Because sahara builds on the compute base layer, it does not need to be
tested in all devstack-gate jobs. Instead create a new job that tests
sahara in a devstack environment as a first step in removing it from the
standard configuration.
Specify that SAHARA should be enabled in the devstack updown job
so that the parts that enable sahara are exercised in devstack.
This is already currently the case, but this updates the job config
so that it will continue to do so later when we remove sahara from
the default config.
Change-Id: Iae4e2439bfdfe7f3fc5f0b75b38583a70d7d78f4
Co-Authored-By: James E. Blair <jeblair@hp.com>
The largeops-src test is used for libraries that can break largeops
(oslo.messaging, oslo.db etc.) and is hitting mysql's connection limit
periodically. Stop running services that large-ops doesn't test to
mitigate this.
Change-Id: I8ab0cee07cbc0649eda37b686a67b81a46586fcb
Partial-Bug: #1403284
This should be OK to run full tempest tests. We've sorted out a
number of issues lately relating to memory and swap usage and making
sure package updates are installed.
Change-Id: I007c7c7ba50f8ea5ab1afe69ab52f6d1be727966
Getting the cells Tempest job to green has been a long process because
there have been constant regressions to keep up with. In order to get
ahead of that the final failing tests have been blacklisted and
annotated with open bugs for them. Work will continue to address them.
Change-Id: Iffccba724c3c45b33a134355fc681e3497b84b3a
aiopcpu is an exceptionally cryptic name that doesn't mean anything to
most people reading test results. multinode is what the job is, and is
much more understandable at first glance.
This changes the naming from aiopcpu => multinode. Should we create a
different kind of multinode test in the future, we can come up with an
appropriate name for it then.
This is safe to change because the devstack-gate logic actually is
fully != 'aio', not == 'aiopcpu'.
Change-Id: Ic7fe31a397af8a14ecf33a4902908ff9c03d11d3
The ec2 test_compute_with_volumes test will not pass with cells since it
relies on security groups which are not supported. It should be skipped
until such a time as cells supports security groups.
Change-Id: I0f29efdd3853ec3038f5944607e9bec9a7615a14
There is a bug in the current regex for excluding slow tests.
Instead of excluding the slow tests, it is excluding every test
except the slow tests -- the inverse of what we want.
Change-Id: I82ccddd1dee87fc6153476419ca15309b7cc9529
This commit fixes a previous oversight to the non-admin tempest job
definition. Previously the job was configured to run the smoke tagged
subset of tempest in parallel. However we need to run the full suite
serially with the non-admin job. This adds the missing flag to the
definition.
Change-Id: Ie54d282d222b79dcc36fd3087665888f5094bf36
Since several patches have merged, the set of tempest test expected
failures with cells have changed. This patch adjusts the list of
tests that should not be run in order for the job to pass.
The tests being removed from the blacklist had been failing in the
past because a flavor created by tempest could not be accessed in
child cells during instance create. Patches have since merged that
enable use of the flavor object stored in the instance object,
eliminating the need for a flavor lookup from the database. So tests
having to do only with an instance create can be removed from the list.
There are some classes of tempest tests still expected to fail in a
Cells V1 environment:
* Networks API
* Resize instance
* Floating ips
* Shelve instance
* Rescue instance
Change-Id: If32b7361521d4747dcd05db47e3bf4955ce1b232
A new macro for capturing all of devstack's logs, the console log and
documentation has been added so we should use that.
Change-Id: I79047847daaa86c63847aedf5e7e808891ef3812
This will make the Neutron's aiopcpu job consistent
with the settings for the nova-networkt aiopcpu job
where multi-node is enabled.
Depends-On: Id8d16877f09a1128c10f5c52438f6b7726c01f8e
Change-Id: Ic90617665feea513a1d0a43850c8e98b33c7a5a0
This commit switches to using a non-existent devstack-gate option
(which was going to be added in an abandoned patch) to using localrc
passthrough to set the option to disable setting the admin creds in
the tempest.conf for the non-admin job.
Change-Id: I21e874231a2265d0665b147dcbd87138d17352c5
Now that tempest-dsvm-neutron-pg-full is passing, make it voting and
drop tempest-dsvm-neutron-pg, since its just a subset of full.
Change-Id: I08d423c50d1af3314838b4e9622302346149cc31
This job is hitting the test timeout, causing it to fail. This change
adds the slow tag exclusion regex for the job.
Change-Id: I958a8ec49b08c00e3aff8c76aa293e09187719a7
This patch is a follow up from 49d059760a. It's creating a new job
pxe_ipa-src that builds the IPA ramdisk from source and it's enabling
running this job (as non-voting) on the openstack/ironic-python-agent
project.
