Closed diegomichel closed 8 months ago
I've been hitting a similar issue:
my-app-1 | 05:50:27 web.1 | Rendered text template (Duration: 0.0ms | Allocations: 24)
my-app-1 | 05:50:27 web.1 | Completed 200 OK in 452ms (Views: 4.8ms | ActiveRecord: 2.8ms | Allocations: 28169)
my-app-1 | 05:50:27 web.1 |
my-app-1 | 05:50:27 web.1 |
my-app-1 | 05:50:27 web.1 | Started GET "/assets/fonts/Graphik-Medium-aeec9e6cbb82ed00c2bf827ef56ac0099bb39112.woff2" for 172.18.0.1 at 2023-08-28 05:50:27 +0000
my-app-1 | 05:50:27 | exited with code 0
my-app-1 | 05:50:27 system | sending SIGTERM to all processes
my-app-1 | 05:50:27 web.1 | [8] === puma shutdown: 2023-08-28 05:50:27 +0000 ===
my-app-1 | 05:50:27 web.1 | [8] - Goodbye!
my-app-1 | 05:50:27 web.1 | [8] - Gracefully shutting down workers...
my-app-1 | 05:50:27 js.1 | exited with code 1
my-app-1 | 05:50:27 web.1 | Not notifying due to an invalid api_key
my-app-1 | 05:50:27 css.1 | exited with code 1
my-app-1 | 05:50:27 web.1 | terminated by SIGTERM
my-app-1 exited with code 0
The exited with code 0
doesn't appear to be any of the sub-processes (as it has no name?) - but it seems to send the SIGTERM killer to all the running processes.
I'm running inside a docker container (with tty: true set) and have removed the ruby debug env var, but it still crashes.
I'm still using webpack, and note that it seems to take 30-40 seconds to build the assets every time I start up (which isn't ideal) - I wonder if that long build time causes something to be unresponsive and causes foreman to be marked for death? I have noted that if "left alone" after starting up, the process seems to no longer crash after 1-2 requests, whereas queueing up 2-3 requests whilst webpack is compiling seems to correlate with the server serving up one page, and then dying.
Nothing crashes when I run all my Procfile tasks in separate containers
This is happening to me too... macOS Sonoma 14.1, Ruby 3.2.2 and Rails 7.1.1. No docker.
Like the OP, it seems to be caused by env RUBY_DEBUG_OPEN=true
in Procfile.dev
. If I remove that environment variable it starts up fine.
Reproducible on any new app that uses Foreman. For example:
rails new debug_test --javascript esbuild
Figured it out @diegomichel. It's because we have long computer usernames. Ruby debug generates a socket name that is too long. Yours is 24 characters (diegomichelrubioramirez) and mine is 23 (matthewlindfieldseager). Suggest you close this issue and plus one the issue there.
@phantomwhale, your issue looks different. Ours was crashing at Rails startup, yours is crashing during the app lifecycle.
On Mac when using the bin/dev script the web server crashes, here are the logs. Please excuse me if this is not the correct gem to report this :).
Looks like env command on
Procfile.dev
might be causing the issue, probably an env var?changing
env RUBY_DEBUG_OPEN=true bin/rails server
to justenv bin/rails server
fixes the issue 🤔 .