Race condition in scaling policy signal handling

[Thomas Herve]

Saw here: http://logs.openstack.org/51/308051/4/gate/gate-heat-dsvm-functional-convg-mysql-lbaasv1/9e3f37c/

Seems like a specific case of convergence: scaling succeeds, and we delete the stack right away. The scaling policy is still trying to handle its work of setting the metadata, but fails because the resource is already removed.

I believe it's okay to ignore NotFound error in finished scaling when calling metadata_set.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/321637

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/321637
Committed: https://git.openstack.org/cgit/openstack/heat/commit/?id=c789a9837e998af2c2ac974c1c911200685a4dfd
Submitter: Jenkins
Branch: master

commit c789a9837e998af2c2ac974c1c911200685a4dfd
Author: Jaime Guerrero <email address hidden>
Date: Thu May 26 16:06:28 2016 +0000

    Ignore NotFound exception in metadata_set

    To avoid a race condition when updating a deleted stack, this catches the
    NotFound exception specifically and ignores it. All other exceptions are
    allowed to be raised.

    Closes bug: 1582236

    Change-Id: I23ddb71c806c0db1ed7ed4ee4ef831ca51605d65

[Doug Hellmann (doug-hellmann)]

This issue was fixed in the openstack/heat 7.0.0.0b1 development milestone.