OpenSearch/vagrant

Vagrant build environment

This provides a vagrant box for building the C++ side of Ml (and the Java side, although that is easily accomplished outside vagrant).

Provisioning the box will take a fair amount of time, since it needs to download and compile a number of dependencies.

Details

• Ubuntu Trusty64 (14.04.5 LTS)
• 25% of host's memory
• 100% host's cores
• Maps ml source repository to /home/vagrant/ml/src
  • Directory is shared with the host, so you can point your IDE to the ml repo and build inside vagrant
• Maps ml build directory to /home/vagrant/ml/build
• Changes into /home/vagrant/ml/src on login

Pre-baked box

Don't feel like compiling the entire box? No fear, there's a pre-baked box available on S3. It is ~1.1gb to download:

# Change into some random directory to download the box.
# Doesn't matter where this goes, but *cannot* go into prelert-legacy/vagrant
$ cd ~/some_directory

# Export the path to your prelert-legacy repo. This is so the box knows where
# to sync the folders
$ export ML_SRC_HOME=/path/to/prelert-legacy

# Download the box from S3
$ s3cmd get s3://ml-elastic-dump/ml_env.box
  # ...
  # Downloading...
  # ...

$ vagrant box add ml ml_env.box
$ vagrant init ml
$ vagrant up
$ vagrant ssh

  Welcome to Ubuntu 14.04.5 LTS (GNU/Linux 3.13.0-96-generic x86_64)
  ...
  ...
  Last login: Tue Oct  4 16:06:32 2016 from 10.0.2.2

vagrant@vagrant-ubuntu-trusty-64:~/ml/src$

Once you've logged into the box, you'll be in the ml source directory. You can build immediately via:

vagrant@vagrant-ubuntu-trusty-64:~/ml/src$ gradle cppmake

The pre-baked box has already compiled ml once, so subsequent compilations should happen considerably faster.

Compiling from Scratch

If you feel like compiling everything from scratch instead of downloading the pre-baked box, simply vagrant up and let the provisioners run:

$ cd prelert-legacy/vagrant
$ vagrant up
  # ...
  # wait while vagrant provisions
  # ...
$ vagrant ssh

  Welcome to Ubuntu 14.04.5 LTS (GNU/Linux 3.13.0-96-generic x86_64)
  ...
  ...
  Last login: Tue Oct  4 16:06:32 2016 from 10.0.2.2

vagrant@vagrant-ubuntu-trusty-64:~/ml/src$

Once you've logged into the box, you'll be in the ml source directory. You can build immediately via:

vagrant@vagrant-ubuntu-trusty-64:~/ml/src$ gradle cppmake
  # ...
  # much building
  # ...

Suspending your box

Once you've provisioned a box, you can use vagrant suspend to "sleep" the box. This has the advantage of rebooting quickly when you vagrant up, and returns you to exactly what you were doing. On the downside, it eats more disk space since it needs to sleep the entire image.

You can alternatively use vagrant halt, which gracefully powers down the machine. Rebooting via vagrant up takes longer since you are rebooting the entire OS, but it saves more disk space.

Fixing a broken box

If you accidentally kill the provisioning process before it completes, you can attempt to reprovision it with vagrant reload --provision. That will run the provisioners that did not complete previously.

If your box is still horribly broken, you can destroy it with vagrant destroy and try again with vagrant up