Greetings Lustre Users - As many of you know, CFS recently commenced the 4th annual Lustre User Group (LUG) which was held on the week of April 17 in (mostly) sunny Hilton Head, North Carolina. For the benefit of the Lustre Community - I wanted to take a moment and summarize the activities and responses we''ve seen from the conference: ----- -Attendance- 2006 marks another growth year for Lustre and CFS - as such, attendance at the LUG was up approximately 40% from the Spring 2005 meeting. It was wonderful to see many of our long time friends, as well as meet some of our newer users. ----- -User Presentations- There were a variety of user and vendor presentations delivered at the User Group. The presentations are being posted to http:// www.clusterfs.com/lustre2006.html as we receive them from the presenters. ----- -Open Development- The topic of open development has circulated around Lustre frequently over the past few months. As such, discussions at the User Group served to lay the foundations for a more concentrated and coordinated community effort. We are seeing some encouraging early demonstrations of these efforts, targeted at improving Lustre and its usability. To name a few: - Myricom is contributing a Lustre Networking Driver for their Myrinet Express (MX) Software Layer for their 10Gb products - Scalable Systems has developed a Lustre Roll for the ROCKS Cluster Distribution Suite Those hoping to contribute are encouraged to subscribe to lustre- devel@clusterfs.com to help shape, monitor and engage this growing community. ----- -Lustre Requirements Forum Results- Toward the end of the conference - Lustre Users were given the opportunity to identify, discuss and prioritize their requirements from the Lustre File System. After some systematic selection of possible Lustre improvements, we were able to identify the following five items (in order of importance) as the most pressing Lustre User Requirements: 1 - Understandable and documented error messages; More simplified troubleshooting 2 - OST stripe management for: - Storage Pools - The ability to join and extend files across multiple OSTs - background migration 3 - Improved Logging, Debugging and Diagnostic tools; NID logic; per- client statistics 4 - Lustre HPSS copy, using DMAPI 5 - I/O & metadata performance profiling, analysis, and reporting Other suggested improvements can be found prioritized at https:// mail.clusterfs.com/wikis/lustre. Over the next few months, CFS will work to align its development priorities with this list to better suit the wishes of its User Community. Additionally, we are actively seeking the contributions of the community to define these efforts and contribute in various ways to their development. ----- I''d like to thank all of the attendees for making this year''s Lustre User Group very productive. Please feel free to contact us with any questions or comments related to this meeting. We look forward to seeing everyone at the next User Group, where we hope to have much to report back to you! Jeffrey Denworth Director of Sales Cluster File Systems, Inc.