Supervise Your Akka Actors
summary
An Akka + Spray Scala architecture replaced fragile Talend FTP batch jobs, centralizing business logic and adding supervision strategies, a deadletter listener, and metrics integrated with Metrics/Graphite and Kamon. The system processes about 20,000 nightly announcements, throttles a weak remote web service, makes failed messages visible and replayable, and cut nightly processing from several hours to about one hour; deployment took roughly five months and has been in production for a month, with further work on webservice scaling and removing the FTP bottleneck.
AI-written summary · about
chapters (20)
- 0:00Table of Contents
- 0:36Outline
- 1:17Context
- 2:40Message Loss
- 4:12Technical Malfunctions
- 5:56Levers
- 6:29Why Akka?
- 7:59Why Java?
- 9:40Supervision with Akka
- 11:10Supervision Example
- 11:54Supervision Management
- 13:37Developed Solution
- 18:12The metricFactor
- 19:35metricFactor Configuration
- 20:40Adding Metrics with SPAR
- 21:40Issues Encountered
- 21:50Actor Design
- 24:28Event Model
- 26:35DDD
- 33:26Q&A
related talks
20mTracing of Large-scale Actor Systems
43mMonitoring Reactive Applications
44mPatterns for Streaming Telemetry with Akka Streams
41mMonitoring Reactive Applications
29m“If you Don’t Monitor your Infrastructure, you Don’t Own it!” Regain Control Thanks to Prometheus
35mMonitoring Distributed Java Applications at Scale
30mRunning Highly Scalable Tracing, Logging, Monitoring & Metric System in Production
15mTRACING AKKA STREAMS