Xen devel - May 2006 - RE: [PATCH] Make "xm mem-set" be lower bound ondomX-min-mem

> I agree with this. I''m also not sure about putting a lower bound
in
the
> balloon driver, but at least there we know that shrinking to <2% of
> original allocation is a very bad idea with very high probability.
Keir,

In lieu of the recent emails about ballooning, do you still want me to
look into making the balloon driver back off when it goes down the OOM
path? Or is it better for that too to be taken care of by some of the
higher level management tools?

Ky,

Does your patch address the situation Keir described above?

[ASIDE]
We are also seeing some other issues with ballooning. Not sure if they
are related. The latest one is described here:
http://bugzilla.xensource.com/bugzilla/show_bug.cgi?id=650

Thanks,
Aravindh Puthiyaparambil 
Xen Development Team 
Unisys, Tredyffrin PA

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

On 20 May 2006, at 16:14, Puthiyaparambil, Aravindh wrote:
> In lieu of the recent emails about ballooning, do you still want me to
> look into making the balloon driver back off when it goes down the OOM
> path? Or is it better for that too to be taken care of by some of the
> higher level management tools?
The OS itself can tell best whether a given memory target is too 
restrictive and will cause major problems. So hooking off the OOM paths 
does make sense. The "2% test" is a suitable static check for now --
it
might make sense to keep it even when we do have a more dynamic 
detection and backoff method. These in-kernel checks can complement 
safety catches in higher-level management tools which might stop a user 
from e.g. setting a target that is too low for Linux and prevent that 
target from ever being passed to the guest.

  -- Keir


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

The patch I sent implements the minimum in the balloon driver and we are
currently building sles10 with my patch.

K. Y
>>> On Sat, May 20, 2006 at 11:14 am, in message
<EF8D308BE33AF54D8934DF26520252D30484B6FC@USTR-EXCH5.na.uis.unisys.com>,
"Puthiyaparambil, Aravindh"
<aravindh.puthiyaparambil@unisys.com>
wrote: 
>>  I agree with this. I''m also not sure about putting a lower
bound
in
> the
>> balloon driver, but at least there we know that shrinking to <2% of
>> original allocation is a very bad idea with very high probability.
> 
> Keir,
> 
> In lieu of the recent emails about ballooning, do you still want me
to
> look into making the balloon driver back off when it goes down the
OOM
> path? Or is it better for that too to be taken care of by some of
the
> higher level management tools?
> 
> Ky,
> 
> Does your patch address the situation Keir described above?
> 
> [ASIDE]
> We are also seeing some other issues with ballooning. Not sure if
they
> are related. The latest one is described here:
> http://bugzilla.xensource.com/bugzilla/show_bug.cgi?id=650
> 
> Thanks,
> Aravindh Puthiyaparambil 
> Xen Development Team 
> Unisys, Tredyffrin PA

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel