The logic in the Gemfile was relying on Zuulv2 variables to find out
whether the spec helper gem was already available on disk, and since
Zuulv3 has changed things it was failing to find it and downloading the
master version instead. This patch ensures the Gemfile looks for the gem
in the right place when running in CI.
Change-Id: I6dbe35f271daa41012045c7425644571f0bbe95b
There are some different changes in standalone.xml between
Zanata version 3 and 4. A version flag is implemented to
rollout the right version from template dir. For backwards
compatibility Zanata version 3 and 4 are supported yet.
Wildfly 9 stuff is removed because we have still Wildfly 10
running. That simplifies the manifest. Switch download URLs
to github, so there is also a standard to fetch files.
Change-Id: I5ada09cad7a19296802b22c2c3488ca547f2f14e
hint: http://docs.zanata.org/en/release/release-notes/#400
Instead of keeping a local copy of spec_helper_acceptance.rb and
requiring updates to all modules for any change, we can move it into the
common helper gem and require it from there. This will make it easier to
create and review changes that affect all puppet modules. Also change
the Gemfile to look for the gem in the local workspace if running in a
zuul environment.
Change-Id: I5c79b9661db5bec8c2de1763dc96568c9b3bba54
Add a xenial nodeset and update the spec helper to install puppet 3 from
the Ubuntu repos instead of from puppetlabs.
Change-Id: I577b70ff634478d28a54f708649269593da8bf73
Wildfly by default rotates logs daily for us. It however does not delete
logs after so many days. This leads disk filling up and unhappy servers.
We had similar issues with elasticsearch as well. Rather than sort out
how to configure logging to delete old files (if even possible) we just
added a daily cron entry that uses find to delete files more than 2
weeks old. Run a similar find via cron against the wildfly logs.
Change-Id: Ia6f262afd723aca0a4d6d981326880878436d518
This installs newer versions of zanata by unpacking published zip files
over the top of the wildfly install instead of manually injecting
specific war files.
Do this because the zip file includes the other dependencies that we
need like hibernate and the mysql connector.
Change-Id: I7632bbb30c8e3780836df65a47d35ca42b0b3f3c
A lot of this is equivalent xml with updated whitespace to reduce the
diff. This made it easier to see the actual differences between
upstream's config and ours.
With the whitespace changes out of the way I was able to update a few
things that were different between the two. The use db connection cache
manager value went from false to true, there was some added websocket
config, a missing weld section, and added back in our config for
specific openid source.
Change-Id: I11dd81a97662e8f24fdc364b353c7375407cfa30
With the upgrade to the wildfly puppet module we should no longer need
this hack to get systemd sysv compat working. Remove it as it is now
dead code and conflicts with the wildfly update.
Change-Id: I39453506821ff7073290a4e5696eda8575b6cb06
Wildfly includes these libs so we don't need to install them separately.
Check what version of wildfly is being installed and only add the libs
if necessary.
Prepare to have a different config file for 10
Change-Id: Idc09ef9d6d7200d86daad586719c2d4f6e143f1c
The different versions of ubuntu packaging use different driver names
for the mysql jdbc connector. Ensure we use the correct one on trusty
and on xenial.
Change-Id: I2cb154925bd67decd21fd440c13d3aafc4d8d982
We need to run daemon-reload and that only works if the path to
systemctl is correct. Change /usr/bin/systemctl to /bin/systemctl.
Change-Id: Ib3fb1b3dfa895b6b56adb19c22adc8b148c8c540
Xenial uses systemd which does have a sysv init script compatbility
layer but you must instruct systemd to reload its config after
installing any init scripts for this to work. Hack this in by adding an
exec for `systemctl daemon-reload` between writing the file and starting
the wildlfy service. Ideally the wildfly module would do this for us but
this should work for now (and eventually will become redundant).
Change-Id: If5c09c0be7714ce7d03ebf364823164a34c29ae4
This will install as depedency the right
Java package:
openjdk-6-jre-headless on Ubuntu 12.04
openjdk-7-jre-headless on Ubuntu 14.04
openjdk-8-jre-headless on Ubuntu 16.04
Java home is already set to /usr/lib/jvm/default-java
Change-Id: I834a3f3eb2f7d509be05aae9011748687f786745
::mysql::server has dropped use of config_hash to tune aspects of the
deployed mysql service. Rewrite the class included inside zanata::mysql
to use overrides.
Change-Id: I6dae4744812dc2469afc947e272cc9aa14e70788
The java_xmx value in Puppet for the -Xmx JAVA_OPTS variable was
increased 10 days ago following a timeout on the dev server with an
upload. Ran into it again during a large branching of Horizon
translations, so now increasing it to 4G following analysis of
server usage to confirm it can take the load.
