yahoo-eng-team team mailing list archive

[Matt Riedemann <mriedem.os@xxxxxxxxx>]

This is the related mailing list thread:

http://lists.openstack.org/pipermail/openstack-
discuss/2019-June/007363.html

I've proposed that when forcefully rebuilding the cache, since we don't
know which ports nova created on behalf of the user, we should detect
that and set preserve_on_delete=True so that the ports are not
accidentally deleted when the server is deleted (or when the ports are
detached). This is a change in behavior for the user, but so is deleting
the ports they didn't want us to delete and that's arguably worse. With
my suggested solution we'd at least give the owner of the port the
chance to clean it up.

** Tags added: neutron

** Description changed:

  Description
  ===========
- We spawn VM with port-id ,the value of preserve_on_delete is True ,When we lose some ports from instance_info_caches ,after periodic task _heal_instance_info_cache the instance_info_caches are recovery from neutron, however the value of preserve_on_delete became False,not the original value False    
+ We spawn VM with port-id ,the value of preserve_on_delete is True ,When we lose some ports from instance_info_caches ,after periodic task _heal_instance_info_cache the instance_info_caches are recovery from neutron, however the value of preserve_on_delete became False,not the original value True.
  
  Steps to reproduce
  ==================
  1.Spawn VM with command :nova boot --flavor xx --nic port-id xx --image xx instance_name
  2.nova interface-list instance_name(Preserve On Delete is True)
  3.Update nova db row, drop one interface from interfaces,such as
  Update instance_info_caches set network_info='[]' where instance_uuid='xxx';
  4.nova show instance_name (we can see ,vm lost network info )
- 5.after periodic task _heal_instance_info_cache , interface is added to the instance 
+ 5.after periodic task _heal_instance_info_cache , interface is added to the instance
  6.nova interface-list instance_name   (Preserve On Delete became False ,not the original value)

** Changed in: nova
       Status: New => Triaged

** Changed in: nova
   Importance: Undecided => Medium

** Also affects: nova/stein
   Importance: Undecided
       Status: New

** Changed in: nova/stein
       Status: New => Triaged

** Changed in: nova/stein
   Importance: Undecided => Medium

-- 
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/1834463

Title:
  Value of perserve_on_delete not corrte after _heal_instance_info_cache

Status in OpenStack Compute (nova):
  Triaged
Status in OpenStack Compute (nova) stein series:
  Triaged

Bug description:
  Description
  ===========
  We spawn VM with port-id ,the value of preserve_on_delete is True ,When we lose some ports from instance_info_caches ,after periodic task _heal_instance_info_cache the instance_info_caches are recovery from neutron, however the value of preserve_on_delete became False,not the original value True.

  Steps to reproduce
  ==================
  1.Spawn VM with command :nova boot --flavor xx --nic port-id xx --image xx instance_name
  2.nova interface-list instance_name(Preserve On Delete is True)
  3.Update nova db row, drop one interface from interfaces,such as
  Update instance_info_caches set network_info='[]' where instance_uuid='xxx';
  4.nova show instance_name (we can see ,vm lost network info )
  5.after periodic task _heal_instance_info_cache , interface is added to the instance
  6.nova interface-list instance_name   (Preserve On Delete became False ,not the original value)

To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/1834463/+subscriptions