build_utils: import ubuntu-toolchain-r's key without keyserver

we are running into following error recently

gpg: requesting key BA9EF27F from hkp server keyserver.ubuntu.com
gpgkeys: key 1E9377A2BA9EF27F can't be retrieved
gpg: no valid OpenPGP data found.
gpg: Total number processed: 0
gpg: keyserver communications error: keyserver helper general error
gpg: keyserver communications error: unknown pubkey algorithm
gpg: keyserver receive failed: unknown pubkey algorithm

so use the local copy of the pubkey instead

Signed-off-by: Kefu Chai <kchai@redhat.com>

Merge pull request #1073 from ceph/evelu-ccache2

ceph-pull-requests: Simplifying run-make-check call

Reviewed-by: Sébastien Han <seb@redhat.com>
Reviewed-by: Kefu Chai <kchai@redhat.com>

ceph-pull-requests: Simplifying run-make-check call

Once https://github.com/ceph/ceph/pull/22847 will be merged, this commit
will remove all the custom hacks from this script as they moved directly
in ./run-make-check.sh

Merge pull request #1070 from ceph/return-0

ceph-ansible-pr-syntax-check: force return 0

ceph-ansible-pr-syntax-check: force return 0

If the test succeeds we return 0.

Signed-off-by: Sébastien Han <seb@redhat.com>

Merge pull request #1069 from ceph/do-not-return

ceph-ansible-pr-syntax-check: return 1 only on failure

ceph-ansible-pr-syntax-check: return 1 only on failure

This was a leftover from an initial attempt with if combinaison. We now
return 1 only if the test line fails.

Signed-off-by: Sébastien Han <seb@redhat.com>

Merge pull request #1068 from ceph/syntax-head-3

ceph-ansible-pr-syntax-check: remove merge commit from count

ceph-ansible-pr-syntax-check: remove merge commit from count

When counting the number of commit you must remove the merge commit so
we get the right number of commits.

Signed-off-by: Sébastien Han <seb@redhat.com>

Merge pull request #1067 from ceph/syntax-head-2

Syntax head 2

ceph-ansible-pr-syntax-check: use target branch to compare

We must use the target branch as a reference when we try to find the
number of commits pushed.
We use 'ghprbTargetBranch' env variable, which comes  from jenkins's
injectedEnvVars.

Signed-off-by: Sébastien Han <seb@redhat.com>

Revert "ceph-ansible-pr-syntax-check: use master instead of HEAD"

This reverts commit 7c971b82771ef6766206b7ef0716a8705fdbe649.
Because this is the wrong fix, if you push a PR on a stable branch, the
comparaison will happen against master, we don't want that.

Merge pull request #1066 from ceph/syntax-head

ceph-ansible-pr-syntax-check: use master instead of HEAD

ceph-ansible-pr-syntax-check: use master instead of HEAD

origin/HEAD does not exist so let's use origin/master instead. This
prevents the following error:

bad revision 'origin/HEAD'

Signed-off-by: Sébastien Han <seb@redhat.com>

Merge pull request #1065 from ceph/collect-ceph-volume-logs

scripts: collect all logs in /var/log/ceph on ansible failures

scripts: collect all logs in /var/log/ceph on ansible failures

On the jobs that use ceph-ansible to test ceph logs are collected from
the vms on failure. However, if the cluster name was not ceph the find
command would fail to collect any ceph-volume logs.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>

Merge pull request #1064 from ceph/wip-no-ask

ceph-pull-requests: Automatically rebuild w/o asking params

ceph-pull-requests: Automatically rebuild w/o asking params

This will carry over our custom sha1 variable when "Rebuild" is clicked.
Previously, it was possible to click "Rebuild," and leave the `sha1`
variable empty which caused every branch/PR to get built.

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #1063 from ceph/signof-pipeline

ceph-ansible-pr-syntax-check: add sign-off check

ceph-ansible-pr-syntax-check: add sign-off check

We now fail if the commits don't contain a Signed-off-by certificate.

