Update compute driver name for Ironic - #1
Open
heidsoft wants to merge 7112 commits into
Open
Conversation
openstack-mirroring
force-pushed
the
master
branch
from
April 30, 2020 20:48
67fd81a to
1147300
Compare
openstack-mirroring
force-pushed
the
master
branch
from
May 27, 2020 06:44
9a6ae34 to
7a0fa4f
Compare
Linux Bridge agent support was removed in the Neutron master branch in [0], let's remove any code here now as well since it will just fail. [0] https://review.opendev.org/c/openstack/neutron/+/927216 Change-Id: Idffa6579000322acfeb860189fb83a317d56bb4e
This change mirrors change Ie02734bb598d27560cf5d674c9e1d9b8dca3801f which ensure that its posible to enable vnc for vms on compute nodes without deploying the vnc proxy. In this change two new flags are added NOVA_SPICE_ENABLED and NOVA_SERIAL_ENABLED to enable configuration of the relevent console create_nova_conf is also modifed to include the db url if the console proxies are deployed on a host. As spice supprot is nolonger avaible in qemu as packged by ubuntu 24.04 and centos 9 a devstack-two-node-debian-bookworm to allow testing with spice in a multinode job. Change-Id: Ie944e518122f2b0059f28acbf68fb7ad0a560ca4
The "logger" mechanism is a testing class that is still calling monkey_patch. This mechanism driver is not relevant nor neccessary for the ML2/OVN CI jobs. Change-Id: I539b202ca81f62f4ae26b5275fd6b245d2066fe7
It may be required to troubleshoot performance related bugs during tests exection, to have ability to inspect environment processes and theirs status during test run this patch installs atop by default if not explicitly disabled. Related-Bug: #2095224 Change-Id: Iedbd61f3ce3cd2255ea5f2a7a93ba2f39ad28ff2
Swift rsync and replication services are up and running, however they fail to replicate any data if needed and if used with more than one replica. This patch removes a deprecated option setting[1] and replaces it with the required setting to use the correct rsync module suffix. Additionally it removes an outdated subdirectory suffix in the rsyncd configuration that has been removed as well[2]. Closes-Bug: #2100272 [1] https://opendev.org/openstack/swift/commit/675145ef4a131d548cc1122689732b9b65e5def4 [2] https://opendev.org/openstack/devstack/commit/0e58d22897457831b9dbf02d66a2f29d43803597 Change-Id: I5283405d00883a4dd11b7c001b1bba3776e576b8
We've LUKSv1 and LUKSv2 tests[1] in tempest that requires the ``extend_attached_encrypted_volume`` option to be True but currently there is no way to set it in devstack. This patch adds the parameter ``TEMPEST_EXTEND_ATTACHED_ENCRYPTED_VOLUME`` to enable the tests. [1] https://github.com/openstack/tempest/blob/cb03598a65f47c51406fc86c9c1503fe42424848/tempest/api/volume/admin/test_encrypted_volumes_extend.py Change-Id: Id3a3483629794ac38cb314812eeff84b677f35cd
If you specify a username, you also need to specify a domain that the user exists in. Failure to do so results in the following error: Expecting to find domain in user. The server could not comply with the request since it is either malformed or otherwise incorrect. The client is assumed to be in error. (HTTP 400)␏ This was mostly being masked for us in python-openstackclient by this little helper in osc-lib [1], but we can't rely on that for openstacksdk (and shouldn't really rely on it elsewhere either). We also deprecate the '--os-identity-api-version' and '--os-volume-api-version' options and will remove them shortly: both services only have v3 APIs nowadays. [1] https://github.com/openstack/osc-lib/blob/3.2.0/osc_lib/cli/client_config.py#L136-L147 Change-Id: I5537b0a7d58efb8a325ed61bad358f677f7a3cdf Signed-off-by: Stephen Finucane <stephenfin@redhat.com>
We have not needed these in years. Change-Id: I4d76a7d3a8513ce5a927d533b34fb609e5dacdfa Signed-off-by: Stephen Finucane <stephenfin@redhat.com>
There is only one volume API and one identity API, and their collective number is 3. Change-Id: Ie269817c5bb0eddd8cfcf279a46cffe4a56377b2 Signed-off-by: Stephen Finucane <stephenfin@redhat.com> Depends-on: https://review.opendev.org/c/openstack/openstacksdk/+/942898
After installation DevStack icons are not displayed. Change-Id: I1bedf97e4d2b7f13b4a0c5b98e29ac53cf502e96 Closes-Bug: #2093844
