This reverts commit 30b484312f.
Reason for revert:
The libvirt bug has been fixed in libvirt 10.0.0.
Change-Id: I62acbb437e894bf3fefb19bbb533aa8ba13a6bcf
The scenario assumed, there is an image in glance, which may
not always be the case. This patch creates a new image based
on path configured in CONF.scenario.img_file. To add this
functionality the TestTelemetryIntegrationPrometheus class
was moved to inherit from tempest.scenario.manager.ScenarioTest.
Requirements on glance and ceilometer were added too. Ceilometer
was required even before this change, but the requirement check
was missing.
The scenario assumed the ceilometer pipeline interval is less
than 2 minutes. This patch adds a config option for this.
By default 300s is used, which is the default interval when
installing devstack. It's then changed to 15s in .zuul.yaml,
which is the interval used in our tests.
Change-Id: Ifae664540fc5a749b2965822267d7c54b34a77a5
This repository was retired as part of retirement of TripleO.
Also filter out some of the failing scenario tests to unblock gate.
Co-Authored-By: Erno Kuvaja <jokke@usr.fi>
Change-Id: Ida1c84c6db338fa5cc7f842f4bf16b5613acdc9f
Aodh can use only one database so having multiple options per backend
software is very redundant.
Also, gnocchi and prometheus are used for the same purpose so can use
a single option instead.
Change-Id: I134cf61d8137409ef5f5d959177e349781110eaa
We have suspected autoscaling tests failing, because the "gabbilive" archive policy can't be
deleted in time. Yet the recent experiments does not support it. This bumps the number of
retries from 360 to 1000 attempts to avoid this happening in the gate where the environment
can be slightly different.
Change-Id: I0936c9c6a12af0ea8ca2d8cb24f0807132125d45
This reverts commit 34e514a482.
In general, we'd want to check autoscaling from end to end.
Until now this plugin could only scale up the stack by adding
an instance. With this change, scaling down of instance is
also tested using a scaledown policy and an alarm which
is triggered if instance CPU usage falls below the threshold.
Depends-On: https://review.opendev.org/865386
Depends-On: https://review.opendev.org/868105
Depends-On: https://review.opendev.org/c/openstack/telemetry-tempest-plugin/+/904746
Co-authored-by: Yadnesh Kulkarni <ykulkarn@redhat.com>
Change-Id: I7f19dcaf327b636ef8328c6634d295414b7c8e65
Adding telemetry_services config group.
Deprecating old gnocchi and sg-core service options.
Splitting the backend specific tests into their own paths so they
can be disabled based on configuration.
Change-Id: I4ca0e217998825589f755d1169aa0600cef14248
If we agree, that we don't need to test anything older than
wallaby, then we can clean .zuul.yaml like this. After we drop
wallaby in the future, we can remove even more.
Change-Id: Ie437138a5befde3fe559fc9e75b7a5772f543f98
The sg-core is not required to run Telemetry services. Enabling tests
requiring sg-core by default is not backword-compatible and affects
testing in distributions(eg. RDO) or any users using tempest to
validate their deployments.
Change-Id: Ib08a42f22325d4029b5148f081e1b095fbfc723e
We have an issue, that tempest doesn't distinguish between
branches. New features aren't available on old branches and
so we can't execute testing, which involves sg-core or prometheus
on old branches. Separating sg-core and prometheus related tests
allows us to use tempest regex to execute those tests only on
branches, which have the required features.
This also fills in service_available requirements, to make it
clear which tests are runable on prometheus based storage
and which are runable on gnocchi based storage.
Change-Id: Ic5bea08295d8c85303c676706415e093625ee16f
* Devstack recently introduced the functionality to use a single venv[1]
but some devstack plugins in telemetry project are not compatible with
this change.
This disables the option as a short term workaround.
* Tempest CI job "telemetry-dsvm-integration-train" for Train release
fails due to
```
Obtaining file:///opt/stack/nova
nova requires Python '>=3.8' but the running Python is 3.6.9
```
Train release of Nova and Heat projects have reached EOL which are
fetched while running this job. If any issues are identified in these
releases fixing them might not be possible.
[1] https://review.opendev.org/c/openstack/devstack/+/558930
Change-Id: I01ed2c96395baf294aa84e4f989ab82f9b387fc4
This option was deprecated in favor of the new aodh option. However
the old name is still used by code which causes the following warning
message.
WARNING oslo_config.cfg [-] Config option service_available.aodh_plugin
is deprecated. Use option service_available.aodh instead.
Change-Id: If10cec87a2f07f9dbe09005bb22070ca7a6e9364
The latest release of gnocchi is not installable
due to the protobuf version we use in upper constraints.
This change adds gnocchi as a required project so that
it will be cloned by Zuul and added to LIBS_FROM_GIT
Depends-On: https://review.opendev.org/c/openstack/devstack/+/872902
Change-Id: Ie70fec0a79b8c9ba1354cf32d6740ac710976c5c
For autoscaling granularity is set to 60 and the heat
stack is expected to scale out after instance reaches
CPU utilization higher than 10%.
According to this configuration the cpu time must be
10**9*60*(10/100) which is 6000000000 ns.
The default threshold for aodh alarms is "10000000"
which comes from ceilometer's devstack settings.
This change overrides the aodh_threshold value in
tempest.conf to the correct cpu time as per
cpu util and granularity values.
Signed-off-by: Yadnesh Kulkarni <ykulkarn@redhat.com>
Change-Id: Ic90613b70556d0ce5bcf64c2305a3351928bf83e
Disable SSL validation to prevent autoscaling tests
from failing with error
"certificate verify failed: unable to get local issuer certificate"
Set parameter "disable_ssl_certificate_validation" to True
which was introduced in 959fc69e49
Signed-off-by: Yadnesh Kulkarni <ykulkarn@redhat.com>
Change-Id: Iaa61eebf0e652b60182f893bba009b30d4734f95
In 2023.1 cycle. we are moving the default distro
version of Ubuntu to Jammy (22.04)[1] so we need to pin
the nodeset for stable branch job in master gate so that
they continue run on their supporting distro version which is
Ubuntu Focal since stable/victoria.
[1] https://governance.openstack.org/tc/goals/selected/migrate-ci-jobs-to-ubuntu-jammy.html
Change-Id: Ia3f329f9f0e4d8f7541d002ad7c8e77083b44bca
Add a variation of the telemetry-dsvm-integration-centos-8s jobs using
branch selectors and provide a config for python 3.6 on older branches
(i.e. pre-yoga)
Change-Id: Ia57b83c7d764e2bb4eda19a46034ff2d3353aff7
This patch sets a zuul variable "nslookup_target" to 'opendev.org'
to fix the FIPS jobs.
The nslookup_target var was added as a required var for the enable-fips
role in [1].
[1] 25caf7ef5e
Change-Id: Ifb0f929cf7cb9dddedcf4f0bd930b4bd7ca2dd72
The latest cinder requires Python >=3.8 which is not available in
CentOS Stream 8. This bumps the base OS version to 9 which has Python
3.9.
Note CentOS Stream 8 jobs are still left now but will be removed once
job definitions in aodh/ceilometer are updated.
Change-Id: I6add67a23a5d39852a1cdca9e2baf3faa1ac0cbb
The failing part of the integration test was disabled until
we figured out what exactly fails and why. In the mean time,
we can take advantage of the rest of the integration tests.
Change-Id: Iaefaa06b37deb5ecf2108c027ac7db98130c8c98