django_ws_bench.py

from collections import Counter
import requests

# Top Django site's from djangoproject.com, stackoverflow.com, and quora.com.
# I've determined the two unknowns, rdio.com and washingtonpost.com, are nginx.
urls = [
    'http://djangoproject.com/',
    'http://disqus.com/',
    'http://instagram.com/',
    'http://addons.mozilla.org/',
    'http://pinterest.com/',
    'http://politifact.com/',
    'http://rdio.com/',
    'http://bitbucket.org/',
    'http://lanyrd.com/',
    'http://ljworld.com',
    'http://mahalo.com/',
    'http://theonion.com/',
    'http://nationalgeographic.com/',
    'http://zoosk.com/',
    'http://washingtonpost.com/',
    'http://guardian.co.uk/',
    'http://giantbomb.com/',
    'http://everyblock.com/',
    'http://comicvine.com/',
    'http://eldarion.com/',
    'http://flavors.me/',
    'http://goodsie.com/',
    'http://grove.io/',
    'http://prezi.com/',
    'http://npr.org/',
    'http://science.nasa.gov/',
    'http://djangopackages.com/',
    'http://readthedocs.org/',
    'http://michaelmoore.com/',
]

server_count = Counter()

for url in urls:
    try:
        r = requests.get(url)
    except requests.exceptions.ConnectionError, e:
        print 'Skipping %s, %s' % (url, e)
        continue

    if r.status_code != 200:
        continue

    server_name = r.headers.get('server', 'unknown').split('/', 1)[0].lower()
    server_count[server_name] += 1

    print '%s: %s' % (server_name, url)

print server_count