RabbIT proxy for a faster web RabbIT logo
SourceForge Logo
Getting started
w3c on HTTP

New release
rabbit 4.0

Current version: 2.0.42
Java/5+ only
jdk/1.4 version: 2.0.37d

jdk/1.1 version: 2.0.26

dnsjava source dnsjava-1.6.2.tar.gz

Mailing lists

Robert Olofsson


Introduction: RabbIT is a proxy for HTTP, it is intended to be HTTP/1.1 compliant. RabbIT is also a package of tools useful for the web.

RabbIT is intended to speed up surfing over slow links by removing unnecessary parts (like background images) while still showing the page mostly like it is (don't ruin page-layout etc).

Since filtering the pages is a "heavy" process RabbIT caches the pages it fetches (the filtered pages that is) but tries to respect things like pragma no-cache and Cache-Control directives. RabbIT also accepts request for nonfiltered pages (by prepending "noproxy" to the adress like this: http://noproxy.www.altavista.digital.com/).

RabbIT is developed and tested under solaris and linux. but since the whole package is written in java the basic proxy should run on any plattform that supports java.

RabbIT converts images into low quality jpegs if possible and only send the smaller image (use noproxy to view the image unfiltered). Image conversion is a heavy process and is done with an external program, I would like to do this in java but have not found any good and fast package to do it. Anyway as standard it uses convert from ImageMagick, but it is configurable to use any command line program that can do the job. Of course RabbIT can be run with image conversion turned off but that is in a way the wrong thing to do since this is the big timesaver.

RabbIT compresses(with gzip) text and HTML pages before they are sent to the browser. This results in less data beeing sent and a nice speedup.

RabbIT is able to act as a proxy for SSL connections. Note however that RabbIT is unable to filter the encrypted streams.

RabbIT uses keepalive both to clients and servers whenever possible.

Whats needed: java 5.x see java.sun.com for the latest java for your plattform.

If you only have access to a java 1.4 engine you can test RabbIT/2.0.37b, There have been some changes and bug fixes since then. That version is probably good enought for most usage.

If you only have access to a java 1.1.x engine you can test RabbIT/2.0.26, that was the last version to support java/1.1.x. There have been a few bug fixes since that version, but that version is probably good enought for an evaluation.

Running: Before you run RabbIT you should configure it, The default configuration is probably ok for testing under linux. RabbIT is packed in a state that allows you to run the proxy directly after unpacking it. after unpacking the file RabbIT2.tar.gz cd into the directory named RabbIT and if your running unix type "jr". If your running windows (95/98/NT) run the proxy by issuing the commando "java rabbit.proxy.Proxy".

Some scripts to help you (probably only useful if your running unix).

  • jmake - (re)compile the proxy.
  • jr - run the proxy, save pid of proxy in a file named by `hostname`
  • jdoc - generate javadoc for the proxy.
  • jr [-h|-?|--help] [-v|--version] [-f inifile|--file inifile]
    -h: gives you some help
    -v: print out the version and exit.
    -f inifile: use inifile instead of conf/rabbit.conf

Configuration: RabbIT is very configurable, with the aim of beeing totaly configurable during runtime. Read the file conf/rabbit.conf (or conf/rabbit.conf.orig if you have change the default configuration) it is commented and is (currently) the best source of information for what you can do with RabbIT. Besides the default rabbit.conf there are two other example config files available, onlycaching.conf and onlycompressing.conf that can provide starting points for a good configuration or rabbit.

Runtime configuration: RabbIT now features reconfiguration during runtime. This means that if you want to rotate the logs, change the adreplacement image or add a advertising site it is very easy to do so. The way to do this is to use the http://<proxy>/FileSender/config.html there you will find some webpages that allows you to reconfigure RabbIT. Edit some changes and press [OK]. Every change you make is written back to disk so if you restart RabbIT it keeps all its setting.

Users and restricted pages: RabbIT provides an easy and very powerfull meta-page system. Therefor it is suggested that you restrict metapages to only authorized users. The current security is the simple "Basic" authentication from rfc2068. Note that this means that passwords are sent in cleartext (well actually uuencoded, but that is cleartext since any person can decode it). RabbIT has a userfile(normally the file users) that is on the form (one entry per line) "userid:password", where password is in cleartext (we dont have access to crypt under windows :-/). In the future maybe we can use Digest authentication to make it safer. Restriction of metapages is in two flavors: you should add a password to them and you should also restrict the set of ip:s that can connect (see access.conf for this). By default for now RabbIT allows access to everyone (that has ip-access rights) to the files found under http:///FileSender/public/, this is to allow NoAd.gif and the rabbit logo to be served from the proxy instead of from the net which used to be standard (both of these images are runtime configurable to either a local proxy file or some other URL).

MetaPages: Definition: a metapage is a service RabbIT provides (in this sense it acts as a CGI-webserver). To access metapages you use: http://<PROXYHOST>:<PROXYPORT>/Metapage (like for me http://magenta11:9666/)

http://<proxy>/Takes you to /FileSender/index.html.
http://<proxy>/StatusCurrent status of the proxy.
http://<proxy>/ConnectionsCurrent active (outbound) connections of the proxy.
http://<proxy>/CacheStatusCurrent status of the proxys cache.
http://<proxy>/ClearCacheClears the cachedir for the proxy.
http://<proxy>/KillTheStupidThingTake a guess!
http://<proxy>/FileSender/filenameMakes RabbIT act as a webproxy. Uses files in the htdocs directory.

It is very simple to write a new MetaHandler and RabbIT doesnt even need to be restarted to accept a new MetaPage. Simply make a class that implements rabbit.meta.MetaHandler and put it in the CLASSPATH for RabbIT. This is an extremly powerful tool, use it with care.

Cache: RabbIT uses a cache to save filtered pages. This is most useful if you are several people running against the same RabbIT proxy since most browsers have there own local cache. RabbIT has two values you can tweak for optimal size of the cache. The values are

maxsize: This value is the Maximum size the cache should take, that is the cache shouldnt be larger than this at any time (not true now, it can be bigger (due to halfcached file(working files), but shouldnt be much bigger). This value should be given in MB:s.

cachetime: This value specifies how many hours RabbIT should keep pages it knows nothing about (that has no expires-field in the httpheader).

Documentation and API: RabbIT should be easy to read and understand. Since RabbIT is written in java the api for RabbIT is accessible in javadoc format. see the directory RabbIT/htdocs/doc/.

Linkchecker: RabbIT has a simple linkchecker, rabbit.linkchecker.LinkVerifier, that can be used to check a sites URL:s it recurses down all links it can find located on the servers given in conf/linkchecker.conf (or user-specified).

Comments & critisism: Please provide some feedback of what you think should be done with RabbIT. That is features you miss, things that can be done better etc. A simple letter of the type "RabbIT is a great product" is also very welcome. Anyway send it to Robert Olofsson, robo@khelekore.org.