Unshelving an offloaded server does not restrict to the original cell

Bug #1702935 reported by Matt Riedemann
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
OpenStack Compute (nova)
Fix Released
High
Matt Riedemann

Bug Description

This was mentioned in bug 1702932 but is a separate issue.

When we create a server it gets put in a specific cell and it's instance_mappings record in the API DB is set to map that instance to that cell.

If we migrate (cold, resize or live) the instance, the conductor task manager restrict the instance move operation to the same cell:

https://github.com/openstack/nova/blob/56cd608d3a199dcb02ac2ae071ff3057241259da/nova/conductor/tasks/migrate.py#L63-L66

However, when we unshelve an offloaded server, there is nothing restricting the scheduler from picking a host in another cell:

https://github.com/openstack/nova/blob/56cd608d3a199dcb02ac2ae071ff3057241259da/nova/conductor/manager.py#L673-L700

Matt Riedemann (mriedem)
Changed in nova:
status: New → Confirmed
importance: Undecided → High
Matt Riedemann (mriedem)
tags: added: pike-rc-potential
Revision history for this message
OpenStack Infra (hudson-openstack) wrote : Fix proposed to nova (master)

Fix proposed to branch: master
Review: https://review.openstack.org/486208

Changed in nova:
assignee: nobody → Chris Friesen (cbf123)
status: Confirmed → In Progress
Changed in nova:
assignee: Chris Friesen (cbf123) → Matt Riedemann (mriedem)
Revision history for this message
OpenStack Infra (hudson-openstack) wrote : Fix merged to nova (master)

Reviewed: https://review.openstack.org/486208
Committed: https://git.openstack.org/cgit/openstack/nova/commit/?id=06b4ff7df857f56d6ecb819895c17e1e708dd071
Submitter: Jenkins
Branch: master

commit 06b4ff7df857f56d6ecb819895c17e1e708dd071
Author: Chris Friesen <email address hidden>
Date: Fri Jul 21 13:25:48 2017 -0600

    Ensure we unshelve in the cell the instance is mapped

    When we create a server it gets put in a specific cell and its
    instance_mappings record in the API DB is set to map that instance
    to that cell.

    When unshelving an instance we want to ensure that the scheduler
    only considers compute nodes in the cell where the instance is
    mapped.

    Change-Id: I96285a1902436cf59b6c6ddf22b69a33503d0f4a
    Closes-Bug: #1702935

Changed in nova:
status: In Progress → Fix Released
Revision history for this message
OpenStack Infra (hudson-openstack) wrote : Fix included in openstack/nova 16.0.0.0b3

This issue was fixed in the openstack/nova 16.0.0.0b3 development milestone.

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.