Running node.js
in production
Ken Perkins,
node.js developer advocate @ rackspace
OK
So what does in production even mean?
It's all about the setup.
Everything starts with
Configuration
If you setup dev the same way you setup prod,
you're gonna have a good time.
What does that even mean?
Use puppet, chef, docker, etc. What you want is repeatable, deterministic infrastructure for every environment.
Node is great at high concurrency, but don't forget about OS tuning
fd limits, open sockets, etc
Now you need to
build & deploy
Git, rsync, grunt, ftp, etc.
A word of caution about
dependencies & npm:
#1 Always use semver
#2 npm deploys break often
#3 checking in node_modules is mostly ok*
* Native modules, pkg bloat can cause difficulties
What about npm shrinkwrap?
No Checksum (--force fucks you)
Still depends on moving parts (registry.npmjs.org)
Lossy conversion (npm shrinkwrap > npm install > npm shrinkwrap yields different npm-shrinkwrap files)
Running your app
or
> service your-app start
Use process management
for your OS
upstart, runit, systemd, etc
Turn your node app
into a proper service
Use tried & true tools that are designed
for exactly what you need: to keep your process running.
Now keep your app healthy
Logging & monitoring can't be afterthoughts
Great tools for logging
Loggly, Papertrail, rsyslog, Winston, Airbrake, Concurix, etc
Use multiple
monitoring strategies
Remote Ping Probes (i.e. pingdom)
App/Svc Level Monitoring (i.e. get /api/service-ping === 200)
Process Level Monitoring (i.e. service your-app status)
System Level Monitoring (mem, cpu, swap, etc)
& more...
The point isn't just to prevent shit from breaking, it's also how do you deal with it when it does.
The key takeaway is that running node in production is all about mitigating liabilities.
fin
@kenperkins (github/twitter)
http://j.mp/node-in-prod