Signed-off-by: Sébastien Han <seb@redhat.com>

Merge pull request #1061 from joscollin/wip-cobertura-fail-fix

ceph-build/cobertura: set *auto-update to false

ceph-build/cobertura: set *auto-update to false

Signed-off-by: Jos Collin <jcollin@redhat.com>

Merge pull request #1060 from ceph/wip-fix-pipeline

ceph-ansible-pipeline: Fix stable-3.0 phase conditional steps

ceph-ansible-pipeline: Fix stable-3.0 phase conditional steps

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #1058 from ceph/evelu-cache

ceph-pull-requests: Improving ccache efficiency

Reviewed-by: Alfredo Deza <adeza@redhat.com>
Reviewed-by: Gregory Farnum <gfarnum@redhat.com>
Reviewed-by: Kefu Chai <kchai@redhat.com>

ceph-pull-requests: Improving ccache efficiency

As of today, ccache is installed by run-make-check but the statistics
reports almost no cache it (<1%) and even worse, this is slowing down
the whole process. The needed time to setup the cache isn't compensated
by the reuse of the cache as every build is rebuilding the cache. This
is leading to a 10% loss of the compilation time.

So the question, is why does two consecutives builds doesn't reuse the
cache. Its usually a matter of variables introduces on each builds like
a sha1 or a __DATE__ or __TIME__ in a header.

As per
https://wiki.debian.org/ReproducibleBuilds/TimestampsProposal, several
tools from the tool chain (like compilers or make tools) are able to use
theses variables to enforce a particular timestamp at build time.

This commit also add a "-D ENABLE_GIT_VERSION=OFF" to avoid having the
commit id in the ceph_ver.h header. Not having the git versioning inside
the CI job is not a by deal as the binaries will be destroyed after the
build procedure.

These changes avoid introducing small variations in the source code
which breaks the build cache (ccache). The expected result is to get a
80 or 90% cache hit between two very close builds.

Signed-off-by: Erwan Velu <erwan@redhat.com>

Merge pull request #1042 from a2batic/test1

mgr/dashboard: Adds unit test coverage report

Reviewed-by: Alfredo Deza <adeza@redhat.com>
Reviewed-by: Jos Collin <jcollin@redhat.com>

mgr/dashboard: Adds unit test coverage report
Enable plugin 'Cobertura' to generate code coverage report, for every PR for ceph-dashboard frontend unit test.
Cobertura jenkins plugin: https://wiki.jenkins.io/display/JENKINS/Cobertura+Plugin

Fixes: http://tracker.ceph.com/issues/24377
Signed-off-by: Kanika Murarka <kmurarka@redhat.com>

Merge pull request #1057 from ceph/wip-arm-name

ceph-pull-requests-arm64: Normalize name to appease JJB

ceph-pull-requests-arm64: Normalize name to appease JJB

Not sure why '!!python/unicode' was there originally but the current
version of JJB doesn't like it.

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #1055 from ceph/wip-ansible-nostable

ceph-ansible-pipeline: Skip *lvm* scenarios if merging into stable-3.0

ceph-ansible-pipeline: Skip *lvm* scenarios if merging into stable-3.0

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #1054 from ceph/wip-ansible-condition

ceph-ansible-pipeline: Only run next phase on SUCCESS

ceph-ansible-pipeline: Only run next phase on SUCCESS

I think I misunderstood the jjb syntax before.

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #1053 from ceph/group-vars-fix

ceph-ansible-pr-syntax-check: fix grep error

ceph-ansible-pr-syntax-check: fix grep error

If grep fails the script exits because of the set -e so let's use ||
true to workaround that issue.

