Why won't Huginn agents act on Webfaction?

Question

I could spool up a VPS or deploy to a PaaS, but I'm determined to get Huginn working on my Webfaction account because I'm pretty sure it's possible and I just don't get rails deployments.

After hours of hacking away, I ended up with a running app whose Agents do nothing. I can't manage to run foreman start, the magical command that should solve the problem, so I'm not sure what else to do.

How I got here

I'm used to PHP and these deployments are foreign to me, so I'm certain the problem's between the chair and the keyboard...

1. Created a new Webfaction Application called huginn with Rails 4.1.8 (nginx 1.6.2/Passenger 4.0.53/Ruby 2.1.2).
2. Cloned huginn into its own folder next to the hello_world example app that Webfaction initializes. (~/webapps/huginn/huginn/)
3. Edited ~/webapps/nginx/conf/nginx.conf (which came from Webfaction's init script) to point to my cloned huginn folder instead of hello_world. Changed GEM_HOME to /home/foo/webapps/huginn/huginn/gems instead of /home/foo/webapps/huginn/gems, root to /home/foo/webapps/huginn/huginn/public, and rails_env to production.
4. Edited ~/webapps/huginn/bin/start to adjust paths (see below).
5. Copied and configured Huginn's .env properly, pointing to a fresh MySQL database with valid credentials. Added a randomly-generated APP_SECRET_TOKEN, set host name and email configuration.
6. Added app bin, path, and gems to PATH using Webfaction's instructions.
7. From ~/webapps/huginn/huginn, ran gem2.1 install bundle, ruby2.1 bin/bundle, ruby2.1 bin/bundle exec rake db:migrate, ruby2.1 bin/bundle exec rake db:seed and all goes well; dependencies installed to huginn's directories, database populated, etc.
8. Ran ruby2.1 bin/bundle exec foreman start and failed:

16:50:03 web.1 | started with pid 7115
16:50:03 jobs.1 | started with pid 7116
16:50:06 jobs.1 | /home/foo/webapps/huginn/huginn/gems/gems/spring-1.1.3/lib/spring/server.rb:22:in `initialize': Permission denied @ rb_sysopen - /tmp/spring/e89e6bd5b6863aeed0fbb5c861b1fc2d.pid (Errno::EACCES)

8. Go back and run ~/webapps/huginn/bin/stop, ~/webapps/huginn/bin/start, and the app miraculously starts working in my browser. Except for the Agents, which do nothing on schedule or an attempt to manually run them.

I suspect that I'm missing something when it comes to environment variables and ruby versions. Various ruby versions are available, but ruby --version outputs ruby 1.8.7 (2013-06-27 patchlevel 374) [x86_64-linux]. Even I figured out that there are symlinks to different versions, so ruby2.1 --version gets me ruby 2.1.2p95 (2014-05-08 revision 45877) [x86_64-linux]. My guess is that the app's using Ruby 1.8.7, and I ran with this trying to install rvm to select a different ruby version—I didn't make it through rvm install X.X.X because I'm not a sudoer and libyaml-devel couldn't be installed.

I was briefly excited to learn that I could run ruby2.1 bin/rails s --port=22766 --environment=production from the deployed app directory and have the app once again running in the browser—but with the exact same issue of non-responsive agents.

It seems like I'm really close, but that I'm missing something that's consistently leaving Huginn's agents crippled. What am I missing?

---

nginx.conf

env               GEM_HOME=/home/foo/webapps/huginn/huginn/gems;
worker_processes  1;

events {
    worker_connections  1024;
}

http {
    access_log  /home/foo/logs/user/access_huginn.log  combined;
    error_log   /home/foo/logs/user/error_huginn.log   crit;

    include         mime.types;
    passenger_root  /home/foo/webapps/huginn/gems/gems/passenger-4.0.53;
    passenger_ruby  /home/foo/webapps/huginn/bin/ruby;
    sendfile        on;

    passenger_max_instances_per_app  1;
    rails_spawn_method               conservative;
    passenger_max_pool_size 2;

    server {
        listen             22766;
        passenger_enabled  on;
        root               /home/foo/webapps/huginn/huginn/public;
        server_name        localhost;
        rails_env          production;
    }
}

~/webapps/huginn/bin/start

#!/bin/bash

PATH=/usr/local/bin:/bin:/usr/bin:/usr/local/sbin:/usr/sbin:/sbin:/home/foo/bin RUBYLIB=/home/foo/webapps/huginn/huginn/lib:$RUBYLIB TMPDIR=/home/foo/webapps/huginn/tmp PASSENGER_TMPDIR=/home/foo/webapps/huginn/tmp GEM_HOME=/home/foo/webapps/huginn/huginn/gems /home/foo/webapps/huginn/nginx/sbin/nginx -p /home/foo/webapps/huginn/nginx/

Answer 1

Thanks to help from John at Webfaction, it boiled down to setting a custom temp directory that could be written to with my account...

mkdir -p $HOME/tmp
export TEMP=$HOME/tmp

Sure enough, running ruby2.1 bin/bundle exec foreman start worked just fine and the Agents sprang to life.

Then I promptly hit the wall of my 512MB memory limit, so it's on to other issues.