roboyeti
commented
3 years ago Same issue with Rails 6, all versions tried so far, after upgrading from 5. Possibly related to this issue: https://github.com/collectiveidea/delayed_job/issues/1099
Closed timdown closed 2 months ago
roboyeti
commented
3 years ago Same issue with Rails 6, all versions tried so far, after upgrading from 5. Possibly related to this issue: https://github.com/collectiveidea/delayed_job/issues/1099
joshfester
commented
3 years ago Same issue here. Happens in both development and production. Using the rake task (rake jobs:work) seems to work ok. But that's not ideal for production mode
soorajhealthplotter
commented
3 years ago I am also facing same issue. Sometimes the delayed job starts.but sometime it is throwing error in both production and development environment Ruby version: 2.6 Rails version: 6.1
marcoschicote
commented
3 years ago Same issue here
jminterwebs
commented
2 years ago Same issue
ajaybhayani741
commented
1 year ago Hey @timdown @roboyeti @marcoschicote @joshfester @jminterwebs
Did anyone get the solution to this issue? I m suffering from the same issue.
joshfester
commented
1 year ago @ajaybhayani741 I ended up moving on to Sidekiq and it's been great. If that's a possibility for you, I would definitely recommend it
jminterwebs
commented
1 year ago @ajaybhayani741, Yes unfortunately I complete forgot how. I was also dealing with isolating the worker into its own docker container as well.
Taking a shot in the dark on this but what ENV are you running in? I would check the config/delayed_job.rb and make sure you are allowing the worker to run in the ENV/
ducpt1012
commented
1 year ago Same issue when I use it in a transaction.
grasshopper1
commented
1 year ago I believe it is closely related to the configuration option config.file_watcher = ActiveSupport::EventedFileUpdateChecker
But I have no solution other than to disable that configuration option.
Since the file_watcher is not active in production environment I do not see the problem occur in that environment.
Ruby version: 2.7.4p191 Rails version: 6.1.7.2 DJ version: 4.1.11 listen version: 3.8.0 daemons version: 1.4.1 rb-inotify version: 0.10.1 linux system: debian 11 (bullseye)
allenjgreer
commented
1 year ago For anyone still encountering this issue, I was able to identify the root issue with inspiration from this stackoverflow answer
When running the delayed job process in daemonized mode bundle exec ruby bin/delayed_job start
From current script generator:
!/usr/bin/env ruby
require File.expand_path(File.join(File.dirname(FILE), '..', 'config', 'environment')) require 'delayed/command' Delayed::Command.new(ARGV).daemonize
I too a look into the daemonize gem and found that when starting up, it actually CLOSES ALL IO STREAMS!!. Modifying global state is always fun 👍 .
Within the script file, it first boots Rails via loading config/environment and especially with the config option eager_load = true means that if there is any code during the bootup process that opens up IOs (e.g. I was opening up cache connections via the Dalli gem) will be closed after the call to daemonize.
My current workaround is to ensure that Rails is loaded AFTER daemonize
#!/usr/bin/env ruby
require "delayed_job"
require "delayed_job_active_record"
require "delayed/command"
module LoadRailsAfterRunProc
def self.included base
after_fork_method = base.method(:after_fork)
base.define_singleton_method :after_fork do |*args|
# after Daemons.run_proc method
require File.expand_path(File.join(File.dirname(__FILE__), '..', 'config', 'environment'))
after_fork_method.call(*args)
end
end
end
Delayed::Worker.include(LoadRailsAfterRunProc)
Delayed::Command.new(ARGV).daemonizeThis explains also why running it via rake rake jobs:work does not have any issues as it does not daemonize the process.
factor4
commented
1 year ago @allenjgreer Your workaround solved it here on macOS and FreeBSD. Thank you very much!
toncid
commented
1 year ago Hello! Any ideas why there is no trace of the error if delayed_job is started with an identifier (./bin/delayed_job -i 1234 start) as indicated in #1172?
albus522
commented
2 months ago Closing stale
philayres
commented
3 weeks ago Hey @albus522 or other maintainers. How do we get the fix proposed by @allenjgreer incorporated in to the delayed_job startup script so it stop affecting new implementations, and update the README or some other upgrade notes for existing apps?
Is it supposed to go in lib/generators/delayed_job/templates/script? I've been fighting with this for ages. It definitely shouldn't be a closed issue in my opinion.
I'd be happy to contribute a PR, if it is likely to be accepted (in a reasonable timescale.)
Steps to reproduce:
gem 'daemons'andgem 'delayed_job_active_record'to Gemfilebundle installbundle exec rails generate delayed_job:active_recordthenbundle exec rake db:migratebundle exec ruby bin/delayed_job startThis does not happen with DJ 4.1.9 in Rails 6.0.3.4. I've tried it on macOS and the error does not appear. I've also tried changing the Ruby version to 3.0.0 and the error still appears.