commit a29a3c9cb16ee0b42f125f8e9b047f22d05c47c4
Author: Matt Riedemann <email address hidden>
Date: Fri Oct 12 12:03:41 2018 -0400
Don't persist RequestSpec.requested_destination
The RequestSpec.requested_destination, similar to the
retry field, is per-request/operation, and persisting
it can caues issues when subsequent move requests.
For example, if you cold migrate a server to a specific
host and then live migrate that server without specifying
a host, the requested target host from the cold migrate
is sent to the scheduler for the live migration, but since
that is where the instance is already running, it's
rejected with NoValidHost.
This is a similar issue to the need to call
RequestSpec.reset_forced_destinations() in all move operations
in conductor. However, rather than try to whack this mole in
every place the request spec is sent to the scheduler, like
reset_forced_destinations() is used, we simply don't need to
persist the requested_destination field since it's just a
vehicle to tell the scheduler which host we want.
The related functional regression test is updated to show
the bug is now fixed.
Conflicts: nova/objects/request_spec.py
NOTE(mriedem): The conflict is due to not having change
Icb295bbd8c83e2e340a7ac3ecc1f159e0db7c7b1 in Queens.
Change-Id: I2a78f0754c63381c57e7e1c610d0938b6df0f537
Closes-Bug: #1797580
(cherry picked from commit ce3af5e33ae6843411e611e81c6ca1c21e0f1e09)
(cherry picked from commit 954b2004a1daa43d4dff0694e6e5cdb9630a441b)
Reviewed: https:/ /review. openstack. org/611945 /git.openstack. org/cgit/ openstack/ nova/commit/ ?id=a29a3c9cb16 ee0b42f125f8e9b 047f22d05c47c4
Committed: https:/
Submitter: Zuul
Branch: stable/queens
commit a29a3c9cb16ee0b 42f125f8e9b047f 22d05c47c4
Author: Matt Riedemann <email address hidden>
Date: Fri Oct 12 12:03:41 2018 -0400
Don't persist RequestSpec. requested_ destination
The RequestSpec. requested_ destination, similar to the operation, and persisting
retry field, is per-request/
it can caues issues when subsequent move requests.
For example, if you cold migrate a server to a specific
host and then live migrate that server without specifying
a host, the requested target host from the cold migrate
is sent to the scheduler for the live migration, but since
that is where the instance is already running, it's
rejected with NoValidHost.
This is a similar issue to the need to call .reset_ forced_ destinations( ) in all move operations forced_ destinations( ) is used, we simply don't need to destination field since it's just a
RequestSpec
in conductor. However, rather than try to whack this mole in
every place the request spec is sent to the scheduler, like
reset_
persist the requested_
vehicle to tell the scheduler which host we want.
The related functional regression test is updated to show
the bug is now fixed.
Conflicts:
nova/ objects/ request_ spec.py
NOTE(mriedem): The conflict is due to not having change 83e2e340a7ac3ec c1f159e0db7c7b1 in Queens.
Icb295bbd8c
Change-Id: I2a78f0754c6338 1c57e7e1c610d09 38b6df0f537 411e611e81c6ca1 c21e0f1e09) d4dff0694e6e5cd b9630a441b)
Closes-Bug: #1797580
(cherry picked from commit ce3af5e33ae6843
(cherry picked from commit 954b2004a1daa43