yahoo-eng-team team mailing list archive

[Jay Pipes <1779725@xxxxxxxxxxxxxxxxxx>]

Public bug reported:

If a call to ``PUT /allocations/{consumer}`` fails, for example with a
409 Conflict due to a resource provider or inventory being concurrently
updated, the consumer record that is auto-created in the handler before
calling AllocationList.create_all() is not being cleaned up properly.

This results in situations like bug #1778591 where a caller can get
seriously confused when attempting to retry creating allocations for a
consumer, since the retry will now expect a non-null consumer generation
when the API is called with microversion 1.28+

The solution is simple: clean up the auto-created consumer record if a
failure occurs when creating allocations for a *new* consumer.

** Affects: nova
     Importance: Medium
     Assignee: Jay Pipes (jaypipes)
         Status: Triaged


** Tags: placement

-- 
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/1779725

Title:
  Auto-created consumer record not cleaned up after failed allocation

Status in OpenStack Compute (nova):
  Triaged

Bug description:
  If a call to ``PUT /allocations/{consumer}`` fails, for example with a
  409 Conflict due to a resource provider or inventory being
  concurrently updated, the consumer record that is auto-created in the
  handler before calling AllocationList.create_all() is not being
  cleaned up properly.

  This results in situations like bug #1778591 where a caller can get
  seriously confused when attempting to retry creating allocations for a
  consumer, since the retry will now expect a non-null consumer
  generation when the API is called with microversion 1.28+

  The solution is simple: clean up the auto-created consumer record if a
  failure occurs when creating allocations for a *new* consumer.

To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/1779725/+subscriptions