virtcontainers: fix invalid CPU topology

[devimc]

sockets * cores * threads should be equal to maxcpus otherwise a
warning is thrown: 'warning: Invalid CPU topology deprecated:
sockets * cores * threads != maxcpus'

This warning in the future will be an error and won't be possible to run
kata containers.

fixes #1605

Signed-off-by: Julio Montes julio.montes@intel.com

[devimc]

/test

[devimc]

/test

[grahamwhaley]

(thought I'd left this question already - maybe it was on a parallel PR or Issue???)
@devimc - does this have any performance impacts? iirc, way back when we first played with topologies there was discussion that different topologies could affect performance. Did you measure with the metrics report by chance?

[devimc]

@grahamwhaley metrics-CI is happy, do you want me to run a specific test?

[grahamwhaley]

@devimc - nope, thats fine. Fair point. Maybe it made more of a difference when we used to set the default ncpu to >=4 etc. Now we set to 1 and use hotplug, maybe we won't notice (at least not in the default container case). Hmm, maybe we need some metrics using ncpu > 1 :-)

Just for ref, metrics looks good with:

21:34:47 +-----+----------------------+-------+--------+--------+-------+--------+--------+-------+------+-----+
21:34:47 | P/F |         NAME         |  FLR  |  MEAN  |  CEIL  |  GAP  |  MIN   |  MAX   |  RNG  | COV  | ITS |
21:34:47 +-----+----------------------+-------+--------+--------+-------+--------+--------+-------+------+-----+
21:34:47 | P   | boot-times           | 95.0% | 102.8% | 105.0% | 10.0% | 93.4%  | 122.1% | 30.7% | 6.0% |  20 |
21:34:47 | P   | memory-footprint     | 95.0% | 100.3% | 105.0% | 10.0% | 100.3% | 100.3% | 0.0%  | 0.0% |   1 |
21:34:47 | P   | memory-footprint-ksm | 95.0% | 101.8% | 105.0% | 10.0% | 101.8% | 101.8% | 0.0%  | 0.0% |   1 |
21:34:47 +-----+----------------------+-------+--------+--------+-------+--------+--------+-------+------+-----+

[grahamwhaley]

lgtm

[grahamwhaley]

Nudged fedora CI rebuild - looked like a net timeout.
Of course, when this lands, we'll also need to tweak the metrics CI bounds checks. Now, it would be great if somebody other than me did this - so we can spread the love - and not have me as the bottleneck...
In my head I'm even thinking 'he who broke it, fixes it'... heh.