Hi there,
We're running virsh on centos and hit a couple of issues. First instance is cpu.time not incrementing even with the VMs on the hypervisor running a stress tool. After running 'virsh domstats' sometimes it'll go into a state where no output is generated at all. Have anyone seen something like this? We're running a pretty recent build of libvirt.
# virsh version
Compiled against library: libvirt 3.9.0
Using library: libvirt 3.9.0
Using API: QEMU 3.9.0
RUNNING HYPERVISOR: QEMU 2.9.0
Through the two domstats we see that the cpu.time is not increasing, but vcpu.0.time is not. It's also not reporting a vcpu.0.wait on this centos hypervisor.
# virsh domstats
Domain: '3a1de3d0-9de3-4876-a51a-1337d2ad0b87'
state.state=1
state.reason=5
cpu.time=18530581887
cpu.user=430000000
cpu.system=2910000000
balloon.current=786432
balloon.maximum=786432
vcpu.current=1
vcpu.maximum=1
vcpu.0.state=1
vcpu.0.time=1145240000000
# virsh domstats
Domain: '3a1de3d0-9de3-4876-a51a-1337d2ad0b87'
state.state=1
state.reason=5
cpu.time=18530581887
cpu.user=430000000
cpu.system=2910000000
balloon.current=786432
balloon.maximum=786432
vcpu.current=1
vcpu.maximum=1
vcpu.0.state=1
vcpu.0.time=1153920000000