extremely slow disk access

extremely slow disk access

[Xen User]

Hi list members,

I have a fresh FC3 with fresh bk clone of xen-2.0.bk.  It builds and 
installs without complaint.

"/lib/tls" has been moved to "/lib/tls.disabled".

My understanding is that in a stand-alone (no vms running) situation the 
performance difference between a standard kernel and a xen enabled 
kernel should be roughly the same.  Is this correct?

2.6.10-xen0 boots with no errors, but the mouse seems jerky and slow to 
respond but is functional.

Networking is up and functional.

The drives are IDE in a raid0 using lvm.  The m/b is typical budget x86 
with 1GB ram and 1GHz Athlon.  It has been in various uses without 
failure for a couple years.

As a test I created a logical volume "Main/test" of 1G rw and formatted 
it with ext3.  The inode creation part took perhaps 1 second but the 
journal creation took over 1 minute.  I was able to have lunch (and do 
the dishes) in the time it took to format 18G.

Reboot into 2.6.9 non-xen and try the same experiment and the entire 
process takes less than 2 seconds.

I looked in /proc/interrupts but nothing there seems awry, i.e. no 
runaway counts.

Does anyone have any idea what may be wrong or where I should look next?

Thanks for any help,
Mike Wright


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: extremely slow disk access

[Steven Hand]

> Hi list members,
> 
> I have a fresh FC3 with fresh bk clone of xen-2.0.bk.  It builds and 
> installs without complaint.
> 
> "/lib/tls" has been moved to "/lib/tls.disabled".
> 
> My understanding is that in a stand-alone (no vms running) situation the 
> performance difference between a standard kernel and a xen enabled 
> kernel should be roughly the same.  Is this correct?
> 
> 2.6.10-xen0 boots with no errors, but the mouse seems jerky and slow to 
> respond but is functional.
> 
> Networking is up and functional.
> 
> The drives are IDE in a raid0 using lvm.  The m/b is typical budget x86 
> with 1GB ram and 1GHz Athlon.  It has been in various uses without 
> failure for a couple years.
> 
> As a test I created a logical volume "Main/test" of 1G rw and formatted 
> it with ext3.  The inode creation part took perhaps 1 second but the 
> journal creation took over 1 minute.  I was able to have lunch (and do 
> the dishes) in the time it took to format 18G.
> 
> Reboot into 2.6.9 non-xen and try the same experiment and the entire 
> process takes less than 2 seconds.
> 
> I looked in /proc/interrupts but nothing there seems awry, i.e. no 
> runaway counts.
> 
> Does anyone have any idea what may be wrong or where I should look next?

Looks like you're falling back to PIO mode under Xen. 

Have you compiled in support for your IDE chipset? 

cheers,

S.


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

RE: extremely slow disk access

[Ian Pratt]

> My understanding is that in a stand-alone (no vms running) 
> situation the 
> performance difference between a standard kernel and a xen enabled 
> kernel should be roughly the same.  Is this correct?

Yes. Xen is clearly very unhappy on your system. 

Are you sure the correct ide driver is compiled into the kernel and is
being used? Compare the boot messages between native and booting the xen
kernel. 

Ian
 
> 2.6.10-xen0 boots with no errors, but the mouse seems jerky 
> and slow to 
> respond but is functional.
> 
> Networking is up and functional.
> 
> The drives are IDE in a raid0 using lvm.  The m/b is typical 
> budget x86 
> with 1GB ram and 1GHz Athlon.  It has been in various uses without 
> failure for a couple years.
> 
> As a test I created a logical volume "Main/test" of 1G rw and 
> formatted 
> it with ext3.  The inode creation part took perhaps 1 second but the 
> journal creation took over 1 minute.  I was able to have 
> lunch (and do 
> the dishes) in the time it took to format 18G.
> 
> Reboot into 2.6.9 non-xen and try the same experiment and the entire 
> process takes less than 2 seconds.
> 
> I looked in /proc/interrupts but nothing there seems awry, i.e. no 
> runaway counts.
> 
> Does anyone have any idea what may be wrong or where I should 
> look next?
> 
> Thanks for any help,
> Mike Wright
> 
> 
> -------------------------------------------------------
> SF email is sponsored by - The IT Product Guide
> Read honest & candid reviews on hundreds of IT Products from 
> real users.
> Discover which products truly live up to the hype. Start reading now.
> http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/xen-devel
> 


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_ide95&alloc_id\x14396&op=click

Re: extremely slow disk access

[Nivedita Singhvi]

Xen User wrote:

> 2.6.10-xen0 boots with no errors, but the mouse seems jerky and slow to 
> respond but is functional.
> 
> Networking is up and functional.
> 
> The drives are IDE in a raid0 using lvm.  The m/b is typical budget x86 
> with 1GB ram and 1GHz Athlon.  It has been in various uses without 
> failure for a couple years.

This may be irrelevant, but how much of the memory are you
allocating to dom0 in your configuration? I had similar
symptoms when running under 64MB.  Increasing the amount
of memory in my system (I happen to have 256MB for dom0 now)
eliminated all of the jerky and slow behaviour.

Others have reported no problems, however, when running in
low mem conditions.

thanks,
Nivedita




-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: extremely slow disk access

[Xen User]

Steven Hand wrote:
>>Hi list members,
>>
>>Does anyone have any idea what may be wrong or where I should look next?
> 
> 
> Looks like you're falling back to PIO mode under Xen. 
> 
> Have you compiled in support for your IDE chipset? 
> 
> cheers,
> 

Steven, you hit the nail on the head.

It took quite a while to recompile in its brain-damaged state but in 
this case the physician really did heal itself!

And, yes, Ian, the performance is now, indeed, on par with the unxenifed 
kernel.

I have only 1G so I will have to apportion it carefully; I'll keep your 
advice in mind, Nivedita.

Now the real fun starts :)

Thanks to all,
Mike Wright


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click