yahoo-eng-team team mailing list archive

[Erik Olof Gunnar Andersson <eandersson@xxxxxxxxxxxx>]

Public bug reported:

This patch that was backported to Mitaka made the assumption that free_ram_mb can never be less than 0.
https://github.com/openstack/nova/commit/016b810f675b20e8ce78f4c82dc9c679c0162b7a

free_ram_mb can be negative when the ram_allocation_ratio is set higher than one. This is also supported by the following comment in the code for the resource tracker.
> # free ram and disk may be negative, depending on policy:

This breaks any Scheduler filter that supports oversubcribe (e.g. memory or disk), as an example see this unit-test for the ram filter.
https://github.com/openstack/nova/blob/master/nova/tests/unit/scheduler/filters/test_ram_filters.py#L43

** Affects: nova
     Importance: Undecided
         Status: New

-- 
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/1635367

Title:
  Ram filter is broken since Mitaka

Status in OpenStack Compute (nova):
  New

Bug description:
  This patch that was backported to Mitaka made the assumption that free_ram_mb can never be less than 0.
  https://github.com/openstack/nova/commit/016b810f675b20e8ce78f4c82dc9c679c0162b7a

  free_ram_mb can be negative when the ram_allocation_ratio is set higher than one. This is also supported by the following comment in the code for the resource tracker.
  > # free ram and disk may be negative, depending on policy:

  This breaks any Scheduler filter that supports oversubcribe (e.g. memory or disk), as an example see this unit-test for the ram filter.
  https://github.com/openstack/nova/blob/master/nova/tests/unit/scheduler/filters/test_ram_filters.py#L43

To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/1635367/+subscriptions