Signed-off-by: Sébastien Han <seb@redhat.com>

Merge pull request #1052 from ceph/group-vars

ceph-ansible-pr-syntax-check: add group_vars check

ceph-ansible-pr-syntax-check: add group_vars check

This commit does 2 things:

* split the structure of the script to use functions
* add a function to test if group_vars weren't generated

Signed-off-by: Sébastien Han <seb@redhat.com>

Merge pull request #1050 from ceph/guits_update_ceph-ansible_nightlies

ceph-ansible: update nightlies definition

ceph-ansible: update nightlies definition

- remove testing of stable-2.2 against ceph jewel
- add testing of stable-3.1 against luminous
- update testing of stable-3.0 against luminous:
  - remove lvm scenarios

Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>

Merge pull request #1051 from ceph/guits-change_osd_bs_project_ceph-ansible

ceph-ansible: move bluestore_osds_container in ceph-ansible-prs-auto

ceph-ansible: move bluestore_osds_container in ceph-ansible-prs-auto

we are seeing this scenario failing because of the race condition,
moving it to a project which requires smithi nodes should help.

Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>

Merge pull request #1049 from ceph/rm-ceph-disk-master

ceph-volume-nightly: remove master from simple, ceph-disk no longer available

ceph-volume-nightly: remove master from simple, ceph-disk no longer available

Signed-off-by: Alfredo Deza <adeza@redhat.com>

Merge pull request #1048 from ceph/ceph-deploy-distros

ceph-deploy-build: remove centos6

ceph-deploy-build: remove centos6

Signed-off-by: Alfredo Deza <adeza@redhat.com>

Merge pull request #1047 from ceph/wip-allow-all

ceph-ansible-*: permit-all to prevent github comment flood

ceph-ansible-*: permit-all to prevent github comment flood

See https://github.com/jenkinsci/ghprb-plugin/issues/608

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #1046 from ceph/guits-revert_80f02250028c

Revert "ceph-ansible-prs: run 'auto' scenarios on smithi or ovh"

Revert "ceph-ansible-prs: run 'auto' scenarios on smithi or ovh"

This reverts commit 80f02250028ce5daae6eed3717787732d06302d1.

Merge pull request #1045 from ceph/wip-ansible-pipeline3

ceph-ansible-pipeline: Implement additional phases

ceph-ansible-pipeline: Implement additional phases

1) Docs and yaml linter (always)
2) Check each infrastructure playbook and run if modified
3) centos7 and docker clusters
4) All other scenarios if basic cluster tests pass

Phases 3 and 4 will NOT be run if infrastructure-playbooks were
modified.

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #1044 from ceph/wip-ansible-pipeline

Implement ceph-ansible pipeline

Let ceph-ansible-pipeline trigger ceph-ansible-* jobs

Signed-off-by: David Galloway <dgallowa@redhat.com>

Add ceph-ansible master multijob

This job will trigger docs, syntax check, and if they pass, then the
actual ceph-ansible jobs will run.

The ceph-ansible-prs-auto project has to move to this job config in order for
the pipeline to start the appropriate jobs for PRs.

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #1043 from ceph/guits-run_ceph-ansible_prs_ovh-smithi

ceph-ansible-prs: run 'auto' scenarios on smithi or ovh

ceph-ansible-prs: run 'auto' scenarios on smithi or ovh

This is slowing down CI testing because we can't run more than 1 or 2
PRs at a time. Since I'm not sure if it's still worth to limit these
scenarios to be run only on smithi node, let's try to temporary remove
this limitation.

Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>

Merge pull request #1038 from ceph/wip-no-docs

ceph-ansible-prs: Don't run auto job if only docs were modified

ceph-ansible-prs: Don't run auto job if only docs were modified

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #1041 from tchaikov/wip-build_utils.sh-cleanup

build_utils.sh: do not try to check non-existent file's size

Reviewed-by: Alfredo Deza <adeza@redhat.com>

build_utils.sh: do not try to check non-existent file's size

it's distracting to read the red alert in jenkins' log like

du: cannot access '/srv/debian-base/xenial.tgz': No such file or
directory

Signed-off-by: Kefu Chai <kchai@redhat.com>

Merge pull request #1040 from ceph/ceph-ansible-prs-dev

build_utils: when running dev-* scenarios set release to nautilus

ceph-build-pull-requests: pin urllib3 to 1.22

We need to do this because jenkins-job-builder installs requests, and if
we don't pin here a version of urllib3 is installed that is incompatible
with the version of requests that jenkins-job-builder requires which is
urllib3<1.23,>=1.21.1

