Feedback

The most visible activity of the NCCS UAC is assisting users in the day to day use of the NCCS resources. A less visible but equally important activity is creating and maintaining a user-friendly environment. To accomplish this goal, we are constantly looking for ways to improve the center’s interaction with the user community.

Often ideas for improvement are gathered from interaction with users and by monitoring system-usage behavior. We also welcome and greatly appreciate direct suggestions and comments from users.

To send a comment or suggestion to the NCCS UAC, you can contact the center directly or complete the suggestions form.

The feedback form gives you the ability to provide your contact information or, if you would prefer, submit anonymously.

Any feedback that will help us provide better service is welcome. For example, if you would benefit from the modification or addition of a policy, please let us know. If you would prefer longer notification periods for outages, please let us know. Or, if you would like to comment on previous assistance you received from the UAC, please let us know.

Below are examples of previous suggestions and the resulting actions.

  • Problem

    • The NCCS sends too many emails detailing outages and system events.

    • The NCCS does not send enough emails detailing outages and system events.

  • Action Taken

    • Different users would like different levels of communication from the center. Some users would like emails each time a system goes down and then another once it is returned to service. Other users believe this is far too many emails. As a solution we chose to send a single weekly email to all NCCS users containing outages, changes, and noteworthy events. The weekly email contains information for each LCF resource. For those who would like more notice of events, we added system-status arrows to the site as well as a system-status page detailing the events of outages. We are also working to provide RSS feeds for each system.

  • Problem

    • Large jobs sit in the queue for long periods of time before running.

  • Action Taken

    • Many factors can cause a job to move slowly through a queue. For example, jobs will naturally move more slowly through the queue of a very heavily used system as opposed to a lightly used system. The NCCS is a capability center with the goal of running large-scale computational science projects requiring capability-class simulations (i.e., occupying a substantial portion of the system’s resources at any given time) for achieving high-impact science. Because of this, large jobs are favored over smaller jobs. We have recently made several changes in the queueing policies with a goal of increasing throughput for larger jobs. These policies are currently being implemented on Jaguar and are documented in Jaguar’s running-jobs section.

  • Problem

    • Are the NCCS systems currently available?

  • Action Taken

    • Due to unforeseen system issues, preventative maintenance, and upgrades, the NCCS systems are occasionally down or unavailable for login. System-status arrows were added to the right sidebar of this website to show the current status of each LCF system. The arrows are automatically updated when the status of a system changes.

  • Problem

    • Can I find a list of previous and upcoming outages and the events that occurred or will occur during the outages?

  • Action Taken

    • Upcoming and previous outages and the events that are planned or occured during the outage are listed on a given machine’s system-status page. These page are updated each time an outage is scheduled and after each outage.

  • Problem

    • How can I determine my current usage?

  • Action Taken

    • The utility showusage can be executed from any allocated NCCS system. The utility will return the year-to-date usage for the calling user as well as any associated
      project(s).

    • Users on one of the allocated INCITE projects can see usage charts for each project for which they are a member at https://users.nccs.gov.

  • Problem

    • How do I plan my Jaguar usage when the system is often taken down for upgrades?

  • Action Taken

    • The NCCS has aggressive plans to increase the capability of its Cray XT supercomputer, known as Jaguar. These upgrades will increase Jaguar’s capacity fivefold, to 250 teraflops, and pave the way for the center’s upcoming petaflop computer. To minimize the disruption to our users and allow them to anticipate changes, the NCCS has outlined a schedule calendar on the Jaguar calendar page.

  • Problem

    • How can I keep up with general system events?

  • Action Taken

    • In a further attempt to communicate system events, the NCCS hosts a monthly teleconference. During the monthly teleconference, general system announcements and other topics of interest are discussed. The call also gives participating users a forum in which to bring up other topics of interest for discussion. The call is open to all NCCS users. Notes from previous calls are listed in the users’ teleconference section.

    • General information is added to the login message of the relevant systems.

    • A weekly email is sent to all NCCS users containing general information.

    • Users have the option to join a more detailed email list.

    • Information is added to the each system’s resources page.

  • Problem

    • I would like to share code and batch scripts with others members of my project.

  • Action Taken

    • A space for each project is created in the format /spin/proj/<projid>.

    • A group has been created for each project named the same as the project ID. Membership to this group is automatically given to each project member.

Users’ Teleconference

The NCCS hosts a monthly teleconference with all NCCS users to discuss pressing issues. All NCCS users are invited to attend the teleconference and express their opinions and ideas. All users receive a monthly reminder, including the date and time of the next teleconference.

Minutes of past teleconferences:

January 2007 jan07concall.pdf

February 2007 feb07concall.pdf

March 2007 mar07concall.pdf

April 2007 apr07concall.pdf

June 2007 jun07concall.pdf