[Libvir] [RFC] Check host's minimum memory.

[Atsushi SAKAI]

Hi, Dan

O.K.

I will asking Xen-devel with policy forcing patch.

Thanks
Atsushi SAKAI

"Daniel P. Berrange" <berrange at redhat.com> wrote:

> On Wed, Mar 07, 2007 at 06:43:14AM -0500, Daniel Veillard wrote:
> > On Tue, Mar 06, 2007 at 02:40:24PM +0900, Masayuki Sunou wrote:
> > > Hi
> > > 
> > > The minimum value of the memory guarded with virsh setmem is 4096KB(4MB).
> > > In general, when host's memory is set to 4MB, the host stops.
> > > 
> > > Therefore, I propose the patch to which host's minimum value of the memory
> > > is guarded by 256MB.
> > 
> >   I have run guests in 64MB of memory, so 256 as an arbitrary limit sounds
> > way too high to me. There is some places in libvirt code where 64 is the limit
> > (e.g. xend_parse_sexp_desc() if ((cur_mem > 63) && (cur_mem != max_mem)) ...)
> > so if we set a limit here too that should be homogenized (using a constant
> > in internal.h preferably).
> 
> This still feels wrong to me - I'm don't like the idea of libvirt enforcing
> policy decisions, particularly when they're compiled in. With the current
> bloated Xen python stack we probably won't go below 64 MB, but we won't always
> be managing Xen, nor will Xen forever be stuck with a python stack (i hope :-). 
> I'd say that if possible XenD should be fixed to honour dom0_min_memory in all
> cases, rather than just when ballooning down to start a guest.
> 
> Regards,
> Dan.
> -- 
> |=- Red Hat, Engineering, Emerging Technologies, Boston.  +1 978 392 2496 -=|
> |=-           Perl modules: http://search.cpan.org/~danberr/              -=|
> |=-               Projects: http://freshmeat.net/~danielpb/               -=|
> |=-  GnuPG: 7D3B9505   F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505  -=| 
> 
> --
> Libvir-list mailing list
> Libvir-list at redhat.com
> https://www.redhat.com/mailman/listinfo/libvir-list