Signed-off-by: Andrew Schoen <aschoen@redhat.com>

build_utils: when running dev-* scenarios set release to nautilus

When running dev-* tests for ceph-ansible PRs we were setting
CEPH_STABLE_RELEASE to 'dev' which is not an actual release and our
functional tests break because of that.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>

Merge pull request #1039 from ceph/ceph-ansible-prs-dev

ceph-ansible-prs: adds the dev-ansible2.4-centos7_cluster scenario

ceph-ansible-prs: adds the dev-ansible2.4-centos7_cluster scenario

This will allow us to test dev installs with a large cluster and many
node types.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>

Merge pull request #1037 from ceph/guits-followup_1036

ceph-ansible: followup on #1036

ceph-ansible: followup on #1036

these tasks are producing some useless output showing error on debian
based distribution. Let's add a condition to avoid this.

Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>

Merge pull request #1036 from ceph/guits-debug_yum_failures

ceph-ansible: add some tasks for debugging purpose

ceph-ansible: add some tasks for debugging purpose

related to ceph/ceph-ansible/pull/2665 and ceph/ceph-ansible/pull/2652

We are seeing too many failures related to yum.
These additional tasks will help us to have more information when a yum
failure occurs.

Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>

Merge pull request #1035 from tchaikov/wip-aarch64-repo

build_utils.sh: add ubuntu-ports repo for arm64 build

build_utils.sh: use arch-neutral path names

Signed-off-by: Kefu Chai <kchai@redhat.com>

build_utils.sh: add ubuntu-ports repo for arm64 build

as it offers updated binutils which does not exist in the default repo.
and the updated binutils is depended by gcc-7. this change address
following failure like:

The following packages have unmet dependencies:
 gcc-7 : Depends: binutils (>= 2.26.1) but 2.26-8ubuntu2 is to be
installed

when installing g++-7.

Signed-off-by: Kefu Chai <kchai@redhat.com>

Merge pull request #1034 from ceph/fix_collect_logs

ceph-ansible: fix teardown issue

ceph-ansible: fix teardown issue

At the moment, the teardown isn't fully executed because of `collect-logs.yml`
playbook.

it is expected the following task can fail according to the context of the main
failure:

```
TASK [find ceph logs] **********************************************************
task path: /home/jenkins-build/build/workspace/ceph-ansible-scenario/collect-logs.yml:4
fatal: [ceph-nfs0]: FAILED! => {
    "changed": true,
    "cmd": [
        "find",
        "/var/log/ceph",
        "-name",
        "test*.log"
    ],
    "delta": "0:00:00.003363",
    "end": "2018-05-23 16:45:22.845782",
    "rc": 1,
    "start": "2018-05-23 16:45:22.842419"
}

STDERR:

find: ‘/var/log/ceph’: No such file or directory

MSG:

non-zero return code
```

therefore, `teardown_vagrant_tests()` doesn't return `0` and causes the
teardown to stop because of `set -e`

```
PLAY RECAP *********************************************************************
ceph-client0               : ok=1    changed=0    unreachable=0    failed=1
ceph-client1               : ok=1    changed=0    unreachable=0    failed=1
ceph-iscsi-gw0             : ok=1    changed=0    unreachable=0    failed=1
ceph-mds0                  : ok=1    changed=0    unreachable=0    failed=1
ceph-mgr0                  : ok=1    changed=0    unreachable=0    failed=1
ceph-mon0                  : ok=3    changed=2    unreachable=0    failed=0
ceph-mon1                  : ok=3    changed=2    unreachable=0    failed=0
ceph-mon2                  : ok=3    changed=2    unreachable=0    failed=0
ceph-nfs0                  : ok=1    changed=0    unreachable=0    failed=1
ceph-osd0                  : ok=1    changed=0    unreachable=0    failed=1
ceph-rbd-mirror0           : ok=1    changed=0    unreachable=0    failed=1
ceph-rgw0                  : ok=1    changed=0    unreachable=0    failed=1

Build step 'Execute Scripts' marked build as failure
Archiving artifacts
Finished: FAILURE
```

adding a `failed_when: false` on this task should avoid this bug.

Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>

Merge pull request #1033 from ceph/remove_fetch_dir_teardown

ceph-ansible: some teardown fixes

ceph-ansible: vagrant clean old cache

Somehow, it's possible sometimes there is some old cache left after a
failure. `vagrant global-status --prune` ensures old caches that were left are
removed.

Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>

ceph-ansible: remove fetch directory in teardown

Remove the fetch directory in teardown for ceph-ansible to avoid
keeping leftover from previous jobs.

Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>

Merge pull request #1032 from ceph/cv-disable-bionic

ceph-volume-ansible-prs: remove bionic

ceph-volume-ansible-prs: remove bionic

ceph-volume doesn't have a 'bionic' factor in tox, so there aren't any
functional scenarios to support this. At the moment, there aren't any
ceph packages to test against either. Disabling this for now.

Signed-off-by: Alfredo Deza <adeza@redhat.com>

Merge pull request #1031 from ceph/wip-pip-nobinary

build_utils.sh: Don't update pip after installing it

build_utils.sh: Don't update pip after installing

Attempting to update pip was resulting in job failures because the wheel
wasn't present in the local pip cache.

Since we pinned pip to 10.0.0 in
https://github.com/ceph/ceph-build/pull/1007, we don't need to update
pip immediately after.

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #1030 from joscollin/wip-fix-signedoff-guide

ceph-pr-commits/build: Fix Sign your work url

ceph-pr-commits/build: Fix Sign your work url

Signed-off-by: Jos Collin <jcollin@redhat.com>

Merge pull request #1029 from ceph/wip-rtslib-fb-pyudev

rtslib-fb: reduce pyudev version requirements

rtslib-fb: reduce pyudev version requirements

Signed-off-by: Jason Dillaman <dillaman@redhat.com>

Merge pull request #1027 from ceph/cv-disable-bionic

ceph-volume-nightly: remove bionic

ceph-volume-nightly: remove bionic

ceph-volume doesn't have a 'bionic' factor in tox, so there aren't any
functional scenarios to support this. At the moment, there aren't any
ceph packages to test against either. Disabling this for now.

Signed-off-by: Alfredo Deza <adeza@redhat.com>

Merge pull request #1028 from tchaikov/bionic-gcc-7

build_utils.sh: help bionic to point its gcc back to 7

build_utils.sh: help bionic to point its gcc back to 7

Signed-off-by: Kefu Chai <kchai@redhat.com>

Merge pull request #1026 from tchaikov/bionic-gcc-7

build_utils.sh: help bionic to point its gcc back to 7.3

build_utils.sh: help bionic to point its gcc back to 7.3

Signed-off-by: Kefu Chai <kchai@redhat.com>

Merge pull request #1025 from ceph/wip-bionic-builddebs

setup_deb: Forgot bionic

setup_deb: Forgot bionic

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #1024 from ceph/wip-bionic-build

build_utils.sh: Add Bionic

build_utils.sh: Add Bionic

Signed-off-by: David Galloway <dgallowa@redhat.com>

Merge pull request #1023 from ceph/branch-regex-fix

ceph-build: use BRANCH instead of non-existing GIT_BRANCH

ceph-build: use BRANCH instead of non-existing GIT_BRANCH

Signed-off-by: Alfredo Deza <adeza@redhat.com>

Merge pull request #1022 from ceph/wip-bionic

Add support for Ubuntu Bionic globally

Add support for Ubuntu Bionic globally

Signed-off-by: David Galloway <dgallowa@redhat.com>