birdhousebuilder.recipe.pywps is a Buildout recipe to install and configure PyWPS with Anaconda. PyWPS is a Python implementation of a Web Processing Service (WPS). PyWPS will be deployed as a Supervisor service and is available behind a Nginx web server. This recipe is used by the Birdhouse project.

The recipe requires that Anaconda is already installed. You can use the buildout option anaconda-home to set the prefix for the anaconda installation. Otherwise the environment variable CONDA_PREFIX (variable is set when activating a conda environment) is used as conda prefix.

It installs the pywps package from a conda channel in a conda environment defined by CONDA_PREFIX. The location of the intallation is given by the prefix buildout option. It setups a PyWPS output folder in ${prefix}/var/lib/pywps. It deploys a Supervisor configuration for PyWPS in ${prefix}/etc/supervisor/conf.d/pywps.conf. Supervisor can be started with ${prefix}/etc/init.d/supervisor start.

The recipe will install the nginx package from a conda channel and deploy a Nginx site configuration for PyWPS. The configuration will be deployed in ${prefix}/etc/nginx/conf.d/pywps.conf. Nginx will be started by supervisor.

By default PyWPS will be available on http://localhost:8091/wps?service=WPS&version=1.0.0&request=GetCapabilities.

The recipe depends on:

• birdhousebuilder.recipe.conda,
• birdhousebuilder.recipe.supervisor,
• birdhousebuilder.recipe.nginx and
• zc.recipe.deployment.

The PyWPS options which are configured by buildout are explained in the PyWPS documentation.

The recipe supports the following buildout options:

anaconda-home

Buildout option pointing to the root folder of the Anaconda installation. Default: $HOME/anaconda.

Buildout options for pywps:

prefix

Deployment option to set the prefix of the installation folder. Default: /

user

Deployment option to set the run user.

etc-user

Deployment option to set the user of the /etc directory. Default: root

name

The name of your WPS project (used for config names and folder path).

hostname

The hostname of the PyWPS service (nginx). Default: localhost

http-port

The http port of the PyWPS service (nginx). Default: 8091

https-port

The https port of the PyWPS service (nginx). Default: 28091

http-output-port, output-port

The http port of the PyWPS output file service (nginx). Default: 8090

https-output-port

The https port of the PyWPS output file service (nginx). Default: 28090

ssl-verify-client

Nginx option to verify SSL client certificates. Possible values: off (default), on, optional. `Nginx ssl_verify_client option`_

ssl-client-certificate

Nginx option with the name of the bundle of CA certificates for the client. Default: esgf-ca-bundle.crt. `Nginx ssl_client_certificate`_

ssl-client-certificate-url

Optional URL to download a bundle of CA certificates for ssl-client-certificate. Default: https://github.com/ESGF/esgf-dist/raw/master/installer/certs/esgf-ca-bundle.crt

workers

The number of gunicorn workers for handling requests. Default: 1

application

PyWPS WSGI Application. Default: ${name}:application.

title

Title used for your WPS service.

abstract

Description of your WPS service.

loglevel

Logging level for PyWPS. Default: WARN

logformat

Logging string format for PyWPS. Default: %(asctime)s] [%(levelname)s] line=%(lineno)s module=%(module)s %(message)s

database

Database where the logs about requests/responses is to be stored. Allowed values are memory or sqlite. Default: sqlite.

parallelprocesses

Maximum number of parallel running processes. The effective number of parallel running processes is limited by the number of cores in the hosting machine. Default: 2

maxprocesses

Maximum number of processes which are accepted in the queue. Default: 30

maxrequestsize

Maximal request size accepted in WPS process. Default: 30mb

allowedinputpaths

List of server paths which are allowed to be accessed by file URL complex input parameters.

mode

Processing mode to run jobs. Allowed values are default (multiprocessing) and slurm. Default: default

The following example buildout.cfg installs PyWPS with Anaconda:

[buildout]
parts = pywps

[pywps]
recipe = birdhousebuilder.recipe.pywps
name = myproject
prefix = /
user = www-data
hostname = localhost
http-port = 8091
https-port = 28091

# pywps options
processes-import = myproject.processes
processes-path = ${buildout:directory}/myproject/processes
title = MyProject ...
abstract = MyProject does ...

After installing with Buildout start the PyWPS service with:

$ cd ${prefix}
$ etc/init.d/supervisord start  # start|stop|restart
$ etc/init.d/nginx start        # start|stop|restart
$ bin/supervisorctl status      # check that pycsw is running
$ less var/log/pywps/myproject.log  # check log file

Open your browser with the following URL:

http://localhost:8091/wps?service=WPS&version=1.0.0&request=GetCapabilities