A Whirlwind Tour of Etsy's Monitoring Stack

It's no secret that at Etsy we are big fans of small, incremental and frequent changes and tight feedback loops. This is how we make it possible to deploy changes to our main codebase more than 50 times a day and also safely apply changes to our infrastructure in a continuous fashion. It enables us to rapidly fix bugs and roll out features in our application stack and infrastructure. This however would not be possible without a tight feedback loop and a myriad of monitoring tools that keep us informed about changes and
possible problems in every nook and cranny of the Etsy stack, no matter if it's a network change event, systems or application level performance or how bad the last week of on-call rotation was.

Transcript

1. A Whirlwind Tour of Etsy's Monitoring Stack Daniel Schauenberg dschauenberg@etsy.com @mrtazz
3. @mrtazz
4. @mrtazz
5. @mrtazz Item by TheBackPackShoppe
6. How comfortable are you deploying a change right now?
7. “If this is your first day at Etsy, you deploy the site”
10. @mrtazz Ganglia • System level metrics • Instance per DC/environment • > 220k RRD files • Fully configured through Chef role attributes
11. @mrtazz Rainbow Graphs!
12. @mrtazz StatsD • Single instance on one server • Traffic mostly from 70 Web & 24 API servers • Node.js • Heavy Sampling • Graphite as backend
13. @mrtazz
14. @mrtazz Graphite • Application level metrics • 96G RAM, 20 Cores, 7.3T SSD RAID 10 • 525k metrics/minute • Mirrored Master/Master Setup • Functionally sharded relays
15. @mrtazz CNAME relays relays caches caches statsdtimers statsdcounts statsd chef logster fqld search generic
16. @mrtazz
17. @mrtazz
18. @mrtazz Syslog-Ng • Web, Search, Gearman, Photos, Nagios, Network, VPN • 1.2GB written/minute • Chef role attribute based config • Rule ordering!
20. @mrtazz github.com/etsy/logster • Extract metrics from log files • Written in Python • Runs every minute via cron
21. @mrtazz Splunk • Indexes all of our log files • Easy search for patterns • Saved searches for interesting ones • Basically using it as a glorified grep
22. @mrtazz Logstash • Experiment status • Makes it easier integrate different sources • Easy to set up in dev environment • Trying to figure out where/how it fits into our infrastructure
23. @mrtazz Eventinator • Tracks all events in our infrastructure • Chef runs and changes • DNS changes • Network • Deploys • Server provisioning and decommissioning • ~ 12 million events in the last 2 years
24. @mrtazz
25. @mrtazz Chef • rules everything around me • Same cookbooks on prod and dev • every node runs Chef every 10 minutes • ton of knife plugins and handlers
26. @mrtazz
27. @mrtazz > 120 recipes
28. @mrtazz
29. @mrtazz Nagios
30. @mrtazz Nagios • 2 instances in each DC/environment • Fully Chef generated configuration • Service checks and contacts in git • Notifications via email->SMS gateway • ~75% ops on-call
31. @mrtazz github.com/lozzd/nagdash
32. @mrtazz
33. @mrtazz
34. @mrtazz
35. @mrtazz Nagios Herald • Add context to nagios alerts • What are the first 5 things you do when you get paged? • You already have the phone in your hand • nagios notification handler
36. @mrtazz
37. @mrtazz The Toys are real
38. @mrtazz There’s another side of heaven
39. @mrtazz Ops Weekly
40. @mrtazz Ops Weekly
41. @mrtazz Summary • Set of trusted tools • Enhance where they come short • Try out new things • Write tools where applicable • Continuous monitoring and adaptation
42. @mrtazz codeascraft.com etsy.com/codeascraft/talks etsy.github.com etsy.com/careers
43. @mrtazz Questions?
44. A Whirlwind Tour of Etsy's Monitoring Stack Daniel Schauenberg dschauenberg@etsy.com @mrtazz