Age | Commit message (Collapse) | Author | Files | Lines |
|
|
|
Change-Id: I09978a08d00c561f17ce3a7d5ba75d80aa49a0f6
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
If there is a CI failure, then we need the VM to be present so we can
ssh to the slave and collect all the logs we can in order to debug
the problem. However, in order to not leave stale VMs around, we should
remove the VM if a job is scheduled again on the same node. As such, if
we need to debug a build on the node, we need to disconnect the node
from Jenkins so the VM is not wiped by a future job.
Change-Id: I93e7d7402fe03c392b63b87e7d438191f6e7e9cc
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Please note that these triggers are temporary and they will be
adjusted to fit promotion based triggering. This will result
in running scenario deployments and testing if any scenario
has changes. Otherwise, daily jobs will be skipped.
Change-Id: I24e8c4d7f1da4d3203d81a8929ce2bc60d398ece
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Daily jobs run on the actual slave so they leave virtualbmc processes
running even when the job is finished. As such, we need to make sure
that such artifacts are gone as soon as we start a new job.
Change-Id: I82f49279cc105e788db466e5040e6a31c537af5c
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
|
|
|
|
Change-Id: Ic3f31fea8170af8b1e9845856976ef57df496e1e
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Change-Id: I206b729c09d30eb81432d11638c2654a96c3b9e8
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Change-Id: I958c8f857c28a45c009daba36c4bcef7adda4bbe
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
|
|
The script does not behave as it should so we need to enable the debug
option in order to debug it properly
Change-Id: I13cd4d027c0d9bb9229466148ea4c913508cba2a
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
There are no scenarios in releng-xci anymore so we must not try and
determine the impacted scenario. Using the default one is good enough
for CI. Users can still override the defaults using the usual methods.
Change-Id: I47207ebddae79fd89615bfd664f211879aabae92
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
It also fixes the results directory where Functest log is cat from.
Change-Id: I3d8ea06d3741c70108cdb1c7f040969dbacf79a5
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Change-Id: Id48d1dcb6c1998f7f6d0f71cb9380ed7fcf7b320
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
In order to activate more test tiers from functest and also
start running yardstick, we need to do the test preparation
at one go and then run what's supposed to be run.
This change is required for below changes to be verified
successfully
https://gerrit.opnfv.org/gerrit/#/c/61639/
https://gerrit.opnfv.org/gerrit/#/c/61641/
Change-Id: Ib3ae2241921383b77c23b05443325aa6240cc176
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Change-Id: I12e1dc7d79ca74df1ba02e16e55e91f4f47a818c
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Change-Id: I2cb52d85a86bf70dfaf151606138d671e618e5d0
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Functest preparation is moved into the deployment job to ensure
things are prepared before we are done and functest job gets
triggered. Making the preparation in functest job will cause
us to either reinitialize venv or depend on deployment job directory.
Apart from that, the env vars necessary for functest are available
within deployment job anyways so reusing that will make running
functest simpler.
See the comments in xci-daily-jobs.yaml builder macros.
Change-Id: I4c5bbe5f4b31bb620986d75ba78de6c0fe442975
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Change-Id: Ie1d652607f8f56f5e1cf0409dc628b1add615cd9
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
The execute shell parts need to be switched to existing
scripts but since the scripts were mainly written for
verify jobs, the adaptations they require has potential
to cause issues so it needs to be done in a separate
change.
Apart from that, the verify and daily jobs can't be run
on same slave at the moment due to a bug in opnfv vm
creation playbook. If clean vm exists on the node, the
playbook fails. Build blocker is enable for verify and
daily jobs to ensure the jobs don't end up on same slave
for the timebeing.
Change-Id: Iefaf5a5c2d4a79bd48a5f9da28d68748bc67f511
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Due to Intel lab maintenance, the number of slaves will go down so we need
to have fewer jobs running. Centos is known to be problematic upstream and
it is better to disable them until the lab is back online to ensure the jobs
for other distros are impacted less.
Change-Id: Ifd8624e45282d68bd7acdf283ef7a499e81990e9
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Change-Id: I62095327781153eaff81999f5f3fdddc652dc550
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
JJB 2.0.3 changed the postbuildscript syntax a bit and requires
changes so that they work properly with version 2.x of PostBuildScript
plugin.
This patch requires that Jenkins also be updated to install the 2.x
release of PostBuildScript.
Change-Id: I65e52c8d9b2efa38a3d16193c176c65e87e704c1
Signed-off-by: Thanh Ha <thanh.ha@linuxfoundation.org>
|
|
Earlier change took the possibility of dynamically determining the default
scenario. This change puts part of it back which is based on the changed
installer and/or other changed files.
Change-Id: Ie6d9a9b0622ac4ae5ff0cf0e679fd26beed51747
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
|
|
|
|
Previously we were determining the impacted generic scenario by looking into
changeset in releng-xci. Since we moved the scenarios to releng-xci-scenarios,
this is not needed anymore and same function can be used for determining the
scenario.
Change-Id: I06a65599e03dff4016d0d9d78116e505d84b5061
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
The WORKSPACE variable is always set so we do not need to check it
in the script.
Change-Id: I33d2e17b9a4a7b262c781c7442cc69c2433f8729
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
global-jjb only supports the .yaml file ending for jjb jobs. Instead of
waiting for a release we're going to rename the files.
