Infrastructure @
Updated 1,701 Days AgoPublic

This wiki (and it's contents) contain all the public information about infrastructure and maintenance.

The Infrastructure Team

The infrastructure team is ultimately responsible for maintaining services. It is not required you be a member of any existing group (for example, the committee) to join. The infrastructure team deals solely with technical aspects of infrastructure, including sensitive matters like domain management, backups, administrative email, security response, service allocation, and domain or URL routing.

Strictly speaking, you do not even need to be a member of the infrastructure team to manage a service - many smaller, independent services are operated by people not on the infrastructure team, many on the same machines we tend to manage. You may certainly run a service with us on your own machine too (and we may provide you a DNS entry, for example). However, the infrastructure team generally carefully overlooks most changes made across the infrastructure to all core services and machines, and we prefer to automate in-house work where applicable. Most questions about the infrastructure are also handled by the team directly.

The current team is:

You can contact the current team at any time with an email to or, or hit them up on Phabricator or IRC (#haskell-infrastructure on freenode).

Infrastructure providers uses many service providers for its infrastructure.

  • Hetzner is by far the oldest provider still around. There is one servers owned by with Hetzner: lambda -- a legacy server that hosts some defunct services, and most importantly,
  • Rackspace provides compute, storage, and networking resources, powering almost all of in several regions around the world. All of our new services are powered by Rackspace, and we're migrating all Hetzner services to Rackspace as well.
  • DataDog provides powerful, customizable 24/7 metrics and monitoring integration, primarily connected with our Nagios instances.
  • Fastly provides low latency access for all of's downloads and highest traffic services, including the primary Hackage server, and Haskell Platform/GHC downloads.
  • CloudFlare provides DNS management, Anti-DDoS, and high-grade frontend SSL and caching for most of
  • powers, and lets us easily tell you when we broke something.
  • DreamHost supports by providing free and unlimited object storage and bandwidth via DreamObjects, which we use to stash logs, perform backups, and generally host all kinds of long-term content that we might need.
  • supports and the Glasgow Haskell Compiler by providing free, ARM-based Cloud Servers for development and testing.
  • (a brand of OVH) supports and the Glasgow Haskell Compiler by providing us incredibly powerful POWER8 based build servers.
  • Webmon supports by providing external monitoring and failure notifications for our core public-facing servers.

Service details

  • - the primary website, powered by rHL, a Haskell web application server.
  • - Our MediaWiki instance, hosting tons of content.
  • - The primary Hackage instance for; this setup is proxied in the front by Fastly, to provide low-latency downloads and caching for Hackage
  • - Serving a variety of haskell-related mailing lists
  • - Our Kibana system, which talks to our backend ElasticSearch.
  • - Our Planet Venus instance, which powers the community RSS feeds.
  • - Our Nagios and DataDog aggregation endpoint, which we use for tactical overviews and public metric reports
  • - This place right here!
  • - A downloads server hosting tons of static content, proxied by Fastly.
  • - Our legacy Darcs repositories.
  • & - The GHC git repositories and Trac system.
  • & - Our legacy community infrastructure
  • - A service providing status updates on infrastructure issues (provided by

More information

For more information, be sure to check out the document hierarchy below.

If you want to see what's been going on recently, check out the Server Admin Log.

Last Author
austin, relrod, gershomb and 3 others