Building Distributed Systems in Elixir: Part 4 — Process Linking

The article shows that process links create a bidirectional failure relationship between BEAM processes. Unlike monitors, links are not just observation; an abnormal exit in a linked process can terminate the other process unless exits are trapped.

It also shows how Process.flag(:trap_exit, true) converts the exit signal into a regular {:EXIT, from_pid, reason} mailbox message. That lets the parent process receive and inspect the worker’s failure, while still leaving restart decisions for later supervisor logic.

Reading notes

#

• spawn_link/1 starts and links a process atomically, avoiding a gap where the child could crash before the link exists.
• A linked crash propagates as an exit signal, so an abnormal worker exit can terminate the parent.
• Normal exits do not ordinarily take down linked processes that are not trapping exits.
• Process.flag(:trap_exit, true) changes the incoming exit signal into a mailbox message.
• The trapped message has the shape {:EXIT, from_pid, reason}.
• ^worker pins the receive pattern to the specific linked worker PID.
• Trapping exits lets the parent observe the failure and stay alive.
• Links and monitors are different: monitors are one-way and produce {:DOWN, ...}, while links are bidirectional and affect process lifecycles.
• The article treats links as a step before supervision, since restart policy still has to be decided separately.
• The next part will build a supervisor from scratch and use the trapped exit to start a replacement worker.

Series

#

Building Distributed Systems in Elixir, by krishnadaspc. Research: Distributed Systems Patterns.

• Part 1 — Stateful Process Without GenServer
• Part 2 — Correlated Request–Reply
• Part 3 — Process Monitoring
• Part 4 — Process Linking
• Part 5 — Supervisor From Scratch
• Part 6 — Named Processes
• Part 7 — Worker Pool
• Part 8 — Publish / Subscribe
• Part 9 — Backpressure