Change-Id: Icf3339eacd2320c583333e02250998cf6b1881f7
Signed-off-by: Trevor Bramwell <tbramwell@linuxfoundation.org>
|
|
We no longer have static inventory files per installer and flavor
so we need to switch to the dynamic inventory option.
Change-Id: I5b8bf800264b1b297b61d09d010552cd22636627
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
The scenarios have been moved from the releng-xci to
releng-xci-scenarios repository so we should connect Jenkins
to that repo as well.
Change-Id: Ifcb4f02b8532dd94c8a622185328f7316e726c6b
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
When passing a YAML variables file to Ansible, we need to prefix it
with '@'.
Change-Id: I884f4698a1033cca9ad55bf3592e1dd817273f81
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
The gerrit variable for the git SHA is GERRIT_PATCHSET_REVISION instead
of GERRIT_CHANGE_ID. We also need to export the XCI_ANSIBLE_PARAMS
variable.
Change-Id: I206820f1155a4876fb7d87cc0b5ae1b8afb06bd6
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Pass scenario information using the new override mechanism.
This patch depends on the matching releng-xci change
https://gerrit.opnfv.org/gerrit/#/c/56643/
Change-Id: Ib8473249e424f259f5757ebdaed7d73fa4b0adce
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
|
|
Commit 33073c5a3426c1867411db446bc9e5d84d2a7747 ("jjb: xci:
xci-run-functest.sh: Ensure env vars are quoted during reload") added
support for reloading the OPNFV host environment in a safe way but that
didn't work because the internal '$var' variable was not escaped so it
was evaluated on the local host instead of the remote one. This fixes
the problem by properly escaping the variable.
Change-Id: I7c90ae18f77aca076579f70c7c0e9a72aaaad7c8
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
The output of the 'env' command does not use quotes when printing the
env variables so trying to source the xci.env file will lead to problems like
the following one:
/home/devuser/releng-xci/.cache/xci.env: line 17: PROFILEREAD: readonly variable
/home/devuser/releng-xci/.cache/xci.env: line 20: 50340: command not found
/home/devuser/releng-xci/.cache/xci.env: line 40: controller00: command not found
/home/devuser/releng-xci/.cache/xci.env: line 48: on: command not found
/home/devuser/releng-xci/.cache/xci.env: line 55: Francisco/O=IT/CN=xci.releng.opnfv.org: No such file or directory
/home/devuser/releng-xci/.cache/xci.env: line 75: a: command not found
/home/devuser/releng-xci/.cache/xci.env: line 88: -I: command not found
/home/devuser/releng-xci/.cache/xci.env: line 90: 50340: command not found
/home/devuser/releng-xci/.cache/xci.env: line 92: -e: command not found
/home/devuser/releng-xci/.cache/xci.env: line 95: fg: no job control
/home/devuser/releng-xci/.cache/xci.env: line 101: fg: no job control
As such, we need to ensure that variables are properly quoted when
exported.
Change-Id: Ie98d8c921bae849d6bffd3858f67708f417b1d0f
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
|
|
The jobs are unmaintained and broken for months. Upstream already has
jobs for all 3 distributions so there is little value to test a similar
scenario on the OPNFV side. As such, lets disable them and save some
hardware resources for XCI.
Change-Id: I0ab7d51f44bdeb6f630671e8d05dd2baf67d756c
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
We only set default scenario if there is not one set already so we
need to check if the array is empty.
Change-Id: Ic5019fb68349adaf52d669aed43574bc5eaeddc3
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Change-Id: I9576417cd2c47fd10153d841dddb334cb47c666e
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
|
|
|
|
|
|
We should reload the XCI environment before running functest in order to
ensure that playbooks have the correct information from the XCI
deployment.
Change-Id: Ieadeca61d3ea3e504e21aa96945ad067f20d1489
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
This change enables Functest for kubernetes scenarios but any change
that impacts kubernetes scenarios will fail and block the change from
merging until the prepare-functest playbook and role are fixed in
releng-xci.
Change-Id: I046e3a0fb9a211cf9d629899017922288d466b0c
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Commands may exit with non-zero exit code but this could be normal (for
example grep does than when it doesn't match anything). As such, lets
not fail on everything but rather capture such failures and handle them
appropriately.
Change-Id: Icc16d0ccd09fd6f262600ae514161bfdf0c593f1
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
The bifrost jobs are using some scripts from the releng-xci repository
so we need to set the XCI_PATH variable appropriately.
Change-Id: Ibfd139d2401029b2ea577c4797b68f29362356ec
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Previously, the code was already setting the default scenario if we had
changes outside the xci/scenario directory. However, this is somewhat
problematic since scenarios may touch other files which do not necessary
affect other scenarios such as anything in xci/playbooks etc. As such,
lets only set a default scenario if no scenario was set up to that
point. However, since installers and scenarios need to match, we also
need to consider installer changes because we fallback to the default
scenario. Finally, since we are filtering results with 'uniq' we can
simply drop the bash regexps prior to adding the scenario to the array.
Change-Id: Ie961071b3e93d04a860bb6655513654208b1a9d4
Signed-off-by: Markos Chandras <mchandras@suse.de>
|