Discovering Processes
summary
Systems run with millions of processes and the concrete problem is getting the pid for message passing. Solutions compared: spawning and passing pids at startup (breaks under restarts), supervisor queries (racey and fragile), simple atom registration (fast but unsafe if atoms grow), and a rich process registry implemented with ETS. The practical recommendation is gproc for local rich registration; it scales for dynamic sessions while global registries remain brittle.
AI-written summary · about
chapters (7)
- 0:00Start
- 2:12Process Discovery
- 2:47Local Process Discovery
- 13:14Process Register
- 20:03Naive Implementation
- 22:39Local Registration
- 25:26Third-Party Registry
related talks
28mKnocking On A Process’s Door: Introspecting Live Python Processes
39mProcess Supervision Strategies
41mThe Future of Tracing
24mCommunity Driven Breakout Session: Continuous Profiling
35mCase Study: A Thundering Herd in the Wild
24mFrom PIDs to Pods: The Life Cycle of an eBPF-Autoinstrumented..
24mDebuggin a Large Node Cluster
18mFind out what your processes are capable of with ebpf