FPX

Standalone service for collecting content from multiple source into single file. Typical usecase is downloading multiple files as archive using single link. Internally FPX fetches content from the specified set of URLs and streams zip-compressed stream to the end users.

AWS EC2 Deploy

1. Install Python 3.8 or newer:

   # install build dependencies
   sudo yum install -y openssl-devel readline-devel zlib-devel bzip2-devel libffi-devel
   
    # install `pyenv`
    git clone https://github.com/pyenv/pyenv.git ~/.pyenv
    # this require `chmod +x $HOME` if you are going to use different user for running services with installed python executable
    echo 'export PYENV_ROOT="$HOME/.pyenv"' >> ~/.bash_profile
    echo 'export PATH="$PYENV_ROOT/bin:$PATH"' >> ~/.bash_profile
    echo -e 'if command -v pyenv 1>/dev/null 2>&1; then\n  eval "$(pyenv init -)"\nfi' >> ~/.bash_profile
   
    # install python
    pyenv install 3.8.2
   

2. Create venv for FPX and install it:

   pyenv shell 3.8.2
   cd /usr/lib/ckan
   python -m venv fpx
   cd fpx
   source bin/activate
   pip install 'fpx~=0.4.0'
   

3. Create config file. It can be created anywhere, as long as it accessible by FPX service:

    echo '
    PORT = 12321
    # DB is not used much, so SQLite can be used as long as you are going to use single instance of FPX service. If you planning to use multiple instances + load balancer, consider using PostgreSQL
    DB_URL = "sqlite:////var/lib/ckan/default/fpx.db"
    # Any other options passed directly to the SQLAlchemy engine constructor(https://docs.sqlalchemy.org/en/13/core/engines.html#sqlalchemy.create_engine)
    DB_EXTRAS = {
    # "pool_size": 10,
    # "max_overflow": 20,
    }
    # Maximum number of simultaneous downloads. In production, value between 10 and 100 should be used, depending on server's bandwidth. Higher value won't affect server perfomance, but will make downloads slower due to bandwidth limitations.
    SIMULTANEOURS_DOWNLOADS_LIMIT = 20
    ' > /etc/ckan/default/fpx.py
   

4. Initialize database and create access token for client. It can be stored later inside CKAN ini file as fpx.client.secret:

    export FPX_CONFIG=/etc/ckan/default/fpx.py
    fpx db up
    fpx client add my-first-fpx-client  # use any name, that match `[\w_-]`
   

   Make sure, db is accessible and writable by FPX service. This manual suggests using apache user when configuring supervisor's process, so following command required:

   chown apache:apache /var/lib/ckan/default/fpx.db
   

5. Test service:

    FPX_CONFIG=/etc/ckan/default/fpx.py fpx server run
    # or, if you want to explicitely use python interpreter
    FPX_CONFIG=/etc/ckan/default/fpx.py python -m fpx
   

6. Configure system.d/supervisor/etc. unit for fpx. Make sure, that fpx server run command, that spins up the service is executed using python>=3.6 (pyenv shell 3.8.2). And, if SQLite is used, fpx process has write access to db file:

    [program:fpx-worker]
   
    ; Use the full paths to the virtualenv and your configuration file here.
    command=/usr/lib/ckan/fpx/bin/python -m fpx
   
    environment=FPX_CONFIG=/etc/ckan/default/fpx.py
   
    ; User the worker runs as.
    user=apache
   
    ; Start just a single worker. Increase this number if you have many or
    ; particularly long running background jobs.
    numprocs=1
    process_name=%(program_name)s-%(process_num)02d
   
    ; Log files.
    stdout_logfile=/var/log/fpx-worker.log
    stderr_logfile=/var/log/fpx-worker.log
   
    ; Make sure that the worker is started on system start and automatically
    ; restarted if it crashes unexpectedly.
    autostart=true
    autorestart=true
   
    ; Number of seconds the process has to run before it is considered to have
    ; started successfully.
    startsecs=10
   
    ; Need to wait for currently executing tasks to finish at shutdown.
    ; Increase this if you have very long running tasks.
    stopwaitsecs = 600
   

7. FPX service must be available via public url(and CKAN ini file requires this URL under fpx.service.url config option). As written in []documentation](https://sanic.readthedocs.io/en/latest/sanic/deploying.html#deploying), no additional layers required. But if you decide to use it with Nginx, the following link may be useful. Note, FPX is using websockets(if it can somehow affect configuration).

   Example of Nginx section for FPX:

   location /fpx/ {
      proxy_pass http://127.0.0.1:12321/;
      proxy_set_header X-Forwarded-For $remote_addr;
      proxy_set_header Host $host;
      proxy_http_version 1.1;
      proxy_request_buffering off;
      proxy_buffering off;
      proxy_set_header connection "upgrade";
      proxy_set_header upgrade $http_upgrade;
      # In emergency comment out line to force caching
      # proxy_ignore_headers X-Accel-Expires Expires Cache-Control;
   }
   

   Example of apache configuration:

   # mod_proxy
   # mod_proxy_http
   # mod_proxy_wstunnel
   # mod_rewrite
   # ...
   ProxyPass /fpx/ http://0.0.0.0:8000/
   ProxyPassReverse /fpx/ http://0.0.0.0:8000/
   
   RewriteEngine on
   RewriteCond %{HTTP:UPGRADE} ^WebSocket$ [NC]
   RewriteCond %{HTTP:CONNECTION} ^Upgrade$ [NC]
   RewriteRule /fpx/(.*) ws://0.0.0.0:8000/$1 [P]