Skip to content

Latest commit

 

History

History
144 lines (101 loc) · 3.86 KB

deployment-production.md

File metadata and controls

144 lines (101 loc) · 3.86 KB
Raw

Deploying browserperfdash for production

This document contains more specific instructions about how to setup the DB with PostgreSQL for production

Install dependendencies and fetch the code

  1. Install dependendencies
$ sudo apt-get install python3 python3-pip python3-dev virtualenvwrapper
$ sudo apt-get install postgresql python-psycopg2
  1. Get the files You can clone it directly from https://github.com/Igalia/browserperfdash
mkdir /srv
cd /srv
git clone https://github.com/Igalia/browserperfdash

Setup the python virtual environment and install pip requirements

  1. Setup python virtual environment
source /usr/share/virtualenvwrapper/virtualenvwrapper.sh
export WORKON_HOME=/srv/browserperfdash/venv
mkdir ${WORKON_HOME}
  1. Create a virtual environment dashboard for our project
mkvirtualenv -p /usr/bin/python3 dashboard
workon dashboard
  1. Install pip requirements on the virtual environment
cd /srv/browserperfdash
pip install -r requirements.txt

Setup the database with PostgreSQL and the app config

See postgres-setup.md.

Deploy HTTP server and running the app

Since the app runs on a python virtual environment we have stored inside the subfolder venv of the rootdir, each time we want to run the app or nanage it via python manage.py we have to enter the virtual env via:

cd /srv/browserperfdash
source venv/dashboard/bin/activate

To run the app via an apache webserver we can use the provided [wsgy.py](../browserperfdash/wsgi.py] module.

First ensure Apache and the wsgi module is installed

$ sudo apt-get install apache2 libapache2-mod-wsgi-py3


Then configure a virtual site, this is an example of how it can be done:

ServerAdmin admin@igalia.com
Alias /favicon.ico /srv/browserperfdash/static/favicon.ico
Alias /static/ /srv/browserperfdash/static/
<Directory /srv/browserperfdash/static>
    Require all granted
</Directory>

WSGIDaemonProcess browserperfdash python-home=/srv/browserperfdash/venv/dashboard python-path=/srv/browserperfdash processes=50 threads=1

WSGIProcessGroup browserperfdash
WSGIApplicationGroup %{GLOBAL}

WSGIScriptAlias / /srv/browserperfdash/browserperfdash/wsgi.py
AllowEncodedSlashes NoDecode

<Directory /srv/browserperfdash/browserperfdash>
    <Files wsgi.py>
        Require all granted
    </Files>
</Directory>

ErrorLog ${APACHE_LOG_DIR}/browserperfdash-error.log
CustomLog ${APACHE_LOG_DIR}/browserperfdash-access.log combined
```

If you need more info or you want to use other webserver like Nginx, please check the Django deployment guide: https://docs.djangoproject.com/en/1.10/howto/deployment/

Set the allowed hosts list

Edit browserperfdash/local_settings.py and set the list of IPs and/or hostnames where the service will be hosted (this should match the hostname set by the webserver)

Example:

# [...]
# SECURITY WARNING: don't run with debug turned on in production!
DEBUG = False

ALLOWED_HOSTS = ["browserperfdash.igalia.com", "browserperfdash.intranet.igalia.com", "localhost", "127.0.0.1", "10.10.1.137"]
CSRF_COOKIE_DOMAIN = '.igalia.com'

# Database
# https://docs.djangoproject.com/en/1.10/ref/settings/#databases
# [...]

Initial configuration of the benchmarks and metric units

You can import this partial dump that will fill the metric units and benchmark plans.

$ su - postgres
$ cat docs/initial_data_postgresql.dump | psql browserperfdash
COPY 8
COPY 13

After that you only need to fill the GPU/CPU/platform/bot/browser entries and start deploying bots to send data. For that you can use the script browserperfdash-benchmark that you can find on the WebKit repository (or, alternatively at https://github.com/Igalia/browserperfrunner)