I tried to use rctl_get_racct("jail:0", ...) to get metrics for the host but cputime seems to include time spent in the idle process. On an idle machine with 16 hardware threads, I ran this:
$ while true; do sudo rctl -u jail:0 | grep cputime; sleep 1; done
cputime=4145725
cputime=4145741
cputime=4145757
cputime=4145773
cputime=4145789
The cputime increases by 16 each time. I was expecting filter jail:0 to show summary cputime metrics for the host plus jails but instead, it always shows 100% cpu usage. If this is intentional, it should probably be mentioned in the documentation.
I can probably work around by getting metrics for the idle proc and subtracting.
I tried to use rctl_get_racct("jail:0", ...) to get metrics for the host but cputime seems to include time spent in the idle process. On an idle machine with 16 hardware threads, I ran this: $ while true; do sudo rctl -u jail:0 | grep cputime; sleep 1; done cputime=4145725 cputime=4145741 cputime=4145757 cputime=4145773 cputime=4145789 The cputime increases by 16 each time. I was expecting filter jail:0 to show summary cputime metrics for the host plus jails but instead, it always shows 100% cpu usage. If this is intentional, it should probably be mentioned in the documentation. I can probably work around by getting metrics for the idle proc and subtracting.