Age | Commit message (Collapse) | Author | Files | Lines |
|
The clean VM normally stays around after the job is completed or killed
due to timeout. Unless a job for the same distro is scheduled on the
same node, that VM will state around for a while consuming valuable
system resources. As such, we should remove it after the job finishes
to free up resources for the new jobs.
Change-Id: I88b5e32797969d9378588199a7ddb013bbe740c9
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
We do not need all the components installed when deploying OpenStack
so we can speed up the jobs by removing some of them.
Change-Id: I19120915d0ef7c59b8b470c5fb87e9c0aa3e3673
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
Change-Id: I883e6e29cf102d1069e3760ab95dd24d76d0a222
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Change-Id: Ie5fae322f98ec6c9c2236486642c0d71c6a04aae
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Change-Id: Ifcea3d364b73d0d19d46aaf2bd2210ec6440f383
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Change-Id: I9366639570315f879a662ca9e65a1894cd905a92
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Change-Id: I231488cd9de58a9051bd289c3d584841e7335417
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Change-Id: I5e8b738d4293040b2792a926cd13708180590bbb
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Change-Id: I40195f08012774c22178a27f146c402b33ecd8fc
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
The envInject complains due to nonexistent scenario.properties so we
set the scenario to os-nosdn-nofeature to suppress the failures since
there is nothing to fail.
Change-Id: Ic259ec53e77cf8afe62638a1b447c8e58863b419
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Change-Id: I261bfafefcd1fb5dcfead20697759e46d3556b46
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Change-Id: Ied1014546f6bd4d80896f29f86374a2865f95dca
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
|
|
Change-Id: I6387ca1f7eaf107f735eab82917d77b2f3a3d90a
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Change-Id: Iccd7ef674c9b80daba511f45cc48f403ca0f3e0b
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Change-Id: Ia2caeef24a199580a9eeb2ba6214fa2008f65c1c
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
When a patch comes in for an external scenario, Gerrit Trigger injects
refs for that patch for the corresponding project. With the previous
configuration, job attempts to get that patch for releng-xci instead
of the actual project.
But switching to the right project causes Jenkins jobs to fail since
the releng-xci will not be checked out to WORKSPACE.
This change makes the clone for the repo/patch work and then it wipes
out the WORKSPACE and clones releng-xci there to get the actual script
parts to work.
Change-Id: Ide31c21a209dff953118f51e9b09de1d206a3085
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Change-Id: Ib58ff1a8b093af550f86d337327114df50da4ae7
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Change-Id: Ic706c120ac879741d5562427a6c3e90162223762
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
|
|
In commit I77d5d66c0146f943d805612ca5b47e2eed3ffc41 the 'ts' utility
was added in order to print timestamps for deployment. However, this
breaks the exit code since 'pipefail' was not set so every job was
returning 'success'.
Change-Id: I2625a655a7334b1e43b690618821e69377075ed1
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
We can use the 'ts' utility to print timestamps on every executed
command so we can keep track on the duration of each task and the
job overall.
Change-Id: I77d5d66c0146f943d805612ca5b47e2eed3ffc41
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
|
|
|
|
The bifrost regexp was wrong leading to conflicts betwen the XCI
and bifrost jobs. Moreover, we need to add the same restrictions for
jobs/node on bifrost jobs similar to the XCI ones. This finally fixes
the 'throttle' property to use the 'xci-verify-virtual' category.
Change-Id: I6d3f8fdd82738c1d079b1a3fdef134753eb4281a
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
The bifrost jobs could run in parallel with the XCI ones so it may take
longer than 90 minutes to complete. Lets double the timeout value in
order to ensure that we don't fail due to a Jenkins timeout.
Change-Id: I53a442f434d839d891bf03e41ba1a6a4459fd878
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
We shouldn't allow the merge jobs to run in parallel with the verify
ones when both are testing the same distro since one may wipe the
'clean VM' of the other one. Moreover, we should restrict the number of
jobs allocated on nodes in order to avoid overloading them.
Change-Id: I9c0b9b5061e41b21c01789fc4676e72e0e082f17
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
This change extracts the scenario that is impacted by the patch
which triggered the verify jobs and saves the scenario name into
java properties file. This file will then be used by envInject
plugin to inject the scenario name into the build environment.
If the triggering change impacts multiple scenarios, the build
will result in failure and the developer will be informed about
a possible way to rectify the issue.
Support for generic scenarios will be introduced in subsequent changes.
Change-Id: I6406299c1352b6e9b8a9297c5441adf03445be33
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Change-Id: If5c5b828312dc588515589b8075b9754537b8f86
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Make sure that failures during bifrost deployments are being
reported properly on Jenkins.
Change-Id: I333d27147caf7598665157c703907572e48041c0
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
CentOS jobs started failing which requires troubleshooting.
In order not to block things, the voting is disabled for CentOS jobs.
