Re: [PATCH 08 of 10 [RFC]] xl: Introduce First Fit memory-wise placement of guests on nodes

From: Dario Faggioli <dario.faggioli@citrix.com>
To: George Dunlap <george.dunlap@eu.citrix.com>
Cc: Andre Przywara <andre.przywara@amd.com>,
	Ian Campbell <Ian.Campbell@citrix.com>,
	Stefano Stabellini <Stefano.Stabellini@eu.citrix.com>,
	Juergen Gross <juergen.gross@ts.fujitsu.com>,
	Ian Jackson <Ian.Jackson@eu.citrix.com>,
	"xen-devel@lists.xen.org" <xen-devel@lists.xen.org>,
	Jan Beulich <JBeulich@suse.com>
Subject: Re: [PATCH 08 of 10 [RFC]] xl: Introduce First Fit memory-wise placement of guests on nodes
Date: Thu, 3 May 2012 03:03:52 +0200	[thread overview]
Message-ID: <1336007032.874.12.camel@Abyss> (raw)
In-Reply-To: <1335976251.2961.123.camel@Abyss>

[-- Attachment #1.1: Type: text/plain, Size: 2000 bytes --]

On Wed, 2012-05-02 at 18:30 +0200, Dario Faggioli wrote:
> > > +
> > > +        if (use_cpus>= b_info->max_vcpus) {
> > > +            rc = 0;
> > > +            break;
> > > +        }
> > Hmm -- there's got to be a better way to find out the minimum number
> > of nodes to house a given number of vcpus than just starting at 1
> > and re-trying until we have enough.
>
> ...
>
> However, if what you mean is I could check beforehand whether or not
> the
> user provided configuration will give us enough CPUs and avoid testing
> scenarios that are guaranteed to fail, then I agree and I'll reshape
> the
> code to look like that. 
>
Actually, thinking more about this, I'm not even sure I can do what I
stated above. In fact, I don't think I can assume the number of CPUs
each node is made up of to be known, without actually checking it for
the specific node(s) I want to try using. What if some CPUs are off-line
and stuff like that?

I might have to recheck, but reading topology information takes
on/offline-ness into account, which is something I lose if I assume some
static x number of CPUs in each node. Also, are we sure archs with
different number of CPUs in different nodes won't ever exist? :-P
Finally, consider this is likely going to change when we will have some
mechanism for taking the actual load on the CPUs into account, so maybe
it's not worth spending much time on it right now.

So, given what we currently export wrt topology, and all the above, I'm
not sure I can see any cleverer way of figuring our how many CPUs/nodes
I need than explicitly counting them (for what it counts, xend is also
doing the same AFAIUI :-) ).

Thoughts?

Thanks and Regards,
Dario
-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)

[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

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