Supervision¶

Category: Communication · Status: stub · Lessons: chapter 05, Message passing (planned)

One line: Letting a failed actor or task crash and having a supervisor restart it, instead of defending against every error inside it — the approach of Erlang and OTP.

Also called: supervision tree, let it crash.

How it connects¶

flowchart LR
  n_actor_model["Actor model"]
  n_otp_behaviours["OTP behaviours"]
  n_supervision["Supervision"]
  n_otp_behaviours -->|uses| n_supervision
  n_supervision -->|uses| n_actor_model
  classDef center stroke-width:3px
  class n_supervision center
  classDef outside stroke-dasharray: 4 3
  class n_actor_model,n_otp_behaviours outside

Is built on: Actor model
Is used by: OTP behaviours
See also: Structured concurrency

In each language¶


Kotlin	`SupervisorJob` and `supervisorScope` ↗: a failing child does not cancel the supervisor or its other children
Erlang and Elixir	OTP's `supervisor` ↗ keeps its children alive by restarting them, with strategies such as `one_for_one`; Elixir's `Supervisor` ↗
The operating system	systemd's `Restart=` ↗, for example `on-failure`, restarts a service automatically
Elsewhere	Akka ↗: a typed actor that throws is stopped unless a supervision strategy, such as restart, says otherwise

Where to read more¶

In the books: The Little Elixir & OTP Guidebook, Benjamin Tan Wei Hao — ch. 5, 'Concurrent error-handling and fault tolerance with links, monitors, and processes'
In the books: Concurrency in Go, Katherine Cox-Buday — ch. 5, 'Concurrency at Scale' → 'Healing Unhealthy Goroutines'
In the books: Async Rust, Maxwell Flitton, Caroline Morton — ch. 8, 'The Actor Model' → 'Creating Actor Supervision'
In the books: Learning Concurrent Programming in Scala, Aleksandar Prokopec — ch. 8, 'Actors' → 'Actor supervision'
In the books: Effective Concurrency in Go, Burak Serdar — ch. 10, 'Troubleshooting Concurrency Issues' → 'Detecting failures and healing'
In the books: Erlang and OTP in Action, Martin Logan, Eric Merritt, Richard Carlsson — ch. 1, 'The foundations of Erlang/OTP' → '– Erlang’s fault tolerance infrastructure'