Change-Id: Idc14b00cf09f0ee75bac17e993dc9f7724a1ad88
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
CentOS is now supported by os-nosdn-nofeature scenario and it is
time to enable voting for it. For scenarios that do not support
CentOS yet, the jobs will skip running real stuff by exit 0 in
order not to fail the patch.
Change-Id: I86c9e1bb4d66ea84ccfc864186a7ca7c696f8663
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
This is yet another messy work to get functest run for xci
as part of merge jobs.
The script is removed from xci-verify jobs in order not to
block patches by -1ing them.
Change-Id: I57254bcfbada7a51da89aba0e7a88f907eb9a525
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Change-Id: I22d2327558ee1d19d04775da6abf1b3eed5b357a
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Change-Id: I48cdbfbd5b7746e75565d318616e9c580b5ba156
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Forgotten to fix phase jobs.
Change-Id: I50c525089d6d3b4f28650ad78b8810817aaf150e
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
Change-Id: Ie1c2f76b6043133c3821272040b1a2a38de29035
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
The cloning of the correct version of the repo which the scenario
resides in will be handled by get-opnfv-scenario-requirements.yml
for all the scenarios.
The change proposed for merge jobs will be reflected to verify jobs
as well once it is verified with this.
Change-Id: Icaa99a8b9e981c839969ce556e5af6a9ef6d9da8
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
This change enables the deployment upon merge of a change.
Followup patches will incorporate functest smoke and promotion
which the promotion will most likely be the first one as functest
parts are a bit tricky due to failures while running testing
in VMs.
Apart from the enablement of the deployments, job concurrency is
disabled as well since we should only test one change at a time
and attempt promotion for any scenario in order to ensure we keep
the promotion consistent.
Change-Id: Ib4b5754b6b88f42eab83c656e03773c83068016f
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
The scripts used by verify jobs to create VMs, start deployments,
run functest, and do cleanup for CI can be reused by other jobs as
well so they are moved out of the yaml file.
Apart from that, if we decide to change the way we run CI (not
using VMs for performance reasons), we don't have to make big changes
to yaml either.
Change-Id: I2df1f2d60f999bcd7d53a5cd879112a5b2b0ca05
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
This change proposes structure for post-merge jobs for the
scenarios that are onboarded to XCI.
The idea with these jobs is to verify the sanity of impacted
scenario to find out if the version that is tested should be
promoted to the next level in XCI; daily on baremetal.
If it succeeds, the metadata regarding the scenario needs to
be stored somewhere (artifact repo, etc.) so daily jobs can
poll that location to see if there is any scenario that is
worth testing on baremetal.
If it fails, the scenario could be demoted instead which still
requires further discussion and thinking. We might perhaps leave
the metadata of the version of the scenario with previous sha
as the candidate for testing instead of demoting it since we
need to have working version of the scenarios at all times.
Change-Id: I85d7176b16d0d6da45cd2462b0cdb990485b6b0d
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
This change takes xci-verify-jobs as base and adjusts osa-periodic
jobs to ensure they do what they are supposed to.
The structure of periodic jobs is pretty similar to xci-verify-jobs and the
differences comparing to them are
- OPENSTACK_OSA_VERSION is set to master.
- SCM type changed to git-scm from gerrit
- Voting is removed.
- Triggering will be done using pollscm to ensure we only run jobs when something
changes upstream. Triggers are not enabled yet until the basics are verified
working.
This change requires https://gerrit.opnfv.org/gerrit/#/c/49275/ in order for
periodic jobs to work as expected.
Further changes are needed to finalize the structure.
Change-Id: I8f6519ccce93e6b8b65509638294419459fd36e8
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
|
|
Change-Id: Ief47cd60d45f2cc35c74a9e8430d6602210bbba1
Signed-off-by: Fatih Degirmenci <fdegir@gmail.com>
|
|
We should set 'disable-strict-forbidden-file-verification' to 'false'
to really prevent any changes to the forbidden paths to trigger a CI
job. When this option is set to 'True', then a patchset that modifies
both allowed and forbidden paths will trigger a job. However, in order
to be on the same side in regards to the start-new-vm.sh script, we
really shouldn't trigger the job. Moreover, we drop the rest of the
forbidden paths since we can use the 'skip-verify' topic name to
skip the deployment.
Change-Id: I2a89e51861c7fb8541fe774b128ea4049e91df43
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
The scripts in xci/scripts/vm are used to manage the VM on the CI so
we shouldn't blindly test a patch that modifies them. It's better to
use manual inspection since that script rarely changes anyway.
Change-Id: I48d81f984a1a8d1500be2cc74ea79420962b9a15
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
|
|
Change-Id: I97d8fac47ec25414045a7761b651019f9f89224f
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
We need to use the XCI specific SSH configuration file so rsync can
tranfer the working directory to the virtual machine. This also
drop the ironic command like we do in the XCI jobs.
Change-Id: I10ce38db6777d5dbfbe62961c88b3902de2160e2
Signed-off-by: Markos Chandras <mchandras@suse.de>
|
|
|