Change-Id: Ia56d2808a9c2d61f737e3ee86aa33a06f2967538
Because heat builds on the compute base layer, it does not need to be
tested in all devstack-gate jobs. Instead create a new job that tests
heat in a devstack environment as a first step in removing it from the
standard configuration.
Specify that heat should be enabled in the devstack updown job
so that the parts that enable heat are exercised in devstack.
This is already currently the case, but this updates the job config
so that it will continue to do so later when we remove heat from
the default config.
Change-Id: I13a80cc5b669f30f55dae67fd82eb6e3f31f9dbc
This patch modifies the job template for devstack plugins by
adding '{pipeline}' and '{branch-designator}' to the job name,
and passing those values as part of 'project' definition in
jenkins/jobs/projects.yaml file.
Also modified job name in Cinder, devstack-plugin-glusterfs
project's check pipeline to "check-tempest-dsvm-full-glusterfs-nv".
And this patch also creates a new voting job
"check-tempest-dsvm-full-glusterfs" for "devstack-plugin-glusterfs"
project.
Co-Authored-By: Deepak C Shetty <deepakcs@redhat.com>
Change-Id: I5939b7b5d2fb3cb29f22e38b1aa392d810c0feec
The dsvm-tempest-neutron-src job is meant to test clients from source, however,
the devstack job it deploys does not actually deploy ironic anywhere
and thus does not actually test client changes. This closes the CI hole by
adding new tempest-dsvm-python-ironicclient-src job that runs the ironic dsvm
Tempest job with python-ironicclient from source. It is non-voting until it
proves to be functional. Once it is, we can follow this up with another commit
to make it voting.
Change-Id: I4dca75e63fef12ebc0790b8e55bcae526c48d605
Add a non-voting job that runs only the database tempest tests
and add it to trove and tempest. This will replace the existing
full runs of tempest on trove for those projects.
Change-Id: I4df41bfff85ccfc2f61a80df8d2e1bb0ea3c7ff9
This drops the ugly tempest regex we were using to target the subset of tests
we knew to be functional against baremetal, and instead uses the 'baremetal'
service tag to run only relevant tempest tests.
This updates the parallel and grenade jobs to do the same. Previously, these
jobs were running the entire Tempest suite against baremetal. This takes a
very long time and runs many tests that are not really relevant.
This is a shift in a previous effort to enable running the entire Tempest suite
against Ironic to running only the smallest subset of tests we care about.
Also, this enables tenant isolation where it is disabled, since that is now
functional against Ironic.
Change-Id: I1a979c644b4950f0fce59c764b115266b8a13ca8
Because trove builds on the compute base layer, it does not need
to be tested in all devstack-gate tests. Instead, create a new
job that tests trove in a devstack environment as a first step
in removing it from the standard configuration.
Specify that TROVE should be enabled in the devstack updown job
so that the parts that enable trove are exercised in devstack.
This is already currently the case, but this updates the job config
so that it will continue to do so later when we remove trove from
the default config.
Change-Id: Ib503e45b08edfb2252d7c56af14f97d630e137d8
Add non-voting tempest dsvm job for Ironic project. This job is intended
to test the pxe_* drivers running with the IPA ramdisk.
Change-Id: I2a06c07a6280a69dd5730fd692a0d6bf5784aa59
Implements: blueprint ipa-as-default-ramdisk
This macro allows a job running on a cloud-based instance to print the
UUID of that instance to the build log. By the time it is determined
that a test failed due to a system or networking error, the instance
has already been deleted. The UUID can be used to locate the physical
system the instance ran on so that analysis can be done at that level.
By providing it in the build log, users will not have to find someone
with access to the nodepool logs to get this level of support for this
type of issue.
Depends-On: Ib84ec30b5c4c5de5d5be80cae2b4230c1aba3bc6
Change-Id: Iddcd56e2e2f2a942573a9868d1c009106b38564a
This patch ensures that the job "tempest-dsvm-full-glusterfs" to
run on ubuntu instead of centos.
And this patch also adds tempest-dsvm-full-glusterfs job as a
experimental job to stackforge/devstack-plugin-glusterfs project.
Change-Id: Ic24daa174fdbd09d20433673cd1e61d6da2e3243
In Ironic we want to be able to use the IPA ramdisk as the default ramdisk
for our deployments. Since that ramdisk is a heavier than our previous one
we need to bump the memory size from 512MB to 1GB to be able to test it
in gate. It's important to note that since the concurrency job needs 3 VMs
we will still run it using the old ramdisk until we can solve the problem.
Plus, for normal jobs where we bumped the RAM size we are lowering the
number of VMs created from 3 to 1, because only 1 VM get's used anyway.
This patch is making use of the new DEVSTACK_LOCAL_CONFIG that allows
passing arbitrary config options to devstack-gate.
Change-Id: Icf011d995640cabcb3677f814b7dd37aab0a303b