Hello again,
Correct, I only have the one OpenBSD VM on that test host at the moment.
I rebooted the VM and started the date loop script. During this time, I
took several trace-cmd samples using the various command line args. Let
me know which of these formats contains the data you need, then I'll
repeat the original experiment.
trace-01-boot.tar.gz (16G -> 1.7G compressed), trace-cmd record -e kvm
trace-02-date-loop-started.tar.gz (4.9M -> 1M compressed), trace-cmd
record -e kvm -c -F -P 19582
trace-03-date-loop-running.tar.gz (5.3M -> 1.1M compressed) trace-cmd
record -e kvm -F -P 19582
http://45.63.6.248/kvm2/trace-01-boot.tar.gz
http://45.63.6.248/kvm2/trace-02-date-loop-started.tar.gz
http://45.63.6.248/kvm2/trace-03-date-loop-running.tar.gz
Best,
Andrew
On 1/2/2018 11:22 AM, Paolo Bonzini wrote:
On 28/12/2017 18:00, Andrew Davis wrote:
Ok, this is interesting. Please gather a trace using
trace-cmd record -e kvm -c -P pid-of-qemu
for about 10 seconds; run it once right after booting, once when it
starts taking 2 seconds, one when it starts taking 4 seconds.
Looks like "-c" didn't have the expected result (the man page says "Used
with either -F (or -P if kernel supports it)" so maybe that's it).
Assuming you have only one VM on the host, can you try again without "-c
-P pid-of-qemu"?
Thanks,
Paolo
