Jenkins job builder only does one pass of variable substitution. Replace
the cmd_erlang variable with its contents, so the jobs don't fail with:
[osmocom-gerrit-debian9] $ /bin/sh -xe /tmp/jenkins7189809471369692535.sh
+ {docker_run} {docker_img_erlang} /build/contrib/jenkins.sh
/tmp/jenkins7189809471369692535.sh: 2: /tmp/jenkins7189809471369692535.sh: {docker_run}: not found
Fixes: aadc60f403 ("jobs/gerrit-verifications.yml: add erlang repos")
Related: OS#4345
Change-Id: I6e121c79ecf872a7058de3a6dc0a940fa8091f68
Run contrib/jenkins.sh in erlang repositories, that were recently
updated (osmo_gsup, osmo_dia2gsup, osmo_ss7).
Depends: docker-playground Ia3eaec6090c9652549b2850de74ee21730374bbd
Related: OS#4345
Change-Id: I05d152de6b7a04dee935d79b9987c511351eca95
Instead of building "osmocom:deb9_amd64" from this repository, build
"$USER/debian-stretch-jenkins" from docker-playground.git (same
Dockerfile). Adjust all jobs to use the new image name.
Add a new "update-osmo-ci-on-slaves-dp" jenkins job, which triggers
the existing "update-osmo-ci-on-slaves" job whenever
docker-playground.git changes.
Replace docker/rebuild_osmocom_jenkins_image.sh with
scripts/osmo-ci-docker-rebuild.sh, so we can get rid of the docker dir.
I thought about dropping the script completely, and directly writing the
two lines into contrib/jenkins.sh. But I kept the extra script for
convenience, when testing locally.
Related: OS#4345
Depends: docker-playground I125ae8a6bcabbd1f485028c79b0abacda0622c3a
Change-Id: I30a61aebcadef5536e74edd35e1c75ef77a2da9f
Place a link to the wiki page, that now documents all available
environment variables.
Related: OS#4243
Change-Id: I770360b671c64369cf6cdb4fb4222921cc6af58e
Don't use the old "osmocom:amd64" image. Remove the entire cmd line, so
it uses the default docker command with the default docker image
"osmocom:deb9_amd64". Also remove concurrent: true, because that default
nowadays too.
Related: OS#3203
Change-Id: I93db12d3652f26db059e0ede52d2b2fdd3e89d34
Now that almost everything runs in docker, we don't need to worry about
problems with running VTY/CTRL tests in parallel for these projects.
Related: OS#3726
Change-Id: I5fdf759faf47ec7e4f9cb2657bd004585a667add
Instead of building some projects in docker, and some directly on the
build slave, build almost all of them in docker with few exceptions, and
make building in docker the default.
This has several advantages:
* build environment of master builds and gerrit verifications is the
same, therefore the same build won't pass in gerrit and then fail in
master
* all build environments are volatile
* no more conflicts while running VTY/CTRL tests
* fix osmo-python-tests gerrit review versions being installed to
/usr/local of build slave (see previous commit message)
Projects that still do not get built in docker:
* openggsn: depends on libgtpnl, but does not build it in jenkins.sh
* osmo-bts: l1headers, various SDKs not installed in the docker image
* osmo-trx: LimeSuite is not in the docker image (and just installing the
debian package does not work)
* osmo-ttcn3-hacks: TTCN-3 not installed in docker image
* pysim: needs to access real sim cards attached to the host
* sysmo-usim-tool: needs to access real sim cards attached to the host
Related: OS#3726, OS#3598
Change-Id: If4b7d2f9ffbc7e92699732d97a8f5829a88c5b35
Otherwise the master job managing the axis can be run on any node,
something which is not desirable. For instance, right now job
master-osmo-ggsn is run on osmo-gsm-tester-rnd which is currently unable
to fetch some git repositories.
Let's avoid running master tests on unintended nodes.
Change-Id: I62abb3c490588adc4c0e13c761de44d442f8cefe
Debian jessie (8) is EOL already, so switch to stretch (9). This is
required to build all Osmocom projects in docker, because at least
libosmocore fails to build with the old image (tests won't build with
GCC-4.9.x from jessie unless -Werror is disabled).
Do not build the jessie image anymore, and update the shared dockerfile.
Remove everything jessie specific, and make it work with stretch by
installing asciidoc-dblatex. Otherwise we get this error when building
manuals:
a2x: ERROR: missing configuration file: /etc/asciidoc/dblatex/asciidoc-dblatex.xsl
Add '--cap-add SYS_PTRACE' argument to the docker run command, because
otherwise the address sanitizer isn't working anymore with stretch
(which makes the OsmoPCU tests fail).
I've tried to split this patch up into multiple ones, but since there is
no asciidoc-dblatex package in jessie, we must add the dependency and
switch the image at the same time.
