[libvirt] [RFC][PATCHv1 2/5] libvirt - new API for getting percpu statistics of VM

[Daniel P. Berrange]

On Fri, Apr 15, 2011 at 04:43:03PM +0900, KAMEZAWA Hiroyuki wrote:
> On Fri, 15 Apr 2011 09:43:15 +0200
> Matthias Bolte <matthias.bolte at googlemail.com> wrote:
> 
> > 2011/4/15 KAMEZAWA Hiroyuki <kamezawa.hiroyu at jp.fujitsu.com>:
> > > Per (host) cpu activity of VMs are very insterested numbers
> > > when running VMs on large SMPs. virt-top -1 mode tries to
> > > provide the information. (But it's not implemented.)
> > >
> > > This patch adds a libvirt interface to get per cpu statistics
> > > of each nodes. This patch just adds an interface and driver
> > > entry points. So,
> > >  - doesn't include patches for python.
> > >  - doesn't include any driver.
> > >
> > > Following patches will add some drivers and codes for python.
> > >
> > > Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu at jp.fujitsu.com>
> > > ---
> > >  include/libvirt/libvirt.h.in |   13 ++++++++++
> > >  src/driver.h                 |    6 ++++
> > >  src/esx/esx_driver.c         |    1 +
> > >  src/libvirt.c                |   55 ++++++++++++++++++++++++++++++++++++++++++
> > >  src/libvirt_public.syms      |    4 +++
> > >  src/libxl/libxl_driver.c     |    1 +
> > >  src/lxc/lxc_driver.c         |    1 +
> > >  src/openvz/openvz_driver.c   |    1 +
> > >  src/phyp/phyp_driver.c       |    1 +
> > >  src/qemu/qemu_driver.c       |    1 +
> > >  src/remote/remote_driver.c   |    1 +
> > >  src/test/test_driver.c       |    1 +
> > >  src/uml/uml_driver.c         |    1 +
> > >  src/vbox/vbox_tmpl.c         |    1 +
> > >  src/vmware/vmware_driver.c   |    1 +
> > >  src/xen/xen_driver.c         |    1 +
> > >  src/xenapi/xenapi_driver.c   |    1 +
> > >  17 files changed, 91 insertions(+), 0 deletions(-)
> > >
> > > diff --git a/include/libvirt/libvirt.h.in b/include/libvirt/libvirt.h.in
> > > index 5783303..6b9292c 100644
> > > --- a/include/libvirt/libvirt.h.in
> > > +++ b/include/libvirt/libvirt.h.in
> > > @@ -400,6 +400,13 @@ struct _virDomainMemoryStat {
> > >
> > >  typedef virDomainMemoryStatStruct *virDomainMemoryStatPtr;
> > >
> > > +typedef struct _virDomainPcpuStat virDomainPcpuStatStruct;
> > > +
> > > +struct _virDomainPcpuStat {
> > > +    unsigned long long cpuTime;
> > > +};
> > > +
> > 
> > NACK to adding another public struct to the API. 
> 
> Oh, yes. I searched a sutiable existing API but cannot found.
> 
> Maybe adding new enum to Usui's work will be good but I need an array.
> Hmm, returning an array of unsigned long long is ok ?
> 
> > It's not expendable.
> > As a stylistic nit pleas don't use the term PCPU as this looks like
> > "Physical CPU". Just call it virDomainPerVcpuStat at least.
> > 
> Ah, no, this is PerPhysicalStat
> 
> 
> > Also do you really need the absolute CPU time? 
> yes, for virt-top -1.

I don't think that is correct. What virt-top ultimately displays
to the end user is a % utilization. So it would be happy to get
data in either % util & display that directly, or get absolute
CPU time and calculate deltas to display % util.

I wonder if we could provide an API that could be used to provide
either per-VCPU or per-PCPU  statistics in one go, so that we
can finally have an API for cpu time that works well for VMWare.

eg

    typedef struct _virDomainCPUTime virDomainCPUTime;
    typedef virDomainCPUTime *virDomainCPUTimePtr;

    #define VIR_DOMAIN_CPUTIME_ABS "abs"
    #define VIR_DOMAIN_CPUTIME_UTIL "util"
    ...define more stats later if desired...

    struct _virDomainCPUTime {
         char field[VIR_DOMAIN_CPU_TIME_FIELD_LENGTH];
         unsigned long long value;
    };

    typedef enum {
       VIR_DOMAIN_CPU_TIME_VIRTUAL  = 0,
       VIR_DOMAIN_CPU_TIME_PHYSICAL = 1,
    } virDomainCPUTimeFlags;

    int virDomainGetCPUTime(virDomainPtr dom,
                            virDomainCPUTimePtr stats,
                            int *nstats,
                            unsigned int flags);


When querying per-virtual CPU time,  nstats would be the number of
virtual CPUs in the guest. eg

    virDomainInfo info;
    virDomainGetInfo(dom, &info);
    virDomainCPUTimePtr stats;

    stats = malloc(sizeof(*stats)*info.nrVirtCPU);

    virDomainGetCPUTime(dom, stats, info.nrVirtCPU,
                        VIR_DOMAIN_CPU_TIME_VIRTUAL);



Or to get the break down per physical CPU, use the VIR_NODEINFO_MAX_CPUS
macro

    virNodeInfo info;
    virNodeGetInfo(conn, &info);
    virDomainCPUTimePtr stats;

    stats = malloc(sizeof(*stats)*VIR_NODEINFO_MAX_CPUS(info))

    virDomainGetCPUTime(dom, stats, VIR_NODEINFO_MAX_CPUS(info),
                        VIR_DOMAIN_CPU_TIME_PHYSICAL);


We could also allow 'nstats' to be 0, in which case the API
would simply populate 'nstats' with the required number and
return. The caller can then re-invoke with correct nstats
and allocated memory.


Daniel
-- 
|: http://berrange.com      -o-    http://www.flickr.com/photos/dberrange/ :|
|: http://libvirt.org              -o-             http://virt-manager.org :|
|: http://autobuild.org       -o-         http://search.cpan.org/~danberr/ :|
|: http://entangle-photo.org       -o-       http://live.gnome.org/gtk-vnc :|