shows these two
quote who=Erik de Castro Lopo
8227 root 26 10 1788 732 504 R 2.3 0.1 0:06.15 updatedb
I don't think I should be seeing those at the top.
You shouldn't be seeing updatedb at all, unless it's being run by cron (in
the morning) or anacron (after boot).
- Jeff
--
GUADEC 2006:
Jeff Waugh wrote:
You shouldn't be seeing updatedb at all, unless it's being run by cron (in
the morning) or anacron (after boot).
That was within about 5 minutes of a boot.
This problem is somewhat elusive. On my machine 2.6.15-25-686 is
definitely borked, but I have seen the problem on
I'd suggest you install and run dstat and look for whether the CPU
is in wait a lot - this would indicate some sort of I'O related issue.
On 6/23/06, Erik de Castro Lopo [EMAIL PROTECTED] wrote:
Erik de Castro Lopo wrote:
Problem disappears with 2.6.15-23-686. Looks like 2.6.15-25-686
is
quote who=Erik de Castro Lopo
Jeff Waugh wrote:
You shouldn't be seeing updatedb at all, unless it's being run by cron
(in the morning) or anacron (after boot).
That was within about 5 minutes of a boot.
Weird.
updatedb running at startup on a laptop isn't very weird. :-)
- Jeff
--
Erik de Castro Lopo wrote:
Erik de Castro Lopo wrote:
Problem disappears with 2.6.15-23-686. Looks like 2.6.15-25-686
is borked.
Ooops, I spoke too soon. First run on my test suite after booting
to 2.6.15-23-686 ran fine. Second run was slow again.
I do notice that when things are running
Martin Visser wrote:
I'd suggest you install and run dstat and look for whether the CPU
is in wait a lot - this would indicate some sort of I'O related issue.
Ok, got it. Running it in 2.6.15-25-686 with my test program an very
little else running I get:
dstat
total-cpu-usage
On 23 Jun 2006, Erik de Castro Lopo [EMAIL PROTECTED] wrote:
Anyway, it does seem to be spending way too much time in wait and idle
when it should doing some bloody processing.
You might want to try changing hdparm parameters; in particular
something like
sudo hdparm /dev/hda
sudo hdparm
Martin Pool wrote:
sudo hdparm /dev/hda
Well the existing setting were not too bad:
/dev/hda:
multcount= 0 (off)
IO_support = 0 (default 16-bit)
unmaskirq= 0 (off)
using_dma= 1 (on)
keepsettings = 0 (off)
readonly = 0 (off)
Yeah, this is a classic - i have one breezy box that is a little old ,
and occasionally the DMA reverts to disabled - performance becomes bog
slow. Running hdparm -d 1 /dev/hda fixies it again.
If you do a simple file copy of a large file, you should be able to
sustain 5MB/s read and write
On Fri, 2006-06-23 at 10:00 +1000, Phil Scarratt wrote:
I stopped it from running altogether, which I believe means you have to
stop using locate (no big deal).
It shouldn't take too long to update anyway. I find locate very useful
to have.
--
Simon Wong [EMAIL PROTECTED]
--
SLUG -
Simon Wong wrote:
It shouldn't take too long to update anyway. I find locate very useful
to have.
I have proved to myself that updated is not the problem so I'll be
keeping it :-).
Erik
--
+---+
Erik de Castro Lopo
Simon Wong wrote:
On Fri, 2006-06-23 at 10:00 +1000, Phil Scarratt wrote:
I stopped it from running altogether, which I believe means you have to
stop using locate (no big deal).
It shouldn't take too long to update anyway. I find locate very useful
to have.
In most cases it doesn't - just
On Fri, Jun 23, 2006 at 01:39:20PM +1000, Phil Scarratt wrote:
In most cases it doesn't - just that in my case it was and was slowing
things right down - something was borked. As I don't use locate that
much, it didn't bother me stopping it.
Bit of random data for y'all: updatedb (almost)
Erik de Castro Lopo wrote:
I think I'll go back to kernel 2.6.15-23-686 and get some dstat
results and then compare those results with what I'm getting with
2.6.15-25-686.
The dstat results are vastly different across the two kernels
even though the hdparm settings are identical.
I've raised
On Fri, 23 Jun 2006 14:20:25 +1000, Erik de Castro Lopo uttered
https://launchpad.net/distros/baltix/+bug/50741
Unfortunately, I screwed up an logged it under Baltix (wtf is that?)
instead of Ubuntu. Is there some way I can fix that?
I just had a look at the bug, and it looks like you
Erik de Castro Lopo wrote:
[...]
I've raised an Ubuntu bug against linux-image-2.6.15-25-686 :
https://launchpad.net/distros/baltix/+bug/50741
Unfortunately, I screwed up an logged it under Baltix (wtf is that?)
instead of Ubuntu. Is there some way I can fix that?
Click the
On 23 Jun 2006, Erik de Castro Lopo [EMAIL PROTECTED] wrote:
The dstat results are vastly different across the two kernels
even though the hdparm settings are identical.
OK, it's good to eliminate it.
I've raised an Ubuntu bug against linux-image-2.6.15-25-686 :
Martin Pool wrote:
To reassign it you need to click on the Affects Baltix link, click
Also affects distribution, then choose Ubuntu. Then you can mark
the report for Baltix as invalid. This is perhaps not the most obvious
way to correct a fairly easy mistake :} but that's how you do it.
22 matches
Mail list logo