Don't make the symlink here since the file won't exist.

Use ansible instead of scp.

Ansible will use the correct ssh params so that we don't have to worry about host key checking
which will become a problem as we build sandboxes at the same domain more than once.

Also check the return code to make sure it succeeded.

'vars' var doesn't work anymore.

Make migration jenkins jobs track migrations in edx-private.txt

Don't override the versions is server-vars.yml.

Turn off new IDA migration checks because they aren't working correctly ...

Add cybersource urls.

We don't need it as a requirement it is in core as of 2.7.

See https://pypi.python.org/pypi/argparse for more details.

Use templated ElasticSearch config for single-instance deployment

Validate certs everywhere

Until now it's only been used for clustered deployment,
but it works fine in single-instance mode, too.

roll back hotfix file size limit

Add more travis testing.

Allow for some overhead in profile image uploads

update maximum profie image size

Minor update and typo fix for mariadb

Add ecommerce, insights, and analyticsapi to pre-supervisor checks

Set default for GIT_REPO_EXPORT_DIR

Require a newer version of pip.

Jibsheet/flower oauth

0.7.0 is from May 2014 and lacks OAuth support and a number of other
updates.

Switch to having Flower use a config file, enable OAuth

Google OAuth requires an additional 4 command line args or environment
variables, so it was time for Flower to get a proper config file.

The broker configuration remains a command line argument because it is
consumed by Celery's configuration parser, not Flower's configuration
parser.

I can't think of a reason our oauth keys need to be world readable

Move templates to live in our new directory layout

Paths under templates/ should mirror their default installation path on
disk to make it easier to correlate deployed files back to source
templates.

Fix the callback plugin for cases where stderr is a boolean and not a string.

Fix profile image backend docs