About

mongrel-esi is meant to make caching easier by distributing the cache logic.

The idea is to represent each part of the page as a unique URL.

The whole page may not be cachable, but most of it probably is. The cache server works by scanning the requested document for specific <esi:* tags before servering the response. MongrelESI currently, only supports a few basic instructions

Tags

esi:include:

src     - request path, response replaces esi:include tag
alt     - if src is unavailable, use alt
timeout - how long to wait on the response
max-age - how long to cache the entity
onerror - whether continue on errors or trigger an exception

esi:invalidate:

encloses invalidation instructions.
these instructions currently only support basic invalidation
e.g.
    <esi:invalidate output="no">
         <?xml version="1.0"?>
         <!DOCTYPE INVALIDATION SYSTEM "internal:///WCSinvalidation.dtd">
         <INVALIDATION VERSION="WCS-1.1">
           <OBJECT>
             <BASICSELECTOR URI="/path/to/invalidate"/>
             <ACTION REMOVALTTL="0"/>
             <INFO VALUE="invalidating fragment test 1"/>
           </OBJECT>
         </INVALIDATION>
    </esi:invalidate>

In the above example, only /path/to/invalidate will be invalidated, advanced regex style selectors are only support with ruby as the 
cache storage.

esi:try/esi:attempt/esi:except

if any tags within the attempt block raise an exception and the onerror attribute is not equal to continue then
the server will fall back to the markup within the except block

Proxy Config

mongrel-esi is a proxy server. To configure where requests should be proxied modify the config/routes.yml file.

Here’s an example:

ESI::Config.define(listeners) do|config|

# define request path routing rules
config.routes do|s|
  s.match( /^\/(content|samples|extras).*/ ) do|r|
    r.servers = ['127.0.0.1:4000']
  end
  s.default do|r|
    r.servers = ['127.0.0.1:3000']
  end
end

end

This sample configuration will route all urls starting with /content, /samples, or /extras to the servers running at 127.0.0.1 on port 4000. Everything else that matches the .* will be routed to the server running on port 3000. Optionally the caching duration can be specificied explicity for each host

s.default do|r|

r.servers = ['127.0.0.1:3000']
r.cache_ttl = 300

end

This example will cache all requests for 300 seconds, but normally you would want to set the ttl on each individual fragment or include.

A typical include would look like the following:

<esi:include src=“/content/1” max-age=“600+600”/>

Requesting /home, might respond with the following:

<html> <head><title>Your Page</title> </head> <body> <div class=“header”><esi:include src=“/content/1 max-age=”600+600“/></div> <div>Some content</div> <div>Some more content</div> </body> </html>

If the uri is not already cached, the cache server will request /content/1, which will respond with:

<div>hello User42</div>

And finally, mongrel-esi will respond to the original client request with the combined documents: <html> <head><title>Your Page</title> </head> <body> <div class=“header”><div>hello User42</div></div> <div>Some content</div> <div>Some more content</div> </body> </html>

Integrating with your applications

mongrel-esi was built to support integrating lots of applications together.

Primarily, you’ll want to integrate by having one base application and many tiny applications that surface within your primary application. In the rails sense think components. In a bigger sense think multiple application technologies all integrating into the same application. Imagine a java application returning up to the minute chat conversations in one part of the page, while rails delivers a reliable listing of current forum discussion, and maybe a python presense indicator for forum posts…

One thing to keep in mind is mongrel-esi is a proxy server. This means for each request it needs to know where to route the requests. To support this config.rb can express any regex to match a url and determine where a request should be forwarded as well as defining a default route.

It is also sometimes convenient to allow multiple applications to store or house static content. For this to work mongrel-esi needs to know in what directories to search for static files. This can be done by using the MultiDirHandler (see rev-config.rb)

Supported Features of ESI

From [www.w3.org/TR/esi-lang esi-lang]

mongrel-esi supports basic include, exception handling and invalidation. It does not include support for Variable or Conditional processing. It does have suppport for COOKIE variables. (e.g. $(HTTP_COOKIEname) will be replaced with the value of the cookie, name)

+ Inclusion - ESI can compose pages by assembling included content, which is fetched from the network. This allows each such fragment to have its own metadata (e.g., cacheability and handling information) seperately associated.

  • Variable support - ESI 1.0 supports the use of variables based on HTTP request attributes in a manner

	reminiscent of the Common Gateway Interface. These variables can be used by ESI statements or written directly into the processed markup.
  • Conditional processing - ESI allows conditional logic with Boolean comparisons to be used to influence how a template is processed.

+ Exception and error handling - ESI provides for specification of alternate and default resources in a number of situations.

HTTP Readings

Adding New Features and Bug Fixing

Before adding a new feature or bug fix, it’s important to add a test to verify the behavior.

Also, help out by keeping the Changelog updated with new features and major bug fixes

Future ideas of improving integration

NOTE: this is a work in progress, below is how I’d like to integrate with rails in the future

cd to/path/to/your/primary/rails/app

mongrel_esi add . –default mongrel_esi add path/to/your/rails/secondary_app2 mongrel_esi add path/to/your/rails/secondary_app3 mongrel_esi add path/to/your/rails/secondary_app4

mongrel_esi start

- starts up mongrel in each of the applications

Credits ==

Support:

Zed A. Shaw        -- For writing [http://mongrel.rubyforge.org/ mongrel]
Adrian D. Thurston -- For writing [http://www.cs.queensu.ca/~thurston/ragel/ ragel]

The [http://www.ruby-lang.org/ ruby] community, for all their insights and help

Author:

Todd Fisher

Co-Authors:

Aaron Batalion -- Design ideas and integrating with Rails, fragment_fu. Ragel advocate
Richard Kilmer -- Thread saftey dispatcher, production ready
Adam Bair      -- Writing the initial testing framework and logger.
Jeff Damick    -- Writing the initial configuration.

And of course thanks to everyone in #ruby-lang for putting up with my questions MenTaLguY for the great help in making good use of ruby