How do you call db:seed on AWS Elastic Beanstalk apps?

Question

I know that you can run tasks as part of each deploy, but I only want to bootstrap the database once.

Answer 1

I used the information provided at this address to create a script that will run AFTER migrations and after each deploy: http://www.emind.co/how-to/how-to-run-rake-dbseed-in-amazon-elastic-beanstalk

I preferred this method so that I could keep track of the file on the EC2 instance. When I initially deployed this on my old servers (which were running Linux <1.0.9) I had no issue. However, I recently had to upgrade the server machines to 64bit Amazon Linux 2014.03 v1.0.9 running Ruby 2.0 (Puma) and the script began to fail. It will likely fail if you are using different Linux versions.

The key here is /usr/local/bin/ to your rake command to use the proper rake. I took this directly from the other scripts found at /opt/elasticbeanstalk/hooks/appdeploy/pre/:

#.ebextensions/db_seed.config
files:
  "/opt/elasticbeanstalk/hooks/appdeploy/pre/13_db_seed.sh":
    mode: "00755"
    owner: root
    group: root
    content: |
      #!/usr/bin/env bash
      . /opt/elasticbeanstalk/containerfiles/envvars
      cd $EB_CONFIG_APP_ONDECK
      su -c "leader_only /usr/local/bin/rake db:seed" $EB_CONFIG_APP_USER ||
      echo "Rake task failed to run, skipping seeding."
      true

It should be noted that this script is still not correct as I am not running "bundle exec rake" which is highly advised if you plan to run any rake command (it's a feature request currently on AWS for other scripts). If you wish to have ALL scripts run the bundle exec before "rake" take a look at the .config files posted here: https://github.com/alienfast/elastic-beanstalk

If this still doesn't work for you, I suggest sshing to one of your EC2 instances that are running your app, or deploy a new one running the same Linux/Ruby version and navigate to the "/opt/elasticbeanstalk/hooks/appdeploy/pre/" to see what the other scripts are doing.

I hope others find this useful!

Answer 2

It's a while since the question was asked, so perhaps you've figured it out. Anyway, you can add a file called (for instance) seed.config in the .beanstalk folder. Enter something like this and your seeds will be run:

container_commands:
  01seed:
    command: rake db:seed

Answer 3

to do commands on your elastic beanstalk instance, you can setup custom commands configurations that will get run when you do update your application. (In fact, there is a whole docs page dedicated to the different types of container commands you can do).

1. if you don't already have an .ebextensions directory in the root of your application, create one.
2. any file you put in here will be run. name it whatever you want as long as it ends with ".config" -- I arbitrarily named mine seed.config

3. borrowing from what hfogel wrote, you can put something in there like this:

   container_commands:
     01seed:
       command: rake db:seed
   

4. add this new code to your repo: git add . then git commit -m 'added seed config script' then git push

5. then push this new code to your aws eb instance(s): git aws.push

To verify that your command actually ran, go into your elastic beanstalk console and expand the details for your environment, and navigate to the logs tab, and refresh your logs and view them. Inside there, simply do a ctrl + f for "seed" until you see your seed command was run. If you don't see it in there somewhere, then it didn't run.

Answer 4

The only way I found was to ssh into the ec2 instance and run "rake db:seed RAILS_ENV=production" manually from /var/app/current.

Answer 5

Kristian's answer is close, but you should also add:

container_commands:
  seeddb:
    command: 'export HOME=/root; rake db:seed'
    leader_only: true

So that the DB is only seeded from 1 EC2 instance, rather than all of them at the same time. Depending on your EB deployment/version, the export HOME may or may not be needed.