There are many reasons why you may want to serve web content directly from Stingray Traffic Manager - simplification, performance, ease of administration and, perhaps most importantly, to host a 'Sorry Page' if your entire web infrastructure has failed and Stingray is all that is left.
The article Using Stingray Traffic Manager as a Webserver describes the rationale in more detail and presents a simple TrafficScript-based webserver. However, we can do a lot more with a more complete programming language - mime types, index pages, more control over the location of the document root are all simple to implement with Python.
Get started with PyRunner.jar
Start with the procedure describe in the article PyRunner.jar: Running Python code in Stingray Traffic Manager. The PyRunner extension lets you run Python code in Stingray, using the local JVM and the Jython implementation.
Note that this example does not work reliably with versions of Jython prior to 2.7 beta1 - I hit problems when a library attempted to import the Jython errno module (possibly related to https://github.com/int3/doppio/issues/177).
webserver.py
Once you've installed PyRunner (using an appropriate version of Jython), upload the following Python script to your Extra Files catalog. Make sure to call the script 'webserver.py', and edit the location of the docroot to an appropriate value:
from javax.servlet.http import HttpServlet
from urllib import url2pathname
from os import listdir
from os.path import normpath,isdir,isfile,getmtime,getsize
import mimetypes
import datetime
docroot = '/tmp'
def dirlist( uri, path, response ):
files = ''
for f in listdir( path ):
if (f[0] == '.') and (f[1] != '.'):
continue # hidden files
if isdir( path+'/'+f ):
size = '<DIR>'
f += '/' # Add trailing slash for directories
else:
size = '{:,d} bytes'.format( getsize( path+'/'+f ))
mtime = datetime.datetime.fromtimestamp( getmtime( path+'/'+f ))
files += '<a href="{f}">{f:<30}</a> {t:14} {s
17}\n'.format(
f=f, t=mtime, s=size )
html = '''''
<html><head><title>{uri}</title></head><body>
<h1>Directory listing for {uri}</h1>
<pre>{files}<a href="../">Up</a></pre>
</body></html>''' .format(uri=uri, files=files )
response.setContentType( 'text/html' )
toClient = response.getWriter()
toClient.write( html )
class webserver(HttpServlet):
def doGet(self, request, response):
uri = request.getRequestURI()
print "Processing "+uri
path = normpath( docroot + url2pathname( uri ) )
# Never access files outside the docroot
if path.startswith( docroot ) == False:
path = docroot
uri = '/'
if isdir( path ):
if uri.endswith( '/' ) == False:
response.sendRedirect( uri+'/' )
if isfile( path + '/index.html' ):
response.sendRedirect( uri + 'index.html' )
else:
dirlist( uri, path, response )
return
try:
c = open( path, 'r' ).read()
except Exception, e:
response.sendError( 400, "Could not read "+path+": "+str( e ) )
return
mtype = mimetypes.guess_type( path )[0]
if mtype == None:
mtype = 'application/octet-stream'
response.setContentType( mtype )
toClient = response.getWriter()
toClient.write( c )
If you want to edit the rule and try some changes, they you'll find the publish.py script in Deploying Python code to Stingray Traffic Manager useful, and you should follow the Stingray Event log (tail -f /opt/zeus/zxtm/log/errors) to catch any problems.
Performance Optimization
The article Using Stingray Traffic Manager as a Webserver describes how to front the web server virtual server with a second virtual server that caches responses and responds directly to web requests.
It also presents an alternative webserver implementation written in Java - take your pick!
