Closed cyring closed 2 years ago
Sources:
CPUSET=/sys/fs/cgroup/cpuset
SETNAME=host
HOSTCPUS="0-1,6-7"
mkdir ${CPUSET}
mount -t cgroup -ocpuset cpuset ${CPUSET}
mkdir ${CPUSET}/${SETNAME}
/bin/echo ${HOSTCPUS} > ${CPUSET}/${SETNAME}/cpuset.cpus
rmdir ${CPUSET}/${SETNAME}
Sources:
CPUSET=/sys/fs/cgroup/cpuset
SETNAME=host
HOSTCPUS="0-1,6-7"
mkdir ${CPUSET}
mount -t cgroup -ocpuset cpuset ${CPUSET}
mkdir ${CPUSET}/${SETNAME}
/bin/echo ${HOSTCPUS} > ${CPUSET}/${SETNAME}/cpuset.cpus
sed -un p < ${CPUSET}/${SETNAME}/tasks > ${CPUSET}/tasks
rmdir ${CPUSET}/${SETNAME}
## Optionally
umount ${CPUSET}
rmdir ${CPUSET}
After deleting the CPUSETS and enabling the offline CPUs through the UI, two CPUs are still blocking monitoring when re-enabled.
EDIT Issue is gone, it was just a matter of time !
Need to add into README.md
and Wiki, the Daemon and Client necessary cgroup binding.
Now part of the README.md
CoreFreq driver is unaware of cpuset actions. Monitoring in the UI starts frozen.
For example with a cpuset
"0-1,6-7"
Go to menu "HotPlug CPU" and disable all cpus which are not part of the cgroup cpuset
Monitoring should start afterward.