Skip to content

Latest commit

 

History

History
111 lines (84 loc) · 4.16 KB

bluemix.md

File metadata and controls

111 lines (84 loc) · 4.16 KB
Raw
permalink
/docs/deploying/bluemix/

Deploying to Bluemix

If you've been following along with Getting Started, it's time to deploy so you can use it beyond just your local machine. IBM Bluemix is a way to deploy hubot as an alternative to Heroku. It is built on the open-source project Cloud Foundry, so we'll be using the cf cli throughout these examples.

Hubot was originally very closely coupled to Heroku, so there are a couple of things to clean up first that we don't need or that might get in the way on another platform:

  • remove Procfile as we'll create the manifest.yml that Bluemix needs in a moment

  • remove the hubot-heroku-keepalive line from external_scripts.json and also remove the related npm module (it causes errors on other platforms):

    npm uninstall --save hubot-heroku-keepalive

In preparation for working with Bluemix, install the Cloud Foundry CLI, and create a Bluemix Account.

First we need to define a manifest.yml file in the root directory. The contents of the manifest at the bare minimum should look like:

applications:
- name: myVeryOwnHubot
  command: ./bin/hubot --adapter slack
  instances: 1
  memory: 512M

In this example, we're using the slack adapter, if you choose slack as your adapter when creating a hubot this will work, otherwise add the hubot-slack module to your package.json. Change the name of your hubot in the manifest.yml file because otherwise your application will clash with someone else's who already deployed an app called this! There are many more useful things you can change about your hubot using the manifest file, so check out these docs for more information.

You then need to connect your hubot project to Bluemix:

$ cd your_hubot_project
$ cf api https://api.ng.bluemix.net
$ cf login

Note that the cf api command changes per Bluemix region so to deploy somewhere other than "US South", replace this api as appropriate. The cf login command will prompt you with your login credentials.

Next, we need to set up our environment variables, but we need to create the app first. It won't work properly without the environment variables it needs, so we'll first of all use the --no-start flag to deploy but not attempt to start it.

$ cf push NAME_OF_YOUR_HUBOT_APP --no-start

Now the app exists, we can set its environment variables. To access slack, you'll need a slack token from the "Apps and Integrations" page; it's visible when you go to create a slackbot. Copy that token and set it as an environment variable called HUBOT_SLACK_TOKEN, like this:

$ cf set-env NAME_OF_YOUR_HUBOT_APP HUBOT_SLACK_TOKEN TOKEN_VALUE

If you have other environment variables to set, such as configuring the REDIS_URL for hubot-redis-brain, this is a good time to do that.

Finally, we're ready to go! Deploy "for real" this time:

$ cf push NAME_OF_YOUR_HUBOT_APP

You should see your bot connect to slack!

Further Reading

Troubleshooting

Bot doesn't connect

Check your logs for more information using the command cf logs YOUR_APP_NAME --recent. If you have NodeJS installed locally, you can also try running the bot on your local machine to inspect any output: simply do bin/hubot from the top level of the project.

Bot crashes repeatedly

It is sometimes necessary to to assign more memory to your hubot, depending which plugins you are using (if your app crashes with error 137, try increasing the memory limit).