First, inital VCPU pinning is set correctly but then it is reset by assigning qemu process to a new cgroup (which contains all CPUs). It's easily fixed by swapping these two actions. --- src/qemu/qemu_driver.c | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) Originally, I thought it would be better to change the initial pinning at cgroup level but then I realized it was not a good idea. AFAIK cgroup olny works for processes and is not usable for per-VCPU pinning. That is, it is fine to have all CPUs in cgroup and fine tune VCPU pinning by setting procss affinity. diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index 5f4adfd..cc3e3b2 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -3081,10 +3081,10 @@ static int qemudSecurityHook(void *data) { /* This must take place before exec(), so that all QEMU * memory allocation is on the correct NUMA node */ - if (qemudInitCpuAffinity(h->vm) < 0) + if (qemuAddToCgroup(h->driver, h->vm->def) < 0) return -1; - if (qemuAddToCgroup(h->driver, h->vm->def) < 0) + if (qemudInitCpuAffinity(h->vm) < 0) return -1; if (h->driver->securityDriver &&