The race may happen and atop can't start due to missing logs directory. This patch pre-creates directory before starting atop process. Closes-Bug: #2100871 Change-Id: I89e3100dc86d60266913b5c5776db65e8882847c
This section is old and doesn't make much sense anymore, let's remove it. Neutron is already mentioned as a default service and has its own config guide. TrivialFix Change-Id: I2a2ed574f9eca7b87fb9bb6422568ed4fc55f057
... because the file does not affect any functionality. Also apply the consistent irrelevant files to skip functional tests to avoid unnecessary jobs. Change-Id: Ibce79d6b7627c26aa69989ed17ae32d7c3b63d19
stable/2025.1 branch has been created now and current master is for 2025.2. Change-Id: Iba81d280ebf1bd488bd590bdc4e31c49782c7099
Systemd limits the total number of restarts that a service can undergo in a short period of time. On faster nodes all of our apache restarts hit that limit and we eventually fail. Mitigate this by removing unnecessary restarts. Change-Id: I425bb9eec525d82372f05edc63e4fb931e5a4887
Rocky Linux 9.5 64 bit httpd -t [Sun Mar 02 01:10:49.272260 2025] [so:warn] [pid 201497:tid 201497] AH01574: module proxy_uwsgi_module is already loaded, skipping Change-Id: Id6a88c2b7958789f7d4947d3259276f120f5f44e
Tempest is installed in venv-tempest virtual environment so that it always uses the correct upper constraints, etc. On the other hand 'tempest verify-config' is called via 'venv' virtualenv, which does not exist, so tox creates it, using upper constraints from master branch (defined in [venv]deps in tox), which can cause pip install failures if there are any conflicting upper constraint on master branch and the set up requirements. To fix this, the same venv-tempest virtual environment should be used, that was set up previously in the script. Change-Id: Ifb542d6f7d937a8a2020d97c10534467dbdf0051 Signed-off-by: Elod Illes <elod.illes@est.tech>
The user was removed when support for glance-search was removed by commit 020586f . Change-Id: Ic08390933ff7a77789a534bc0e3af751f94312cc Signed-off-by: Takashi Kajinami <kajinamit@oss.nttdata.com>
Depends-On: https://review.opendev.org/c/openstack/devstack-plugin-ceph/+/951810 Change-Id: If366cd10f7858b4238c8ebfad16c129a4cb61a6a Signed-off-by: Rajat Dhasmana <rajatdhasmana@gmail.com>
This patch adds support to configure cinder replication with ceph by executing the following steps: 1. Setting replication_device config in cinder.conf 2. Setting the replication_enabled="<is> True" in ceph volume type Change-Id: Idaee83d051d496e3f9e411c19c0ea7a51bbd76bf Signed-off-by: Rajat Dhasmana <rajatdhasmana@gmail.com>
oslo_policy[enforce_scope] config option was added for the temporary transition to SRBAC and was deprecated. - https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/M72AY5ABQFXQ7XHLVEGHLBBK4XFQGVFK/ To remove it from oslo.policy, we need to remove setting of it from devstack which makes scope enforcement enabled always without providing a way to disable it. oslo_policy[enforce_new_defaults] is still there and will be used to enable/disable the new defaults (until old defaults are not removed). Needed-By: https://review.opendev.org/c/openstack/oslo.policy/+/986475 Change-Id: Ib5afc734b711dd8fa22c6c91b5083888973b2815 Signed-off-by: Ghanshyam Maan <gmaan.os14@gmail.com>
Ubuntu 26.04 (resolute) is the next Ubuntu LTS target for DevStack platform coverage. Add it to the supported distro list and define a non-voting resolute platform job so it can be exercised in check and periodic queues. Resolute uses sudo-rs, which does not support requiretty, and exposes some optional kernel tunables differently. Detect sudo-rs before writing sudoers options, make optional sysfs and sysctl tuning best-effort, and use the C locale for OVN extension sorting to keep setup portable. Disable the noVNC proxy in the resolute platform job until Nova's Python 3.14 noVNC compatibility fix is available. Depends-On: https://review.opendev.org/c/openstack/keystone/+/992782 Assisted-By: pi gpt-5.5 Change-Id: I48518a67f06e2e875bb2fc2034e1517a7e0164b0 Signed-off-by: Sean Mooney <work@seanmooney.info>