Related: OS#3726
Change-Id: I61fe93863d5367e3b9c01ab4726bb95bd16d891d
Move the common "docker run" invocation and image name into one place,
so it is not duplicated over and over. This prepares for changing the
image and building all Osmocom projects in docker (follow up patches).
Related: OS#3726
Change-Id: If529e6a93bb2227230981ee22f2c75bd0f1f4518
Re-enable concurrent builds for OsmoPCU, and run all builds in
docker. This should prevent previous problems with VTY/CTRL test
collisions, which lead to the concurrent builds being disabled
in I23d5b75825a667e4f043d16a12b841cd8f01af5e.
Related: OS#4204
Change-Id: I4f90a601e41dca18a6fcbab56bdf5663167d2353
The asciidoc warnings check fails the build if asciidoc prints a warning
while building the documentation. We had this enabled by default, but
since OBS uses various asciidoc versions, we could get unexpected build
failures (different asciidoc versions print different warnings).
Now that it is disabled by default, explicitly enable it for jenkins
again.
Related: OS#4140
Depends: Iac993a0d1b17205397a1f1ef1a7bd3f9df739e36 (osmo-gsm-manuals)
Change-Id: I8775d45b4dc1a5a46e2ba042fc7ee7d9f10f18cc
Make sure that we don't test yesterday's packages. Adjust timers to run
the OBS job around 1 o'clock, and the debian-install test around 4
o'clock. Use the H (hash) symbol for minute to spread the load.
Change-Id: I503058018172d50e9585d4804dfec1b4ece7644a
Allow changing the branch in the jenkins UI when starting the build.
We already have this in other jobs, and it is very useful to test
changes.
Change-Id: Ice4662551625b3c96ca105629a934a0a081db2e4
Replace "Generated by job-builder" with a proper description. The job
builder automatically adds "<!-- Managed by Jenkins Job Builder -->" to
the description (which is visible when editing the job), so we don't
need to mention it in the description at all.
Related: OS#3369
Change-Id: I17d2e9e0ac7fe77fb0306724a1f1e15641d76854
I was quite surprised, that the job used "osmith/pdfs-in-doc-subpackage"
instead of "master" today [1]. Fix this randomness by adding a branch
parameter that defaults to "master".
Interestingly, we have to set the default branch to
"refs/remotes/origin/master". When using "*/master" (like we do in other
jobs), Jenkins would still chose my branch over "origin/master", even
though mine does not have "master" in the name. I have also tried
"chosing-strategy: 'default'", in case it was using the gerrit strategy
for some reason (like in [2]), and setting 'refspec:', but both did not
have any effect.
[1] https://jenkins.osmocom.org/jenkins/job/Osmocom_OBS_nightly/98/
[2] https://issues.jenkins-ci.org/browse/JENKINS-26981
Change-Id: I6d1f0e70d5c82ffd551d00ffdf2712cbfaec16cf
Add the links that the obsolete Osmocom_nightly_packages job has, before
we remove it. Also link to the binary packages wiki page.
Change-Id: Idbc7ccd0156d9c3eb6d30059384686849a36f49f
Adjust README.adoc to mention "cmd: |" instead of the workaround and
replace this:
cmd: >
# keep first line with less indent to preserve newlines
docker run ...
With that:
cmd: |
docker run ...
Change-Id: I65a06acb76e5d693aa187f9ac38970b73e7fead6
This requires Change-ID I800ebe0f33cf7c86eea65d2e186c6af12fee6db7 from
osmo-ttcn3-hacks.
Fixes: OS#3179
Change-Id: Idee5a986e4cda8db601982ac3f2662bc20fe3f5f
We had some load issues with many things running at once on jenkins.
This will at least make it easier from having colliding ttcn3 jobs.
I also changed ttcn3-sip-test to run at 7:30 instead of 7:00 where
ttcn3-bts-test runs.
Change-Id: I65f346a3fae2811c8314c26d4a5b3a528b8101f9
With mobile, virtphy, osmo-bts-virtual we can run e2e tests without
the need of any hardware. Create a dedicated job for it. This avoids
having to share resources (e.g. lock state/ip allocations) with other
builds.
Change-Id: I19993b287cc948dbdfba9d375f80a7904f96ff4a
The repository currently doesn't yet have a contrib/jenkins.sh
script, so all verifications will fail. This is intentional,
as it will force the addition of a related script in the first merged
commit.
Change-Id: I713d345240dec319ee07d99e9381fc1af6a2d86d
Jenkins job builder does not support string replacing in variables,
so the only way to make each ttcn3-foo-test-latest job get blocked by
their ttcn3-foo-test counterpart was specifying them each time.
Related: OS#3767
Change-Id: I124286d4b996e79a7b77603c7b59473b016c3bd5