User:Aude/Toolserver
 | This is a draft article being written by User:Aude for the Wikipedia Signpost |
The Toolserver (or more accurately, the Toolserver cluster) is a project run by Wikimedia Germany that allows programmers to develop web-based tools and facilitate other such tasks. This week, the Signpost interviewed Daniel Kinzler, an employee of Wikimedia Germany, to find out more about how the Toolserver operates.
What is the Toolserver and how does it operate?
The Toolserver allows interested programmers to work with a live copy of the wiki databases. The Toolserver is used mainly to provide specialized web based tools to the wiki communities, but also to run bots and similar services. The Toolserver is also open to researchers that want to study Wikimedia's projects.
The Toolserver cluster, is located in Amsterdam, has several types of machines.
Mainly, there are three:
- The database servers, which contain a live copy of all the wiki databases — however without the full article text. Article text is stored in compressed blocks on a special cluster. There is a copy of this in Amsterdam, but it is not accessible directly to Toolserver users. The main reason for this is that these compressed blocks may contain private data.
- Then there is the web server, where users can provide web based tools. This is what the wiki communities usually see.
- And there are the login servers, machines where programmers can run bots and other "offline" scripts, or run commands directly.
Who runs it? and how is it funded?
The Toolserver is run by Wikimedia Germany, it is funded entirely from donations. The Wikimedia Foundation helps with hosting the machines in Amsterdam.
What is your role with the toolserver?
I am an employee of Wikimedia Germany, and I'm responsible for planning the toolserver budget, among other things. I also have root access, and can approve new accounts, but I'm a lousy system administrator. So mainly my role is planning and organizing. I am of course also a toolserver user and provide several tools, the most well known of wich are probably CatScan and CheckUsage.
How did the toolserver get started?
To be honest, I don't know the details of the very beginning. I only joined when it was already up for a couple of months. But I believe that there was some frustration by people who had the skill to develop helpful tools for the wiki communities, but did not get access to the main server cluster, for security reasons. So the idea arose to create a place for such people to work with the data.
How do Wikipedians get accounts on the toolserver? What are the requirements and approval process?
Wikipedians (or anyone else, actually) can request an account by simply stating what you want to do there. The requests are reviewed by DaB (and sometimes by me), and if the project falls into the Toolserver's scope (which is: supporting Wikimedia and, more recently, OpenStreetMap projects), seems technically feasible and complies to our terms of use, the request is granted.
Until now, account requests have been handled at new accounts. However, we are currently reworking the process, to make it more swift and more transparent.
The terms of use for the toolserver are defined by rules, the conditions for account approval are at https://wiki.toolserver.org/view/Account_approval_policy/en.
If one gets an account, what are the responsibilities?
It's the user's responsibility to comply to the rules, that is, to do no evil. There are no hard limits on resources, users are asked to be considerate. Heavy tasks that slow down the system may be killed by an administrator. Generally, we try to address problems on a case by case basis, directly with the user.
Who is the person to contact when things break?
That depends on what is broken. We have over a hundred active users, many of which have several tools in diverse states of completeness or repair. Generally, when something is broken, contact the author of that something - Toolserver administrators usually can not fix it.
The matter is different when the Toolserver as such has a problem - for example, when one of the servers is down. In that case, write an email to ts-admins@toolserver.org. You can also find us on IRC, in the channel #wikimedia-toolserver.
Recently some new hardware was recently ordered. What's the purpose of the new hardware?
The latest hardware order included five servers, three of them for the Toolserver cluster:
- We are replacing our oldest database server, which has been constantly overloaded recently.
- We are putting user data (home directories) onto a new machine and move other things around a bit. In the end, there will be a second login server, for running bots and the like.
- And last but not least: the map toolserver, a place for the OpenStreetMap community to develop exciting new projects.
In addition, two servers are provided for integrating "live" interactive maps on Wikipedia articles, based on OpenStreetMap data.
What's coming in the pipeline?
We are mainly working on the reliability of the toolserver - for instance, we hope to have two copies of every wiki database in the future, so tools using that data can keep running even when one system fails.
Generally, we are trying to make the Toolserver cluster a mature part of the Wikimedia landscape.
Anything else to add about the Toolserver?
We should all say thanks to River, our senior system administrator, for keeping the Toolserver running!
More information
[edit]- http://lists.wikimedia.org/pipermail/toolserver-l/2009-June/thread.html (more to add to this article)