So you can't even manually balloon up? The blog seems to suggest that
this can be necessary with some kernel+Xen combinations.
Beyond that I'm afraid I've no other ideas.
On Thu, 2014-09-04 at 18:17 +0800, kevin.zhang@xxxxxxxxxxx wrote:
> Ian Campbellï
> Thank you for your reply.
> I've read the page you provided, but my situation is different.
> Even I set dom0_mem=6G,max:8G, the displayed domain-0 memory is
> still 3979. It seems that it can provide no more than 4G
> I don't know why the same option differ from 1 socket and 2 socket
> servers.
>
>
> This time dom0_mem=6G,max:8G
>
>
> root@ServerV3:~# xm list 0
> Name ID Mem VCPUs State Time(s)
> Domain-0 0 3979 40 r----- 1549.2
>
>
> root@ServerV3:~# cat /proc/meminfo
> MemTotal: 3737508 kB
> MemFree: 2603036 kB
>
>
> root@erverV3:~# free -m
> total used free shared buffers cached
> Mem: 3649 1120 2529 0 41 260
>
> By the way, my 2 sockets server is using E5 2650v3, DDR4 64G, it's
> from Intel Lab, while my 1 socket server is I5 CPU DDR3 16G.
>
>
>
> ______________________________________________________________________
> Best Regards
>
>
>
> From: Ian Campbell
> Date: 2014-09-04 17:34
> To: kevin.zhang@xxxxxxxxxxx
> CC: xen-users
> Subject: Re: [Xen-users] dom0_mem not working in debian wheezy
> On Thu, 2014-09-04 at 11:32 +0800, kevin.zhang@xxxxxxxxxxx
> wrote:
> >
> > Hi all,
> >
> >
> > Today I found that dom0_mem option does not work in my
> 64G 2 cpu
> > server , but works in my 16G 1 CPU server.
> > I am using debian wheezy and xen 4.3.3.
> > Firstly I set GRUB_CMDLINE_XEN="dom0_mem=6G,max:6G"
> > in /etc/default/grub,
> > Then execute:
> > update-grub
> > After rebooting, execute:
> > xm list 0
> > Only see no more than 3967M reserved for Domain-0 in my
> 64G 2CPU
> > server, however, my 16G 1CPU server display correctly.
> > If I use xm mem-set 0 5000, dom0 memory also can be
> raised.
> > But why it does not work with dom0_mem option?
>
> Does
> http://blog.xen.org/index.php/2012/04/30/memory-where-it-has-not-gone/
> relate to your situation?
>
> Ian.
>
>
>
