Bug#904303: /sys/fs/cgroup/cpu, cpuacct/machine.slice changed after systemctl daemon-reload
Michael Biebl
biebl at debian.org
Sat Nov 9 01:26:40 GMT 2019
Control: tags -1 + moreinfo
On Mon, 23 Jul 2018 10:11:15 +0800 Zhu Yijun <lovemrd at gmail.com> wrote:
> Package: systemd
> Version: 232-25+deb9u4
>
>
> Description of problem:
> After a 'systemctl daemon-reload' followed by any service 'systemctl
> restart xxx' the blkio, cpu,cpuacct、memory cgroup path changed.
>
> Steps to Reproduce:
> 1. Create and launch a basic VM
> 2. Check that the cgroups are mapped properly for example with 'virsh
> memtune VM_NAME'
> 3. Execute 'systemctl daemon-reload'
> 4. Restart _any_ service on your machine, such as : systemctl restart libvirtd
>
> before:
> 10:perf_event:/machine/qemu-1462-zyjtest.libvirt-qemu
> 9:blkio:/machine/qemu-1462-zyjtest.libvirt-qemu
> 8:cpu,cpuacct:/machine/qemu-1462-zyjtest.libvirt-qemu/emulator
> 7:memory:/machine/qemu-1462-zyjtest.libvirt-qemu
> 6:freezer:/machine/qemu-1462-zyjtest.libvirt-qemu
> 5:pids:/system.slice/libvirtd.service
> 4:devices:/machine/qemu-1462-zyjtest.libvirt-qemu
> 3:cpuset:/machine/qemu-1462-zyjtest.libvirt-qemu/emulator
> 2:net_cls,net_prio:/machine/qemu-1462-zyjtest.libvirt-qemu
> 1:name=systemd:/system.slice/libvirtd.service
>
> after results:
> 10:perf_event:/machine/qemu-1462-zyjtest.libvirt-qemu
> 9:blkio:/
> 8:cpu,cpuacct:/
> 7:memory:/
> 6:freezer:/machine/qemu-1462-zyjtest.libvirt-qemu
> 5:pids:/system.slice/libvirtd.service
> 4:devices:/system.slice/libvirtd.service
> 3:cpuset:/machine/qemu-1462-zyjtest.libvirt-qemu/emulator
> 2:net_cls,net_prio:/machine/qemu-1462-zyjtest.libvirt-qemu
> 1:name=systemd:/system.slice/libvirtd.servi
>
> # virsh memtune zyj_test
> error: Unable to get memory parameters
> error: Requested operation is not valid: cgroup memory controller is not mounted
>
> I found a similar description at :
> https://bugzilla.redhat.com/show_bug.cgi?id=1140215
>
> And I had checked this systemd version had merged the patches.
> core: introduce new Delegate=yes/no property controlling creation of
> cgroup subhierarchies
> core: don't migrate PIDs for units that may contain subcgroups, do
> this only for leaf units
>
> However, this issue is still exist.
Is the Delegate= option set for those cgroup hierarchies created by libvirt?
--
Why is it that all of the instruments seeking intelligent life in the
universe are pointed away from Earth?
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: OpenPGP digital signature
URL: <http://alioth-lists.debian.net/pipermail/pkg-systemd-maintainers/attachments/20191109/d26314fb/attachment.sig>
More information about the Pkg-systemd-maintainers
mailing list