Debian Patches

Status for nova/2:34.0.0-4

Patch Description Author Forwarded Bugs Origin Last update
fix-failing-timebased-tests.patch Fixed failing timebased tests If the updates are made on the same second, these unit tests are failing.
Adding a time.sleep(2) fixes it.

===================================================================		 Thomas Goirand <zigo@debian.org> no 2025-12-04
testtools-2.8.2-TestMatchersInterface-is-removed.patch Compat with testtools 2.8.2: TestMatchersInterface is removed Thomas Goirand <zigo@debian.org> no debian 2025-12-29
Install-missed-files.patch Install missed files
diff --git a/MANIFEST.in b/MANIFEST.in
new file mode 100644
index 0000000..cf3032d		 Thomas Goirand <zigo@debian.org> not-needed 2021-09-28
remove-svg-converter-from-doc-conf.py.patch Remove sphinxcontrib.rsvgconverter from doc conf.py
===================================================================		 Thomas Goirand <zigo@debian.org> not-needed 2019-09-28
fix-exception.NovaException.patch Fix exception.NovaException
===================================================================		 Thomas Goirand <zigo@debian.org> no 2024-11-19
weights_debug_log_stop_using_tuples_as_dict_keys.patch weights debug log: stop using tuples as dict keys While Python supports using tuples as dict keys, json.dumps() (which
is what oslo_serialization.jsonutils.dumps() calls under the hood)
does not. Stop using tuples as keys in our weight debug log.

diff --git a/nova/weights.py b/nova/weights.py
index 75d51dc..7983a90 100644		 Artom Lifshitz <alifshit@redhat.com> yes upstream https://review.opendev.org/c/openstack/nova/+/904060 2026-05-18
Retrieve_the_image_metadata_from_instance.image_meta_during_resize.patch Retrieve the image metadata from instance.image_meta during resize In the current process, image metadata is obtained from `request_spec`
during resize or migrate, retaining the values from when the instance
was initially launched. As a result, if the image metadata is updated
and the instance is rebuilt, the latest values are not retrieved from
`request_spec`. This patch addresses this issue.
Additionally, it also resolves the following problem:
Changing the `hw_machine_type` of an image from None to a specific
value and rebuilding the instance causes the `image_hw_machine_type`
in `instance_system_metadata` to revert to its default value during
resize/migrate.

===================================================================		 Masanori Kuroha <mkuroha@lycorp.co.jp> yes upstream upstream, https://review.opendev.org/c/openstack/nova/+/936910 2026-05-26
Stop_periodic_tasks_during_shutdown.patch Stop periodic tasks during shutdown The oslo.service periodic tasks runner expects that tasks are returning
a number of seconds indicating when to run next time. During graceful
shutdown we want to stop the periodic tasks but we returned None which
breaks the protocol and triggers a stack trace. Instead this patch makes
sure we raise LoopingCallDone to signal that no more tasks needs to be
run during shutdown.

===================================================================		 Balazs Gibizer <gibi@redhat.com> yes upstream upstream, https://review.opendev.org/c/openstack/nova/+/1005880 2026-09-20
Add-a-healthcheck-endpoint-to-the-metadata-API.patch =================================================================== no
Lock-the-info-cache-row-in-instance_info_cache_update.patch Lock the info cache row in instance_info_cache_update MariaDB 11.x enables innodb_snapshot_isolation by default (MDEV-26642),
making InnoDB detect write-write conflicts under REPEATABLE-READ: an
UPDATE fails with ER_CHECKREAD (1020) when the row was modified by
another transaction after the current transaction took its snapshot.
.
instance_info_cache_update() reads the row then updates it in the same
transaction, so a concurrent cache update (for example the Neutron vif
events racing the delete flow) can commit in between. The failing
UPDATE also discards the nested savepoint, so the savepoint teardown
raises ER_SAVEPOINT_NOT_EXIST (1305), and that DBError escapes both the
DBDuplicateEntry handler and wrap_db_retry, leaving the server in
ERROR state during delete (observed in the OpenStack Debian CI against
MariaDB 11.8.6).

===================================================================		 Thomas Goirand <zigo@debian.org> yes 2026-10-16

All known versions for source package 'nova'

Links