Change-Id: I43a04fa8820bb3c1578f574d286cc1b9ca87ad30
Signed-off-by: OpenStack Release Bot <infra-root@openstack.org>
Generated-By: openstack/project-config:roles/copy-release-tools-scripts/files/release-tools/functions

Update the URL to the upper-constraints file to point to the redirect
rule on releases.openstack.org so that anyone working on this branch
will switch to the correct upper-constraints list automatically when
the requirements repository branches.

Until the requirements repository has as stable/2026.1 branch, tests will
continue to use the upper-constraints list on master.

Change-Id: I3c1a9310782ac2968403e94998f87855807552d5
Signed-off-by: OpenStack Release Bot <infra-root@openstack.org>
Generated-By: openstack/project-config:roles/copy-release-tools-scripts/files/release-tools/functions

The UT test_thread_local_sessions_are_isolated fails intermittently
and causes zuul to fail

unit test job failures are observed due to below test:
cinder.tests.unit.volume.drivers.hpe.test_hpe3par.
TestHPE3PARDriverBase.test_thread_local_sessions_are_isolated

Failures are intermittent and seen mostly on the openstack-tox-py313 job

test is introduced by change If3deb4a2791a05b21aa8
https://review.opendev.org/c/openstack/cinder/+/976596

This patch skips the test "test_thread_local_sessions_are_isolated".

Related-Bug: #2146339
Change-Id: I11be960fc8d95e7e193bfb7bc0ef6277c54f62fe
Signed-off-by: Raghavendra Tilay <raghavendra-uddhav.tilay@hpe.com>
(cherry picked from commit 9d290e8)

The changes to fix LP#1527324 and LP#1823445 have as side effect that
'properties': {} may be sent to Glance when creating an image from a
volume which fails. This change ensures that the code in
_translate_to_glance() effectively removes the 'properties' key from
the metadata dict if it is empty.

Closes-Bug: #2144550
Change-Id: Ifd632433ce5759494172b74ce311816e10728d21
Signed-off-by: Alejandro Santoyo <alejandro.santoyo@canonical.com>
(cherry picked from commit 2096a78)

ONTAP 9.15.1 changed the default LUN space-allocation setting from
disabled to enabled. The block driver was unconditionally creating
LUNs with space-allocation disabled (unless the
netapp:space_allocation extra spec was set), causing inconsistency
between Cinder-managed volumes and ONTAP's own defaults.

The driver now checks the ONTAP version when no extra spec is present:
- ONTAP >= 9.15.1: space-allocation enabled (matches ONTAP default)
- ONTAP < 9.15.1: space-allocation disabled (preserves existing behaviour)

The explicit netapp:space_allocation extra spec continues to take
precedence over version-based defaults.

Closes-Bug: #2152031

Change-Id: I4dd3bbb1d634befe02cdcde0a20fab7f2f4c046a
Signed-off-by: hongp <inyong.hong@samsung.com>
(cherry picked from commit fd7d056)

Restoring a backup to a new volume sets discard_zeros, so _transfer_data()
skips all-zero chunks. The Ceph driver skipped them without moving the
destination offset, so every following non-zero chunk was written too
early and the restored volume came back shifted and corrupted.

An encrypted (LUKS) volume hits this reliably: it is sparse, the restore
takes the full non-diff path, and a freshly created target sets
discard_zeros=True. The restore reports success and the corruption only
shows up later when the guest cannot mount the volume.

Advance the destination offset by the number of bytes just read whenever
a zero chunk is discarded, the same bookkeeping the chunked driver
already does on its non-zero writes.

Closes-Bug: #2155612
Change-Id: Idb7b709ba3f6c42ab237e717b0e58db365a7567c
Signed-off-by: kurtyoon <kr.dev.kurtyoon@gmail.com>
(cherry picked from commit 8d4a9e9)

py-pure-client 1.86.0 (FlashArray REST API 2.52) changed the Patch
models so that ``VolumePatch.qos`` and ``VolumeGroupPatch.qos`` no longer
accept a flat ``Qos`` object with integer ``iops_limit``/``bandwidth_limit``
values. They now require a ``QosPatch`` whose limits are wrapped in
``QosIopsLimitPatch`` and ``QosBandwidthLimitPatch`` objects.

As a result, any operation that set QoS through a volume *patch* -- most
visibly creating a volume from a snapshot, which always patches the new
volume's QoS -- failed with::

    pydantic.v1.error_wrappers.ValidationError: 2 validation errors for
    VolumePatch
    qos -> bandwidth_limit
      value is not a valid dict (type=type_error.dict)
    qos -> iops_limit
      value is not a valid dict (type=type_error.dict)

leaving the volume in ERROR state.

Build the QoS object for the ``*Patch`` requests with the new
``QosPatch``/``QosIopsLimitPatch``/``QosBandwidthLimitPatch`` models via a
small helper. The ``*Post`` models (``VolumePost``/``VolumeGroupPost``)
still take a flat ``Qos`` with integer limits in all supported SDK
releases, so those call sites are unchanged. The minimum py-pure-client
version is raised to 1.89.0, which provides the new models.

Add a unit test that exercises ``set_qos`` across the four IOPS/bandwidth
combinations and asserts the ``QosPatch``/``QosIopsLimitPatch``/
``QosBandwidthLimitPatch`` models are used (and the flat ``Qos`` is not),
so a regression back to the flat ``Qos`` form is caught.  This test was
contributed by Jon Bernard as a follow-up change and squashed in here.

Closes-Bug: #2158313

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Co-Authored-By: Jon Bernard <jobernar@redhat.com>
Change-Id: Ife6a7f42778c065a70af4752e12661cec34bf7b2
Signed-off-by: Simon Dodsley <simon@purestorage.com>
(cherry picked from commit 635e41d)

We merged a second entry for "packaging" to
requirements.txt, now the requirements-check
gate job is failing with:

ERROR: Package 'packaging' requirement does not match number of lines (1) in openstack/requirements

Change-Id: I3b5c62aa356f6bcffe2d927c18c3b8c24c68ad54
Signed-off-by: Eric Harney <eharney@redhat.com>
(cherry picked from commit cee90b0)

…able/2026.1

When deleting a group with the delete-volumes flag set to True,
the volumes associated with the group are deleted, but the
'volume.delete.end' notifications are not emitted. This occurs
because the group delete flow in the manager does not trigger
the standard notification logic used in individual volume deletes.

This patch ensures that 'volume.delete.start' and
'volume.delete.end' notifications are correctly sent for each
volume processed during a group deletion.

Closes-Bug: #2139068
Change-Id: I2abff84798f54f207ce53ab5afedf41e26348648
Signed-off-by: hongp <inyong.hong@samsung.com>
(cherry picked from commit cb63d6f)

…tion" into stable/2026.1

…stable/2026.1