BACKWARDS INCOMPATIBILITIES
- Requires Vagrant 1.7.2 or later.
- Because of the upgrade to Vagrant 1.7.2, we were forced to change the locations/names of the Puppet manifests and
modules directories on the target machines. See the "BREAKING CHANGES" section in the
Vagrant Changelog for 1.7.2.
Vagrant 1.7.2+ adds MD5-hash based suffixes to the manifests and modules directories, which broke our assumption
on where to find these directories on the target machines (so we actually belong to the "minor number of people" who
are affected by this change in Vagrant). The new directories look like:
modules-8048a9ab32fda4a984b584e79dfe8cb7
. Thankfully, in the current Vagrant implementation these suffixes are identical across deployments in our case, so the current fix simply renames the respective directory locations. - We use Java 7 by default.
IMPROVEMENTS
- Deploy the Avro schema registry of
Confluent Platform on
kafka1
(first Kafka broker). - Update puppet_wirbelsturm-common to 1.0.5, which by default installs netcat on each machine.
- Update puppet-storm to 1.0.14.
- Update puppet-zookeeper to 1.0.8, which e.g. fixes ZK quorum deployments.
BUG FIXES
vagrant-scp.sh
now exits with a non-zero exit code in case of errors.
BACKWARDS INCOMPATIBILITIES
- Requires Vagrant 1.6.1 or later.
- Requires vagrant-hosts plugin 2.1.4 or later.
IMPROVEMENTS
- Add
ansible-playbook
wrapper script. - bootstrap: Use new GitHub.com URL for retrieving raw user content.
- WS-10: source rvm script after ruby-bootstrap so that libarian-puppet is found (thanks neeleshs)
- WS-6: Fix Puppet installation on latest Amazon Linux AMI when creating custom AMIs
- WS-5:
$MYDIR
in AWS scripts should be$MY_DIR
(thanks pbwilmot) - WS-2:
bootstrap
now supports the--skip-ruby
parameter. If set,bootstrap
will skip the installation of Ruby/rvm/bundler/gems. (thanks bzz) - Improve
update-wirbelsturm.sh
- Environment
default_environment
:- Enforce 'UTC' timezone
- Enforce the use of Puppet 3.3.1 to have more deterministic builds compared to the previous setup where we always installed the latest Puppet version currently available.
- Update puppet-diamond to 1.0.2.
- Update puppet-graphite to 0.0.5.
- Update puppet-kafka to 2.0.3, which e.g. adds experimental support for writing Kafka data to tmpfs.
- Update puppet-redis to 1.0.4.
- Update puppet-statsd to 1.0.4.
- Update puppet-storm to 1.0.12, which e.g. supports running the new logviewer daemon and DRPC support.
- Update puppet-supervisor to 1.0.6.
- Update puppet-wirbelsturm_build to 1.0.3.
- Update puppet-wirbelsturm_common to 1.0.4, which hardens the NTP configuration (if NTP management is enabled).
- Update puppet-wirbelsturm_yumrepos to 1.0.2.
- Update puppet-zookeeper to 1.0.7.
BUG FIXES
deploy
now correctly writes its log files to the directoryprovisioning-logs/
in the base Wirbelsturm directory, even whendeploy
is executed from within sub-directories of Wirbelsturm.- Read
/etc/environment
to makeFACTER_*
env vars available on reboot. This makes clusters work properly upon instance reboots on AWS. (thanks stevencdavis) [GH-20]
- Initial release