I’ve been experimenting with headless chrome for a Link Unshortener tool I’ve built to take screenshots of websites. I’ve been using BrowserShot which is great. It’s a php wrapper around Puppeteer which makes it simple to use in Laravel. To experiment more with Puppeteer, I wanted to get a node app running on Heroku. Overall it’s pretty straight forward but there are a few gotchas.
Here is a sample project that should get you started. Clone this and take a look at the source. You’ll notice I specified the node.js version in the package.json file. This is required for Heroku to know what version of node to use. Another issue I ran into is to run Puppeteer on Heroku you have to specify –no-sandbox. They last hurdle was adding the puppeteer Heroku buildpack. Follow the steps below and you should have a working screenshot app running locally and on Heroku. These instructions assume you are familiar with node and Heroku.
Setup Steps
- Clone sample project: git [email protected]:timleland/headless-chrome.git
- Make sure you are in the correct directory: cd headless-chrome
- Install node dependencies: npm install
- Run the app to test: npm start
- Open your browser to: http://localhost:8080/?url=google.com
- You should have download a png screenshot of google.com
- To deploy to Heroku make sure you have the Heroku cli tool installed
- Create a new app on Heroku: heroku create APPNAMEHERE
- Add the puppeteer heroku buildpack: heroku buildpacks:add https://github.com/jontewks/puppeteer-heroku-buildpack
- Deploy to Heroku: git push heroku master
- Now just append the query string url=site.
- You should now have a working Heroku app that will screenshot any url you send it.
Leave your questions and feedback in the comments below.
Also published on Medium.
I had to add the Node buildpack as well:
`heroku buildpacks:set heroku/nodejs`
Did you get it to work?
@Tim
Yeah, sorry I meant: `heroku buildpacks:add heroku/nodejs` (Note the `add` instead of `set` as `set` seems to overwrite the existing buildpacks)
Thanks
Thanks! It has been updated.
Hello!
Thanks for the tutorial in the first place. I’ve followed the steps but I got the following error in my console:
/app/node_modules/puppeteer/.local-chromium/linux-508693/chrome-linux/chrome: error while loading shared libraries: libcairo-gobject.so.2: cannot open shared object file: No such file or directory
Do you know anything about this issue?
Thanks! Take a look at step 9 and make sure you added the build pack.
I’ve already done that and it’s still not working. I’ve added the libcairo-gobject to the buildpack and I’ve managed to get it run.
Ok thanks for letting me know.
Thank you,.
#1
2018-01-08T21:50:44.412485+00:00 heroku[router]: at=error code=H14 desc=”No web processes running” method=GET path=”/?url=google.com” host=infinite-headland-44709.herokuapp.com request_id=85fa1725-7ee0-4404-bb19-8dd54fbfffbe fwd=”24.5.137.98″ dyno= connect= service= status=503 bytes= protocol=https
2018-01-08T21:50:44.862414+00:00 heroku[router]: at=error code=H14 desc=”No web processes running” method=GET path=”/favicon.ico” host=infinite-headland-44709.herokuapp.com request_id=46644015-4aed-4038-a941-80321c466eb0 fwd=”24.5.137.98″ dyno= connect= service= status=503 bytes= protocol=https
#2 Tried doing this
heroku ps:scale web=1
got
Scaling dynos… !
▸ Couldn’t find that process type.
#4 Tried removing buildpacks
heroku buildpacks:remove heroku/nodejs
heroku buildpacks:remove https://github.com/jontewks/puppeteer-heroku-buildpack
#5 Tried doing this now
heroku ps:scale web=1
got
Scaling dynos… !
▸ Couldn’t find that process type.
Not sure if any issues with https://github.com/timleland/headless-chrome.git
Were you able to get it to work? Did you try cloning my project?
I had this same problem and fixed it by running the following:
heroku buildpacks:set heroku/nodejs
Thanks for sharing!
After following your instructions I’m getting this weird error when launching puppeteer:
TypeError: input.on is not a function
2018-01-19T06:16:08.710918+00:00 app[web.1]: at new Interface (readline.js:181:11)
2018-01-19T06:16:08.710919+00:00 app[web.1]: at Object.createInterface (readline.js:64:10)
2018-01-19T06:16:08.710922+00:00 app[web.1]: at waitForWSEndpoint (/app/node_modules/puppeteer/lib/Launcher.js:195:10)
2018-01-19T06:16:08.710921+00:00 app[web.1]: at new Promise ()
2018-01-19T06:16:08.710923+00:00 app[web.1]: at Function.launch (/app/node_modules/puppeteer/lib/Launcher.js:133:39)
2018-01-19T06:16:08.710924+00:00 app[web.1]: at
2018-01-19T06:16:08.805193+00:00 app[web.1]: events.js:183
2018-01-19T06:16:08.805197+00:00 app[web.1]: throw er; // Unhandled ‘error’ event
2018-01-19T06:16:08.710920+00:00 app[web.1]: at Promise (/app/node_modules/puppeteer/lib/Launcher.js:196:25)
Any idea what’s going on? Your help is much appreciated :).
Nothing stands out from the error log. You could try opening an issue on the github page.
If anyone is getting page crashes (only in Heroku), try adding the –disable-dev-shm-usage flag. I put a little example code here: https://github.com/heroku/heroku-buildpack-google-chrome/issues/37#issuecomment-380877297
Thanks for sharing
For those having problems, firstly go on the Heroku web dashboard, click on Settings and scroll down to build packs, ensure you have the following 2 packs added:
https://github.com/jontewks/puppeteer-heroku-buildpack
heroku/nodejs
Secondly, you’ll need a Procfile. Create one with the following content:
web: node index.js
…and use git push heroku master to deploy it.
Good tip thanks!
Has anyone tried to start puppeteer with –proxy-server option on Heroku? I’m getting always timeout when attempting to connect (not a problem of the proxy, it works on my local env)
Thanks,
Alessandro
Thanks for this. You saved my day.
No problem!