A hubot script that keeps the hubot Heroko web dyno alive
CoffeeScript JavaScript Shell
Latest commit 880370a Jan 9, 2016 @treby treby Prepared to release 1.0.2.
Permalink
Failed to load latest commit information.
script first commit Oct 10, 2014
src Keepalive in auth Dec 8, 2015
test first commit Oct 10, 2014
.editorconfig first commit Oct 10, 2014
.gitignore first commit Oct 10, 2014
.travis.yml first commit Oct 10, 2014
Gruntfile.js first commit Oct 10, 2014
README.md Merge pull request #15 from abeger/readme_timezone_fix Dec 5, 2015
index.coffee first commit Oct 10, 2014
package.json Prepared to release 1.0.2. Jan 9, 2016

README.md

hubot-heroku-keepalive

A hubot script that keeps the hubot Heroku free web dyno alive.

Note that a free Heroku dyno can only run for 18 hours/day, so it will be required to sleep for at least 6 hours. Accessing your Hubot during a sleep period will wake it, but it will return to sleep after 30 minutes.

Installation

In hubot project repository, run:

npm install hubot-heroku-keepalive --save

Then add hubot-heroku-keepalive to your external-scripts.json:

[
  "hubot-heroku-keepalive"
]

Configuring

hubot-heroku-keepalive is configured by four environment variables:

  • HUBOT_HEROKU_KEEPALIVE_URL - required, the complete URL to keepalive, including a trailing slash.
  • HUBOT_HEROKU_WAKEUP_TIME - optional, the time of day (HH:MM) when hubot should wake up. Default: 6:00 (6 am)
  • HUBOT_HEROKU_SLEEP_TIME - optional, the time of day (HH:MM) when hubot should go to sleep. Default: 22:00 (10 pm)
  • HUBOT_HEROKU_KEEPALIVE_INTERVAL - the interval in which to keepalive, in minutes. Default: 5

You must set HUBOT_HEROKU_KEEPALIVE_URL and it must include a trailing slash – otherwise the script won't run. You can find out the value for this by running heroku apps:info. Copy the Web URL and run:

heroku config:set HUBOT_HEROKU_KEEPALIVE_URL=PASTE_WEB_URL_HERE

If you want to trust a shell snippet from the Internet, here's a one-liner:

heroku config:set HUBOT_HEROKU_KEEPALIVE_URL=$(heroku apps:info -s | grep web-url | cut -d= -f2)

HUBOT_HEROKU_WAKEUP_TIME and HUBOT_HEROKU_SLEEP_TIME define the waking hours - between these times the keepalive will ping your Heroku app. Outside of those times, the ping will be suppressed, allowing the dyno to shut down. These times are based on the timezone of your Heroku application which defaults to UTC. You can change this with:

heroku config:add TZ="America/New_York"

Waking Hubot Up

This script will keep the dyno alive once it is awake, but something needs to wake it up. You can use the Heroku Scheduler to wake the dyno up. Add the scheduler addon by running:

heroku addons:create scheduler:standard

The scheduler must be manually configured from the web interface, so run heroku addons:open scheduler and configure it to run curl ${HUBOT_HEROKU_KEEPALIVE_URL}heroku/keepalive at the time configured for HUBOT_HEROKU_WAKEUP_TIME.

Heroku Scheduler Screenshot

Note that the Scheduler's time is in UTC. If you changed your application's timezone with TZ, you'll need to convert that time to UTC for the wakup job. For example, if HUBOT_HEROKU_WAKEUP_TIME is set to 06:00 and TZ is set to America/New_York, you'll need to set the Scheduler to run at 10:00 AM UTC.

Legacy Support

Hubot has for a long time had it's own builtin way to keep its web dyno alive, but this is an extraction of that behavior.

The legacy support uses the HEROKU_URL environment variable instead of HUBOT_HEROKU_KEEPALIVE_URL, so for forward compatability, hubot-heroku-keepalive will also use HEROKU_URL if it's present, and will also disable the legacy keepalive behavior if it's present.

Development

The best way is to use npm link and make sure to point HUBOT_HEROKU_KEEPALIVE_URL at the right place:

hubot-heroku-keepalive$ npm link
hubot-heroku-keepalive$ cd /path/to/your/hubot
hubot$ npm link hubot-heroku-keepalive
hubot$ export HUBOT_HEROKU_KEEPALIVE_URL=http://localhost:8080/
hubot$ bin/hubot