GNOME System Adminstration Projects
Here's a partial list of some of the pressing short term and longer term projects that the GNOME sysadmin team could and should tackle given capacity. Italics mark thoughts about what would be needed to move forward.
Short Term
- Catch up on routine tasks
the syadmin team is currently having a lot of trouble keeping up with routine tasks such as account and mailing list creation. The big need here is better coordination. See /FormalTeam
- Version control migration
there is a strong community desire to move from Subversion to a more modern version control system. There is current work to explore the details of migrating to git (GitMigration). Mostly well in hand; some question about available system capacity.
Medium Term
- Self service
Increase self-service capabilities for password resets, email changes and so forth to reduce the amount of routine work. Needs a volunteer to pick up work on mango.gnome.org
- Configuration management
There is very little management of changes to the GNOME servers - configuration files are usually just modified directly making moving services between servers or reinstalling very difficult. The "puppet" system has been prototyped on some of the newer systems such as online.gnome.org and the playpen git.gnome.org. Needs someone dedicated spending a good amount of time on it or a well-coordinated team of volunteers. /JobDescription? /FormalTeam?
- Hardware Upgrades
Project critical services, including the bugzilla database, home directories, and the mail server are running on machines from 2003 and 2004 that are out of their service contracts. An hardware failure on any of these services would cripple GNOME development until replacement capacity was found. Hardware donations or fundraising toward new modern servers where services could be consolidated (virtualization)
- Bugzilla performance
Bugzilla can be quite slow at times (it can completely freeze up for minutes at times.) Bugzilla and the Bugzilla database share a pair of old and small servers with live.gnome.org and other service. Additional hardware could be very helpful here - throw memory at the database
Long term
- End-user services
- Modern computing is largely not about stand-alone desktop apps but rather apps with desktop and distributed components. GNOME associated projects (gnome-games, Conduit, Abiword, etc.) have taken footsteps into this area, however, we've not been able to provide support within the framework of the GNOME systems, and the projects have either been stymied or created a system from scratch with external hosting.
- Distributing hosting
- Find ways to take advantage of hosting outside the Red Hat colo.There's a good potential for significant donations of resources by ISPs and others (Fedora does great on this), but our systems are generally set up assuming a shared physical back-channel network. (svn.gnome.org and l10n.gnome.org are already hosted separately, but there are some awkward aspects.) This would like be important for scaling up to handle offering services to large numbers of end-users.