Particular Particulars about "Zero Downtime Deploys"
summary
Claim: fully automated zero downtime deploys run for months with no scheduled outages, surviving Ruby upgrades and dozens of commits. How: Unicorn + NGINX + Runit + RVM + Capistrano + Chef, send USR2 restart to spawn new master and phased workers, write pid outside current release and demonize (-d), precompile assets before symlink, use unix sockets to nginx, keep seven releases, scale workers with TTIN/TTLO and verify via rack-statsd.
AI-written summary · about
chapters (15)
- 0:00Zero downtime deploys intro
- 2:1024/7 needs zero downtime
- 4:42Unicorn zero downtime server
- 7:15Zero downtime for uptime
- 9:23Continuous deployment without downtime
- 10:07Journey to zero downtime
- 12:34Unicorn zero downtime config
- 16:28USR2 signal seamless restarts
- 17:53runit and signals management
- 20:25Unicorn config insights
- 21:50Scaling workers without downtime
- 22:53Monitoring workers with rack statsd
- 24:41Zero downtime DB migrations
- 26:52Concurrent indexing shadow envs
- 28:39Summary and Q&A
related talks
40mStrategies for Zero Down Time, Frequent Deployments
32mZero Downtime Deploys Made Easy
39mZero downtime deploys for Rails apps
36mContinuous Deployment with Rails
32mZero downtime Rails upgrades
35mTen Years of Rails Deployment
30mZero Down Time Operations: How to Get Atleast 3 9's of Uptime
45mLarge-Scaled Deploy Over 100 Servers in 3 Minutes