Re: [Xen-devel] [PATCH] libxl: Make 'xl vcpu-set' work properly on overcommited hosts.

[Ian Jackson <Ian.Jackson@xxxxxxxxxxxxx>]

George Dunlap writes ("Re: [Xen-devel] [PATCH] libxl: Make 'xl vcpu-set' work 
properly on overcommited hosts."):
> Yes, I think IanC and I are disagreeing about ideal xl behavior.  We 
> both agree that "vcpus > pcpus" is a bad configuration.  I think ideally 
> we should support it (because administrators should be allowed to shoot 
> themselves in the foot) and Ian seems to be making the case that we 
> shouldn't support it.

I'm with George on this one.

> > I think what you and Ian would like me to do:
> >
> > For Xen 4.3:
> >   - Add a warning to the vcpu-set that the user might be doing something
> >     silly if they are trying to set 'xl vcpu-set vcpus' where vcpus >
> >     pcpus. Something along: "You are overcommiting. You have x PCPUs and
> >     you want y vCPUS. Aborting - if you would like to do this please use
> >     --no-seatbelt'.
> >   - If user is trying to do vcpu-set on more than pcpus abort the request
> >     unless the '--no-seatbelt' option is provided.
> >
> > For Xen 4.4:
> >   - Remove those two things above and make vcpu-set work irregardless of
> >     the pcpus value? No that does not sound right.
> 
> I think for 4.3, if this patch is the quickest way to get rid of the bug 
> of xl crashing, we should just accept it.

Yes, absolutely.  I think the patch should go into 4.3.

Ian.

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