opentracker is a open and free bittorrent tracker project. It aims for minimal resource usage and is intended to run at your wlan router. Currently it is deployed as an open and free tracker instance. Read our free and open tracker blog and announce your torrents there (but do not hesitate to setup your own free trackers!).
opentracker is currently only available from CVS. A tar ball will be rolled soon. Use git clone git://erdgeist.org/opentracker or the legacy view cvs -d :pserver:email@example.com:/home/cvsroot co opentracker with an empty passwort to check it out. An *opentracker* gitweb and an *opentracker* cvsweb (deprecated) are available.
Until opentracker is declared official release ready, the way to install it is:
cvs -d :pserver:firstname.lastname@example.org:/cvs -z9 co libowfat cd libowfat make cd .. # LEGACY: cvs -d:pserver:email@example.com:/home/cvsroot co opentracker git clone git://erdgeist.org/opentracker cd opentracker make
That should leave you with an exectuable called opentracker and one debug version opentracker.debug.
Some variables in opentracker's Makefile control features and behaviour of opentracker. Here they are:
- -DWANT_V6 makes opentracker an IPv6-only tracker. More in the v6-section below.
- opentracker can deliver gzip compressed full scrapes. Enable this with -DWANT_COMPRESSION_GZIP option.
- Normally opentracker tracks any torrent announced to it. You can change that behaviour by enabling ONE of -DWANT_ACCESSLIST_BLACK or -DWANT_ACCESSLIST_WHITE. Note, that you have to provide a whitelist file in order to make opentracker do anything in the latter case. More in the closed mode section below.
- opentracker can run in a cluster. Enable this behaviour by enabling -DWANT_SYNC_LIVE. Note, that you have to configure your cluster before you can use opentracker when this option is on.
- Some statistics opentracker can provide are sensitive. You can restrict access to these statistics by enabling -DWANT_RESTRICT_STATS. See section statistics for more details.
- Fullscrapes is bittorrent's way to query a tracker for all tracked torrents. Since it's in the standard, it is enabled by default. Disable it by commenting out -DWANT_FULLSCRAPE.
- By default opentracker will only allow the connecting endpoint's IP address to be announced. Bittorrent standard allows clients to provide an IP address in its query string. You can make opentracker use this IP address by enabling -DWANT_IP_FROM_QUERY_STRING.
- Some experimental or older, deprecated features can be enabled by the -DWANT_LOG_NETWORKS, -DWANT_SYNC_SCRAPE or -DWANT_IP_FROM_PROXY switch.
- Currently there is some packages for some linux distributions and OpenBSD around, but some of them patch Makefile and default config to make opentracker closed by default. I explicitly don't endorse those packages and will not give support for problems stemming from these missconfigurations.
opentracker can be run by just typing ./opentracker. This will make opentracker bind to 0.0.0.0:6969 and happily serve all torrents presented to it. If ran as root, opentracker will immediately chroot to . (or any directory given with the -d option) and drop all priviliges after binding to whatever tcp or udp ports it is requested.
When options were few, opentracker used to accept all of them from command line. While this still is possible for most options, using them is quite unhandy: an example invocation would look like ./opentracker -i 188.8.131.52 -p 80 -P 80 -p 6969 -i 184.108.40.206 -p 80 -r http://www.mytorrentsite.com/ -d /usr/local/etc/opentracker -w mytorrents.list -A 127.0.0.1.
opentracker now uses a config file that you can provide with the -f switch.
opentracker's config file is very straight forward and a very well documented example config can be found in the file opentracker.conf.sample.
While personally I like my tracker to be open, I can see that there's people that want to control what torrents to track – or not to track. If you've compiled opentracker with one of the accesslist-options (see Build instructions above), you can control which torrents are tracked by providing a file that contains a list of human readable info_hashes. An example whitelist file would look like
To make opentracker reload it's white/blacklist, send a SIGHUP unix signal.
Given its very network centric approach, talking to opentracker via http comes very naturally. Besides the /announce and /scrape paths, there is a third path you can access the tracker by: /stats. This request takes parameters, for a quick overview just inquire /stats?mode=everything`.
Statistics have grown over time and are currently not very tidied up. Most modes were written to dump legacy-SNMP-style blocks that can easily be monitored by MRTG. These modes are: peer, conn, scrp, udp4, tcp4, busy, torr, fscr, completed, syncs. I'm not going to explain these here.
The statedump mode dumps non-recreatable states of the tracker so you can later reconstruct an opentracker session with the -l option. This is beta and wildly undocumented.
You can inquire opentracker's version (i.e. CVS versions of all its objects) using the version mode.
A torrent tracker basically is an http-Server that collects all clients ip addresses into pools sorted by one of the request strings parameters and answers all other clients that specified this exact same parameter a list of all other recent clients. All technologies to implement this are around for more than twenty years. Still most implementations suck performancewise.
Utilizing the highly scalable server framework from libowfat, opentracker can easily serve multiple thousands of requests on a standard plastic WLAN-router, limited only by your kernels capabilities ;)
One important design decision of opentracker was to not store any data persistently. This reduces wear&tear on hard disks and eliminates problems with corrupt databases.
A project like this one is impossible without lots of help from friends. It is powered by beer, much energy and love, batches of bug reports and support on the operating system integration side by denis, taklamakan, cryx, supergrobi and – his libowfat always on the bleeding edge – Fefe. Thanks are also due to Hannes for helping me out with designing internal data representation.