Change-Id: Iee63c91456ad8f17b0a32732f6610416b6c5b987
Zanata can sometimes run into a "GC overhead limit exceeded" error
when running with the default 512m of RAM for the -Xmx JAVA_OPTS
variable.
Fortunately the Wildfly module we use has this as a variable for
it, using that (see manifests/install.pp for in the biemond-wildfly
module and associated standalone template for details).
Change-Id: I5d1f03c9d5cd57bb6bbf6650292173a3325cec4d
Wildfly modules and the Zanata ROOT.war file installations were written with
execs for downloads and extraction. Switch to using the archive module.
Note: Use of md5sum for the war file is good practice and it's easy to find on
sourceforge when we update the module, but also has the added benefit of making
the archive module update our ROOT.war file if the md5sum changes, as it would
when we update the download URL for a new version. Just changing the URL won't
trigger an update to this file.
We should consider and test adding md5sum check for the Wildfly modules later,
but they aren't required right now.
Change-Id: I4223d37f5dcff5c53bd7166291eb2f3915cd9546
Wildfly has released a new major version, which is verified
working by Zanata upstream, we should switch to it.
Change-Id: I612add9df4c300b9c1108296608b999f42fc728e
Use same target directory for zuul-cloner and
the regular git command.
Change-Id: I35f5397c943f4e1024182bf6c75fd34fa11f8edd
Co-Authored-By: Fabien Boucher <fabien.boucher@enovance.com>
Currently the second requirement for the ROOT.war deployment is an
Exec requirement that comes deep from within biemond-wildfly. This
is a layering violation and makes it impossible to upgrade the module.
Require the install class itself, rather than a part of its internals.
Change-Id: Ie381fcddd98a65c019a44cbde9054db6fafb88f4
In anticipation of puppet 4, start trying to deal with puppet 4 things
that can be helpfully predicted by puppet lint plugins. Also fix errors
caught by the puppet-lint-absolute_classname-check gem and arrow
alignment errors now caught by the bundle-installed puppet-lint. Avoid
trying to catch empty_string errors for now.
Change-Id: Ie881a8ac7539fbcbc283c3a572efe2e7562ac426
Java does not like to make connections to untrusted HTTPS hosts, and
makes it rather difficult to not verify the certificate and its CA
chain, like our Python scripts are currently doing. To that end, drag
down the CRT for the configured Zanata server, and inject it into the
default Java keystore -- since passing a non-default keystore would
also require changes to Zanata.
Change-Id: If6aaf4d560acc25e626027d82ed103dd46328802
Depends-On: I32ef72eba436c338052be2eb83ce39e4400d047c
Zanata has released 3.7.1, we should use it. I have tested it on a VM,
and aside from unable to login, it looks fine. Carlos has checked the
configuration changes for the move, and none look to be required.
Change-Id: Iaba85f6b049a191c0a753a95c07e7a875848fcb6
Deploying puppet-zanata from scratch can sometimes fail because
wildfly has yet to be unpacked when puppet goes to copy the ROOT.war
into the deployment directory, which does not yet exist. Fix this
issue by adding a requirement for tar to run as well.
Change-Id: Id0229864b64ce39c70b559e1eb5d94822f1988ec
Co-Authored-By: Colleen Murphy <colleen@gazlene.net>
puppet-httpd is the openstack-infra version of puppetlabs-apache
(0.0.4) release.
This patchset will remove the puppetlabs-apache namespace from -infra
allowing for possible future patchsets to use newer puppetlabs-apache
modules.
Change-Id: Id258092e1c1e21800c24512b5bdfee45b0939bca
Signed-off-by: Paul Belanger <pabelanger@redhat.com>
The jboss.bind.address.management variable keeps being
automatically flipped to 0.0.0.0 so it's triggering a
template refresh with every puppet run. Just put that
in the template.
Change-Id: Id7faae2b9937bd1bd016b2cea7329702d0b951e2
Due to using an undefined variable when determining the server id,
the client manifest was writing an incorrectly formatted zanata.ini
file. Furthermore, the intent of the change did not go far enough,
since it would not strip off the URI scheme or port. Due to the
complexities of parsing URIs in regular expressions, use a Puppet
function for the heavy lifting.
Change-Id: I754ee54f805c91f5548b2cf270b23c68eed3959c
Update the version of zanata that the module installs to 3.6.2. I've
tested the resulting install and Zanata launches.
Change-Id: I7c02cf208bb744fd0f259e85d6757ed924470d54