Nova has fixed Python 3.14 noVNC incompatibility, so the resolute platform job can exercise the default noVNC deployment again Depends-On: https://review.opendev.org/c/openstack/nova/+/987815 Change-Id: I5892db8e51fc01793a50d904dcd132312f30b4d3 Signed-off-by: Sean Mooney <work@seanmooney.info>
Configure the resolute platform job to run n-novnc-cell1 with OS_NOVA_DISABLE_EVENTLET_PATCHING=true. This keeps noVNC enabled with explicit threaded-mode coverage in advance of nova's default change. Change-Id: I314c001ab01767c1fb281a853135c5d13f653189 Signed-off-by: Sean Mooney <work@seanmooney.info>
The value of `KEYSTONE_ENFORCE_SCOPE` has been changed to `True` by default. This envrionment variable is used to set the values of `enforce_new_defaults` in the `[oslo_policy]` section of the keysonte configuration file. The default value for both oslo.policy options was changed to `True` in the Dalmatian (2024.2) cycle release. [1] But devstack continued to override this value by defaulting to `False`. With this change, keystone will use the same default as newer releases of oslo.policy. [1] https://docs.openstack.org/releasenotes/oslo.policy/2024.2.html Signed-off-by: Douglas Mendizábal <dmendiza@redhat.com> Change-Id: I7edce1519db4fcb38f37082ddf1aa06d440c983b Signed-off-by: Ghanshyam Maan <gmaan.os14@gmail.com>
In RHEL-based environments, the _install_rdo function uses wget to fetch the delorean-deps repository file. However, wget might not be present in a minimal OS installation (e.g., CentOS Stream 10 Cloud images), which causes the download to fail and subsequent tasks like rabbitmq-server installation to break. This patch adds a check to ensure wget is installed using the is_package_installed and install_package helper functions before attempting to download the repository file. Change-Id: I3e99096e2892c2b1bf04e618aa521d665790d27d Signed-off-by: Masayoshi Mizuma <m.mizuma@fujitsu.com>
The argparse help string contains "fe80::1%eth0" where "%e" is interpreted as a printf-style format specifier. Python 3.14 added eager validation of help strings at add_argument() time (cpython commit eb2d268ac7), causing the script to crash with "ValueError: badly formed help string" before it even parses the address. This potentially breaks all IPv6 Zuul jobs on Resolute nodes. Escape "%" as "%%" per the argparse documentation. Closes-Bug: #2158736 Change-Id: Ieda0388548d5964ec9dbc3cff52540a19e9f5625 Signed-off-by: Omer <oschwart@redhat.com>
We are in the process of changing the defaults in each service True and will eventually remove this option entirely. Until that happens, manually force it to True here and remove notes about defaults which are either already out of date or soon will be. Change-Id: Ifd965427dc92db8a590cc3e87e6d6691e11ba344 Signed-off-by: Stephen Finucane <stephenfin@redhat.com> Signed-off-by: Ghanshyam Maan <gmaan.os14@gmail.com>
Cinder uses barbican (via castellan) to manage encryption keys for
encrypted volumes. Some of these operations run as the cinder service
user rather than as the end user -- notably the key transfer performed
during an encrypted volume transfer, where transfer_create stores a new
secret using the cinder service context.
Barbican's legacy policy authorized secret creation via the "creator"
role:
secrets:post -> rule:admin_or_creator
and devstack therefore grants the cinder service user the "creator" role
when barbican is enabled. Barbican's new RBAC defaults instead require
the "member" role:
secrets:post -> True:%(enforce_new_defaults)s and role:member
Now that enforce_scope / enforce_new_defaults are enabled by default,
the cinder service user (which holds only "service" and "creator") is no
longer permitted to create secrets, so every encrypted volume transfer
fails with:
Forbidden: Secret creation attempt not allowed - please review your
user/project privileges
oslo_policy.policy.PolicyNotAuthorized: secrets:post is disallowed by
policy
This breaks the cinder-tempest-plugin *-barbican jobs
(TransferEncryptedVolumeTest.test_create_accept_volume_transfer) across
all cinder changes.
Additionally grant the cinder service user the "member" role in the
service project when barbican is enabled. The existing "creator" grant
is retained so the path keeps working under the deprecated policy as
well.
