Scaling Chef Emotionally

on YouTube ↗33m

" summary

Slack runs Chef on a single stateful backend then migrates to AWS managed services (Aurora, managed Elasticsearch, NLB) and stateless EC2 frontends to stop outages; automated testing lands via Test Kitchen plus serverspec and a CI workflow that uses a custom gchen wrapper to run convergence and new-host tests pre-merge; release controls include feature flags and zonal rollouts, a CAB then a delegated deputy model, and an 84,000 cookbook versions incident that forces cleanup and drives a plan to shard Chef servers for scale.

AI-written summary · about

" chapters (12)

1. 0:00Scaling Chef at Slack
2. 2:31Manual scaling challenges
3. 6:34Re-architecting Chef backend
4. 9:42Developer workflow limits
5. 13:13Breaking point and freeze
6. 15:42Safety processes
7. 18:31Automated testing with gchen
8. 23:29Measuring infra safety
9. 26:14Upgrading Chef versions
10. 29:04Future sharding plans
11. 31:12Big events in scaling
12. 32:56Q&A

" related talks

• 18mChef On The TrainHeike Hallenberger · ChefConf 2018
• 34m"Acing" Infrastructure Testing with ChefBrian O'Connell · ChefConf 2017
• 42mCaptial One's Newly Designed Fully Automated, Scalable, and...John Casanova · ChefConf 2018
• 33mDeveloping for Multi-Component Environments While Keeping Your SanityFred Simon · Jenkins User Conference US West (JUC San Francisco) 2014
• 38mAn Approach to Air-Gapped DeploymentJoseph E. Gonzalez · ChefConf 2017
• 25mCooking with Chef - Your servers will thank youJames Golick · MountainWest RubyConf 2010
• 40mJenkins and Chef: Infrastructure CI and Application DeploymentDan Stine · Jenkins User Conference Boston 2014
• 38mChef for OpenStack OverviewMatt Ray · OpenStack Summit Portland 2013