openstack/nova
Code Issues Proposed changes

Set instance host and drop migration under lock

Browse Source 
The _update_available_resources periodic makes resource allocation
adjustments while holding the COMPUTE_RESOURCE_SEMAPHORE based on the
list of instances assigned to this host of the resource tracker and
based on the migrations where the source or the target host is the host
of the resource tracker. So if the instance.host or the migration
context changes without holding the COMPUTE_RESOURCE_SEMAPHORE while
the _update_available_resources task is running there there will be data
inconsistency in the resource tracker.

This patch makes sure that during evacuation the instance.host and the
migration context is changed while holding the semaphore.

Change-Id: Ica180165184b319651d22fe77e076af036228860
Closes-Bug: #1896463
This commit is contained in:
Balazs Gibizer
2020-09-28 18:44:02 +02:00
parent 3f348602ae
commit 7675964af8
4 changed files with 35 additions and 33 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
nova/tests/functional/regressions/test_bug_1896463.py
View File

@@ -219,13 +219,4 @@ class TestEvacuateResourceTrackerRace(
server, {'OS-EXT-SRV-ATTR:host': 'host2', 'status': 'ACTIVE'})
self._assert_pci_device_allocated(server['id'], self.compute1_id)
# This is bug #1896463 as the PCI allocation was deleted by the racing
# _update_available_resource periodic task.
self._assert_pci_device_allocated(
server['id'], self.compute2_id, num=0)
# FIXME(gibi): When this bug is fixed (or if you remove the sleeps
# above to avoid the race condition) then we expect that the PCI
# allocation exists on the destination host too.
# self._assert_pci_device_allocated(server['id'], self.compute2_id)
self._assert_pci_device_allocated(server['id'], self.compute2_id)