yahoo-eng-team team mailing list archive

[DeyunLuo <2079790@xxxxxxxxxxxxxxxxxx>]

** Project changed: kolla-ansible => nova

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to OpenStack Compute (nova).
https://bugs.launchpad.net/bugs/2079790

Title:
  nova_libvirt restart cause vm restart

Status in OpenStack Compute (nova):
  New

Bug description:
  Enviroment information:

  Rocky Linux 9.3
  koll-ansible version is 17.1.0
  openstack_release: "master"
  --------------

  when nova_libvirt restart, the vm will restart.

  May be the qemu-kvm process ppid is not 1, is nova_libvirt containerd-
  shim process

  UID          PID    PPID    PGID     SID  C STIME TTY          TIME CMD
  42436    1531817  926797 1531816 1531816  1 Sep05 ?        00:20:34 /usr/libexec/qemu-kvm -name guest=instance-00000481,debug-threads=on 

  UID          PID    PPID    PGID     SID  C STIME TTY          TIME CMD
  root      926797       1  926797   63101  0 Sep05 ?        00:00:12 /usr/bin/containerd-shim-runc-v2 -namespace moby -id 89aea1c0f3710f9fdc85d30f9012a6c1b43c151d283019aa1bc329658ad71944 -address /run/

  89aea1c0f3710f9fdc85d30 this is nova_libvirt containerd-shim process

  
  If an vm's process(qemu-kvm) ppid is not 1(systemd)

  nova_libvirt container resetart will cause vm restart

  in libvirt source code like this 
  https://github.com/libvirt/libvirt/blob/6f10d15ca18c927e6533577421d437826314548e/src/util/vircommand.c#L808
  create a vm, libvirt will use setsid syscall to change the qemu-kvm process ppid to 1

  I find releated information in
  https://review.opendev.org/c/openstack/kolla-ansible/+/806476

  But in kolla-ansible version 17.1.0

  /sys/fs/cgroup is already map to nova_libvirt container

To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/2079790/+subscriptions