There is a new Rackspace region coming online that may be fast enough to
use in the gate. Add a devstack job which is configured to run on nodes
in the new region in the silent queue, so that we can check out
runtimes without really bothering people.
Change-Id: I9b8a92ac92562581dab3818196e0006e2d69e08d
* modules/jenkins/files/slave_scripts/run-tox.sh: Feeding subunit v1
through a subunit v2 parser then running the leftovers through a subunit
v1 parser is a really slow way to support subunit v1 and subunit v2.
Instead convert subunit v1 to subunit v2 to speed up parsing. Local
testing indicates this should shave off minutes of test run times.
Closes-Bug: #1221094
Change-Id: Ie17708ac217e7468bb1c92cfda7aa50994178817
neutron 26 jobs are actually regularly hitting the 40 minute
timeout on unit tests, as their actual unit tests take ~32 minutes
and there is current a performance issue on subunit report
processing that makes it take ~9 minutes (on 27, 26 might be slower
still).
This means that completely passing unit test runs are reseting the
gate due to a timeout limit.
60 minutes should be sufficient to get us past this issue, and let
us figure out the subunit performance issue. However given neutron
unit test growth, 40 minutes is probably too low anyway.
Change-Id: If4bb799d9b6fe8815280aca9f28fee8a65a477c6
* doc/source/gerrit.rst: When renaming projects in the changes
table, set created_on to itself so that it won't get updated to the
current time. This column is a timestamp datatype in MySQL and has
the "on update" property set, probably for no good reason but it's
how Gerrit's installer sets the schema.
Change-Id: I5b99d2869e55ab0f8b036fd08f248f406aff41bb
* doc/source/gerrit.rst: Add sections to the document about
resetting Gerrit SSH usernames and cleaning up duplicate preferred
E-mail addresses. Also a section demonstrating the dreaded account
merge process for posterity, even though we do not offer to perform
this for our users.
Change-Id: I43883ef9558abbdf5051aa617def36d1fc158100
* modules/openstack_project/manifests/logstash.pp: Rotate jenkins log
client logs with logrotate. Keep one week of logs and rotate daily.
Change-Id: I9b6a320afda76cc7285c959196a24853a5a9561e
This project is intented to provide community and software development
metrics of the OpenStack community from several perspectives.
Change-Id: Ie3aa1adbf7093b988ea58af68da4332fcce70536
In preparation to split the run_mirror.py script into
a different project from jeepyb, we need to create the
new home for the run_mirror.py script.
Change-Id: I1dd467afe8ef6a153d8a5f0fc95707c864861f9c
This is a minor step towards PEP-8 as well, but mainly makes it easier
to document what third party modules are needed.
Change-Id: Id901ef437fcdeb11c8dd555e8db7c13daa86c4b2
While credentials for clouds are uploaded by hand, the place they are
put is constant and has one correct setting for mode and owner, so
lets automate it.
Change-Id: I98ce4493c260f97c3ac1028d7065536178760d72
Puppetmaster bringup will often have the wrong hostname in a cloud
(e.g. a uuid or the shortname given to nova). But if it's wrong the
puppetmaster CA created by the puppetmaster module will be broken.
Additional nodes can have this handled by automation, but the
puppetmaster is special - this is part of the bootstrap process.
Change-Id: I8e0e2ea4a5d27fb248b88d6be57abb171d81f13f
Packages install it with mode 0755 which then fails to chmod in the
webservice, so lets set it up from within puppet itself while we are
privileged.
Fixes bug: 1218697
Change-Id: I3a005ab2495502047fc8b18e385b421a9afc35d4
Derived infrastructure deployments may well live as a set of rebased
patches floating on the upstream infrastructure. Simple 'pull -q' will
conflict if the reference branch is rebased.
Change-Id: I8127bb6305208879957c01e1c81b6329d74b27bf
