Facepalm to Foolproof: Avoiding Common Production Pitfalls
summary
A checklist of concrete fixes: stop running WEBrick in production and use Puma with a sane worker count (typical Rails process ≈180–200MB; too many workers causes swap and huge slowdowns). Set force_ssl or redirect in nginx, use the asset pipeline helpers (image_tag / asset-url) and add files to the precompile list, log to stdout per twelve‑factor, deploy with Mina or git push Heroku, never commit S3 keys — prefer secrets.yml or dotenv-rails/ENV, add DB indexes, fix N+1s with includes, and offload slow ops to background jobs.
AI-written summary · about
chapters (5)
- 0:00Start
- 8:35Memory Enemies
- 10:17Watch Your Memory Usage
- 13:11The Asset Pipeline
- 23:52Composite Indexes
related talks
32mRails Sustainable Productivity
20mMaster the Rails Asset Pipeline: Best Practices for Apps & Gems
38m"Application Security for Rails Engineers"
43mSecuring Your Site
33mDo the Right Thing: Practical lessons for your..
31mTricks and treats for new developers
31mOperating Rails: What About After You Deploy?
22mKeep Ops Happy, Designing for Production