Toolserver/Presentation

From Meta, a Wikimedia project coordination wiki

This is going to be a presentation about the Toolserver for Wikimania 2006 http://wikimania2006.wikimedia.org/. Everybody actively working with the Toolserver is invited to prepare 1-2 slides with a description of his tool, frameworks, statistics etc. for the benefit of all people interesed. You can also write about the general status, problems, plans and visions.

Submission[edit]

See Wikimania's CfP.

Title
Tools for Wikimedia projects - the Toolserver approach
Abstract
The open nature of Wikimedia projects supports the creation of additional tools like bots, statistics and scripts for cleanup and research. To collect and support these tools the idea of a dedicated server with access to the Wikimedia project's databases rose in 2005. When Sun donated a server to Wikimedia Deutschland it was installed in Amsterdam as Toolserver. Since then a lot of tools are developed, for instance daniel's CatScan and kate's editcounter. The presentation will give an overview of the Toolserver, its tools, users, problems and future possibilities.
Format
Presentation, 30 minutes, notes at http://meta.wikimedia.org/wiki/Toolserver/Presentation

coordination[edit]

  • Gregory Maxwell
  • Jakob Voß

topics[edit]

history[edit]

The server was donated by Sun in Summer 2005 to Wikimedia Germany. It is hosted together with Kennisnet's servers and Wikimedia squid servers in Amsterdam.

usage statistics[edit]

  • access...
  • number of projects/accounts
  • growth of disk usage
  • ...

organisation[edit]

  • mostly self-organized
  • account-policy
  • ...


technical architecture[edit]

Hardware[edit]

  • Sun v40z
    • 2* Opteron 848, 2.2 GHz
    • 8 GB RAM
    • 6*146 GB disk in RAID-1 mirror for / and RAID-0 stripe for /u01 (incl. user homedirs and MySQL data)
    • built in hardware RAID controller (not used)
    • External RAID array (12*400 GB)

Software[edit]

  • Existing disks use SVM to mirror /, and stripe /u01. /u01 contains all the important data. (this will change once the array is installed)
  • MySQL runs in the global zone, all user accounts + apache/tomcat/etc are on the login-services zone, for security reasons.
  • It runs Solaris 10 3/05
  • What else?

Usage stats[edit]

Future plans[edit]

  • What hardware do you need? What architecture makes sense and what is maintainable?

Problems[edit]

Gregory?

  • "It's a bit crazy, we want to replicate the data of a whole cluster on one single server -- and we want it not to lag" (Leon)
  • lag, lag, lag
  • long queries vs. short queries?
  • An additional RAID was ordered in December 2005 and installed in April 2006 because of several delays - this upgrade is far too slow.

frameworks[edit]

daniel, interiot - can you please describe your frameworks in some sentences?

  • WikiSense: PHP
  • Interiot: Perl

Many minor scripts in

  • php, perl, python, shell, C, SQL, STX, Tcl .. please add your language here to get a list of all programming languages used on the toolserver

tools[edit]

describe your favorite tool here

funny notes[edit]

...

images[edit]

replag: http://tools.wikimedia.org/~leon/stats/replag/replag-daily.png and http://tools.wikimedia.org/~leon/stats/replag/replag-both-daily.png