Reps Infrastructure migration report

(Yousef Alam) #1

As of early this week, is now running on ParSys infrastructure. This has allowed us to modernize the deployment and processes used which means less issues and quicker times to develop and deploy new features.

A lot of work has gone into this:
Our docker setup has been overhauled to work for production deployments

Whitenoise has been added to serve static files directly, which should speed up loading times

We’re now using Amazon SES to send emails, adding SPF and DKIM support. The insecure practice of sending emails as our users is gone, and all emails will now come from a single address.


We’re getting closer to our ideal deployment flow and our remo developers @johngian and @akatsoulas now only need to push to the production branch to do a production deploy.

Our last prod deploy took only 3 minutes and 18 seconds from the push to GitHub! The (super simple) deployment file we use is also public:

The future
Looking forward we’re planning on fixing some of the performance regressions from the migration (mainly gzipping static content) and then possibly supercharging that with a CDN. We’re also planning on moving tests to our own infrastructure so contributors spend less time waiting to see if their PRs have passed. Following currently ongoing work for general security hardening of our AWS instance, we will resubmit the Reps portal for a Rapid Risk Assessment with the InfoSec team to ensure we’re up to Mozilla’s data and operational security standards.

(Henrik Mitsch) #2

Super awesome!! Honored to be part of this team. :heart: