yahoo-eng-team team mailing list archive

[Amit Uniyal <2062523@xxxxxxxxxxxxxxxxxx>]

Public bug reported:

Description
===========
After evacuation, when original host/node become seriveable (like nova is running properly in host ), VM status changed in nova DB as 'SHUTOFF' automatically from 'ACTIVE'.

This is because on nova initialization nova checks for all evacuated
instances on original server-host, and delete local copy from host. and
also update status at DB.

Steps to reproduce (reporduced 100%)
====================================

- create a VM (you must have multinode setup to run evacuation)
- get server-host
    openstack server list --long

- server-host -  stop service
    systemctl stop devstack@n-cpu

- server-host - force-down compute service
    openstack compute service --down <host> vm1

- stop VM
    openstack server stop vm

- look for VM power state - stuck at powering off
    openstack server list --long

- open logs api/cpu in both nodes
- evacuate VM
     openstack server evacuate --host=<target-host>  --os-compute-api-version 2.29

- server moved to new host and become active

- orginal server-host start compute service (systemctl)

- look for VM status 
    openstack server list --long


Expected result
===============
After service start at original host, it should not affect VM in any way.

Actual result
=============
openstack server list shows  -

VM went to shutoff,  task state=None, power=no state, stays at expected
new host

although VM can be used, can login to server (tried with virsh)

Environment
===========

Nova: current master or future 2024.2

** Affects: nova
     Importance: Undecided
         Status: New


** Tags: db evacuate

** Tags added: evacuate

-- 
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/2062523

Title:
  evacuation, running VM shown stop after compute service started at
  firsh host.

Status in OpenStack Compute (nova):
  New

Bug description:
  Description
  ===========
  After evacuation, when original host/node become seriveable (like nova is running properly in host ), VM status changed in nova DB as 'SHUTOFF' automatically from 'ACTIVE'.

  This is because on nova initialization nova checks for all evacuated
  instances on original server-host, and delete local copy from host.
  and also update status at DB.

  Steps to reproduce (reporduced 100%)
  ====================================

  - create a VM (you must have multinode setup to run evacuation)
  - get server-host
      openstack server list --long

  - server-host -  stop service
      systemctl stop devstack@n-cpu

  - server-host - force-down compute service
      openstack compute service --down <host> vm1

  - stop VM
      openstack server stop vm

  - look for VM power state - stuck at powering off
      openstack server list --long

  - open logs api/cpu in both nodes
  - evacuate VM
       openstack server evacuate --host=<target-host>  --os-compute-api-version 2.29

  - server moved to new host and become active

  - orginal server-host start compute service (systemctl)

  - look for VM status 
      openstack server list --long

  
  Expected result
  ===============
  After service start at original host, it should not affect VM in any way.

  Actual result
  =============
  openstack server list shows  -

  VM went to shutoff,  task state=None, power=no state, stays at
  expected new host

  although VM can be used, can login to server (tried with virsh)

  Environment
  ===========

  Nova: current master or future 2024.2

To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/2062523/+subscriptions