Change-Id: I1e902b77c73717a2c2b3a727a8f737e1f94193cf
Signed-off-by: Simon Dodsley <simon@purestorage.com>
When the ovn-evpn agent extension is enabled (q-ovn-evpn service), configure the [ovn_evpn] section in the OVN agent config file and set the required OVS external_ids for EVPN operation. New devstack variables: - OVN_EVPN_BGP_AS: BGP AS number for the EVPN extension - OVN_EVPN_BGP_LOCAL_INTERFACE: local interface for BGP peering - OVN_EVPN_VXLAN_PORT: VXLAN port for the SVD parent device (default 4789, sets ovn-evpn-vxlan-ports external_id) - OVN_EVPN_LOCAL_IP: local VTEP IP for EVPN (sets ovn-evpn-local-ip external_id) Also adds the evpn service plugin file, sourced from lib/neutron, which registers the "evpn" neutron service plugin class when the q-ovn-evpn service is enabled. Change-Id: I2c3ee1e6d576098f780cbd949702b06dea570a5b Signed-off-by: Eduardo Olivares <eolivare@redhat.com> Assisted-By: Claude Opus 4.6
Currently, Guru Meditation Reports (green thread stack dumps) are only collected when stack.sh itself fails, via the worlddump.py exit trap. When tempest tests fail due to a service hang (e.g. an eventlet green thread stuck in a privsep RPC call), no GMR is captured and the only way to diagnose the hang is indirect evidence from libvirtd logs, process trees, and LVM metadata. Add a post-run task that sends SIGUSR2 to OpenStack services on failed job runs, triggering their GMR handlers. The task runs before export-devstack-journal so the thread dumps appear in the collected service logs. SIGUSR2 does not terminate the processes. Assisted-by: Claude claude-opus-4-6 2.1.210 Change-Id: I55487cf774c43523fe60f703e2abb170f376803b Signed-off-by: melanie witt <melwittt@gmail.com>
OVS dropped support for building out-of-tree kernel modules after kernel 5.5. All currently supported distros ship kernels well above that version, so ``OVN_BUILD_MODULES`` and the ``build_modules`` parameter in ``compile_ovs`` / ``prepare_for_ovs_compilation`` are dead code. Remove ``OVN_BUILD_MODULES`` from ``ovn_agent``, strip the ``build_modules`` parameter and all its conditionals from ``compile_ovs`` and ``prepare_for_ovs_compilation`` (kernel version guard, ``--with-linux`` configure flag, ``modules_install`` target, and kernel header package installation). Closes-Bug: #2161600 Assisted-By: Claude Opus 4.6 Signed-off-by: Rodolfo Alonso Hernandez <ralonsoh@redhat.com> Change-Id: Ic3d3fc6e2016af6c191ea35eef081b0dec41621a
In [1], the ``compile_ovs`` method removed the first input parameter, named ``build_modules`` inside. This patch is updating the method call with the missing leftover. [1]https://review.opendev.org/c/openstack/devstack/+/998414 Related-Bug: #2161600 Signed-off-by: Rodolfo Alonso Hernandez <ralonsoh@redhat.com> Change-Id: Icef5b10edfecdcc6456133d6b7f88c57b47e14b0
``compile_ovs`` was called with ``--localstatedir=/var``, causing
all compiled OVS binaries to hardcode ``/var/run/openvswitch`` as
the default runtime directory. However, the ``ovn_agent`` plugin
derives ``OVS_RUNDIR`` from ``OVS_PREFIX`` as
``/usr/local/var/run/openvswitch``, creating a path mismatch that
makes ``ovs-vsctl`` unable to reach the running ``ovsdb-server``.
Additionally, ``start_new_ovs`` was unconditionally starting OVS
during the install phase. When OVN is the backend,
``ovn_agent:_start_ovs`` later starts its own OVS instance with a
separate database, but the orphaned ``ovs-vswitchd`` from
``start_new_ovs`` still holds the kernel datapath, preventing
bridge creation.
Fix both issues:
- Remove the ``--localstatedir=/var`` override so the compiled
default matches ``OVS_RUNDIR`` (autotools defaults localstatedir
to ``${prefix}/var``).
- Skip ``start_new_ovs`` when OVN services are enabled, since
``_start_ovs`` handles the full OVS lifecycle.
- Fix the hardcoded DB path in ``start_new_ovs`` to match the
compiled ``--prefix=/usr/local`` sysconfdir.
Closes-Bug: #2158954
Assisted-By: Claude Opus 4.6
Signed-off-by: Rodolfo Alonso Hernandez <ralonsoh@redhat.com>
Change-Id: I8f3a2c5e7d1b4a9f6e0c3d8b2a5f7e1d4c6b9a0e
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
20 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Commit 08448e3 mistakenly updated the location of the Ironic compute
driver to be loaded from within the Nova tree. However, no such module
exists as the compute and scheduler drivers both live in the Ironic tree.
This reverts that and adds a note explaining the setting.
Change-Id: I7f32f77bd7e5fd1f6b41014ac71bbf6e1a7d84cf