script/delayed_job background process immediately dies silently

[gchiaren]

This is in development. It seems delayed job immediately dies as soon as I start it, and fails silently without writing anything to the log (that's the most frustrating part bc I'm completely stuck). Below are more details:

I start DJ in the console using script/delayed_job start. The PID file is created, but when I do a script/delayed_job status it says:

pid-file for killed process 47552 found (/Users/me/Rails/projects/tmp/pids/delayed_job.pid), deleting.
delayed_job: no instances running

Sure enough, the jobs in the database table just sit there, with the first one up getting its locked_at timestamp set.

My DJ log file only logs the attempted start, but nothing about the stop. It just says:

Starting job worker
Job job_name (id=1537) RUNNING

I followed the instructions in Common Problems to no avail. If I manually run rake jobs:work the jobs are indeed processed and disappear from the db.

I thought it might be a serialization problem, but I can do this successfully from the console:

j = Delayed::Job.last
YAML.load(j.handler)

Versions of everything:

• Rails (3.2.22)
• Ruby (2.2.10)
• delayed_job (4.1.5)
• delayed_job_active_record (4.1.3)
• daemons (1.3.1)

My DJ config looks like this:

Delayed::Worker.max_attempts = 5
Delayed::Worker.max_run_time = 5.minutes
Delayed::Worker.logger = Logger.new(File.join(Rails.root, 'log', 'delayed_job.log'))
Delayed::Worker.destroy_failed_jobs = false

[dignoe]

@gchiaren Did you ever resolve this? The same thing is happening to me.

[gchiaren]

@dignoe Nope! But I had to handle it somehow so I'm using bluepill to bring DJ back up when it goes down. Lmk if you need help with the bluepill config.

[dignoe]

Ok, thanks. I've noticed that certain jobs cause DJ to fail, and sometimes DJ will process multiple jobs and just keep on going. All the jobs pass our tests and work fine in production. I'm slowly migrating everything over to ActiveJob and then I'm probably going to switch to Resque anyway, so I'll just live with this until I get everything moved over.

[dignoe]

FYI - This issue might be related: #935

I've switched to using foreman in dev mode, so instead of running bin/delayed_job start I'm now just using bin/rails jobs:work in my Procfile. This seems to have solved the issue for me.

[johnnyshields]

This is expected behavior of the daemons gem. You can see here the first thing it does is to fork the parent and exit I forget exactly why.

If you're ok to run everything in the foreground (e.g. for a container) try my "forking" mode PR: #1138. This is similar to the Rake task except it can now run multiple workers.

[albus522]

Closing stale