aaronrenner
Best practices for connecting Supervisor's child processes to each other
I am trying to figure out if there’s an established best practice for connecting child processes within the same supervisor without using name registration.
I have the following processes
PortCacheStateServerWebServer- requires aPortCacheandStateServerfor startup
I am using a rest_for_one supervision strategy and trying to create a supervision tree like this.
I’m curious what the best way is to give WebServer a way to send messages to PortCache and WebServer without using name registration.
Here is what I currently have:
defmodule MyApp.Supervisor do
def start_link(_init_opt) do
# Start the supervisor with no children
with {:ok, sup} <- Supervisor.start_link([], strategy: :rest_for_one) do
{:ok, port_cache} = Supervisor.start_child(sup, PortCache)
{:ok, state_server} = Supervisor.start_child(sup, StateServer)
{:ok, _web_server} =
Supervisor.start_child(sup, {
WebServer,
port_cache: port_cache,
state_server: state_server
})
{:ok, sup}
end
end
end
I know generally we pass a list of children to a supervisor, but then we don’t have a way to reference the other process’s pids. Are there drawbacks to this approach or a better way of doing this?
Trending in Questions
Other Trending Topics
Latest on Elixir Forum
Sponsor Spotlight
Bringing learning, work and compliance together since 1987.
Categories:
Sub Categories:
Forums
Popular Tags
- #ecto
- #liveview
- #troubleshooting
- #learning-elixir
- #deployment
- #library
- #erlang
- #testing
- #genserver
- #mix
- #absinthe
- #remote-other
- #otp
- #plug
- #how-to-question
- #macros
- #postgres
- #channels
- #elixirconf
- #exunit
- #discussion
- #code-sync
- #javascript
- #podcasts
- #onsite
- #dialyzer
- #docker
- #authentication
- #umbrella
- #full-time-contract
- #podcasts-by-brainlid
- #ecto-query
- #elixir-ls
- #phoenix_html
- #iex
- #blog-post
- #graphql
- #genstage
- #ai
- #elixirconf-us
- #websockets
- #supervisor
- #advent-of-code
- #distillery
- #processes
- #api
- #forms
- #metaprogramming
- #hex
- #performance
Our Sponsors
Build Elixir applications with speed and confidence.
Supporting innovation across the BEAM ecosystem.
We build reliable cloud platforms for business-critical systems.
Catch errors, track performance, monitor hosts and more.
Error tracking, logs, uptime, and Just Enough APM. Built for developers.
Practical resources that improve the lives of professional developers.
Producing high quality Elixir screencasts since 2017.
The deployment platform built for Elixir: PaaS ease, VPS control.
The team behind Membrane, Popcorn, LiveDebugger. Available for hire.
Courses that'll move you from confusion to "Aha, now I get it!"
First 4 of 4 Posts
mpope
An alternative (not necessarily better) is using the
WebServeras the interface for theSupervisor. What I mean is first you’d start the Supervisor, then start the WebServer as a child and pass the Supervisor’s pid to the WebServer. Then implement aWebServer.start_childfunction. TheWebServercould then call out to the Supervisor to start the child process, store the child’s pid in it’s state, as well as link them, etc.I don’t see a ton of benefit compared to what you have unless you need to start different child types in different scenarios, like if you have different implementations for the
StateServerdepending on configuration values or you have a complex chain of processes.EDIT: This would move the WebServer as the first child, and might not work well with the
rest_for_onerequirement (I missed that on first read).kokolegorille
I would do something similar, but not with WebServer, with an additional GenServer used to start others processes, and keeping pid in his state.
I guess I am a little bit influenced by how poolboy does this.
lud
Why not name registration?
rcavanaugh
Maybe overkill but what about
Registry? Start a registry namedMyApp.Supervisor.Registryin yourMyApp.Supervisorbefore anything else and then each supervised process can register itself ininitand dispatch calls via the registry.