2017-01-11 17:38 GMT+08:00 Daniel P. Berrange <berrange@redhat.com>:
On Wed, Jan 11, 2017 at 06:52:45AM +0000, Qiao, Liyong wrote:
Hi,
I observe that virsh capabilities give wrong cpu topology on a multiple sockets host
taget@jfz1r04h13:~/libvirt$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 72 On-line CPU(s) list: 0-71 Thread(s) per core: 2 Core(s) per socket: 18 Socket(s): 2 <------------------------ NUMA node(s): 2 Vendor ID: GenuineIntel CPU family: 6 Model: 63 Model name: Intel(R) Xeon(R) CPU E5-2699 v3 @ 2.30GHz Stepping: 2 CPU MHz: 1201.660 CPU max MHz: 3600.0000 CPU min MHz: 1200.0000 BogoMIPS: 4590.78 Virtualization: VT-x L1d cache: 32K L1i cache: 32K L2 cache: 256K L3 cache: 46080K NUMA node0 CPU(s): 0-17,36-53 NUMA node1 CPU(s): 18-35,54-71
But output of virsh capabilities only gives.
<topology sockets='1' cores='18' threads='2'/>
The 'sockets' value is basically "sockets-per-NUMA-node".
looking into code and got this: https://libvirt.org/git/?p=libvirt.git;a=blob;f=src/util/virhostcpu.c;h=
f29f3122acee018b9fd7dca06fd7ae1fc118b210;hb=HEAD#l703
should we change it into
util/virhostcpu.c;h=f29f3122acee018b9fd7dca06fd7ae1fc118b210;hb=HEAD#l704> *sockets += nodesockets;
This also affect nodeinfo.sockets.
BOth the <topology> summary and nodeinfo data is really a broken design as it can't cope with asymetric topologies. We recommend apps to completely ignore this data, and instead look at the fine grained topology info in the capabilities:
<topology> <cells num='2'> <cell id='0'> ... <cpus num='6'> <cpu id='0' socket_id='0' core_id='0' siblings='0'/> <cpu id='1' socket_id='0' core_id='2' siblings='1'/> <cpu id='2' socket_id='0' core_id='4' siblings='2'/> <cpu id='6' socket_id='0' core_id='1' siblings='6'/> <cpu id='7' socket_id='0' core_id='3' siblings='7'/> <cpu id='8' socket_id='0' core_id='5' siblings='8'/> </cpus> </cell> ... </cells> </topology>
Thanks Daniel, it's really good to know this.
Regards, Daniel -- |: http://berrange.com -o- http://www.flickr.com/photos/dberrange/ :| |: http://libvirt.org -o- http://virt-manager.org :| |: http://entangle-photo.org -o- http://search.cpan.org/~danberr/ :|
-- libvir-list mailing list libvir-list@redhat.com https://www.redhat.com/mailman/listinfo/libvir-list
-- 天涯无处不重逢 a leaf duckweed belongs to the sea , where not to meet in life