Messages & Announcements

  • 2010-09-08:  Firefly: Unplanned partial shutdown 9/9/10
    Category:  Maintenance

    A remodeling job will begin tomorrow morning at PKI which will force at least a partial shutdown of the Firefly cluster. Further details will be added to the web site.

    Sincerely,
    David Swanson


    Significant remodeling work required to upgrade the condition of a conference room at PKI will impact the adjacent machine room that houses Firefly. It is scheduled to begin tomorrow morning, Thursday, 9/9/10. The time frame of this project is up to 2 weeks, although it is hoped it will be shorter. The best case scenario is that we will be able to maintain operation of roughly 75% of the cluster during this process, but that will have to be verified as work proceeds. Further details will be posted here as information becomes available.

  • 2010-09-03:  Firefly unplanned network outage 9/3
    Category:  System Failure

    Firefly's fiber connection between PKI (where Firefly resides in) and Nebraska Hall in Lincoln was cut.

    The network engineers are working on restoring service.


    Firefly's fiber connection between PKI (where Firefly resides in) and Nebraska Hall in Lincoln was cut.

    The network engineers are working on restoring service.

  • 2010-08-20:  Scheduler change for PrairieFire Monday, Aug. 23
    Category:  General Announcement

    For historical reasons, the schedulers running on PrairieFire and Firefly have not been identical to date. After over a semester of use, the Maui scheduler has proven adequate for our current needs on Firefy, and we will be switching from SGE to Maui on Prairiefire on Monday of next week. HCC staff will be available to answer questions or meet with researchers if any problems are encountered.


    Reasons for using identical schedulers on by PrairieFire and Firefly include the following:

    Increased ease of use of both machines for HCC researchers.
    Simplified administration for HCC staff.
    Potential integration (unification) of the two clusters' scheduling environment.

    Reasons for selecting Maui include the following:

    Maui is open source and free.
    Maui is built on Torque, which is more similar to PBS Pro, currently running on Merritt. (Merritt is a special case and will not be changed.)
    SGE was formerly a product of Sun, which was purchased by Oracle. Its future is thus less certain.
    The commercial version of Maui (Moab) is used at several of the largest supercomputing centers in the country.

    Details for using Maui on PrairieFire may be found here:
    <a href='http://hcc.unl.edu/hcccreditedit/faq.php?tp=PFSched'>http://hcc.unl.edu/hcccreditedit/faq.php?tp=PFSched</a> .

    <i> Existing SGE scripts will no longer function. You will have to change these to a PBS-style script as discussed in the above link. </i>

    Please notify HCC staff if you have any difficulty or questions.

  • 2010-07-30:  Firefly fiber maintenance, July 30, 2010
    Category:  Maintenance

    A temporary fiber outage for Firefly is scheduled from 1:00 PM to 2:00 PM July 30, 2010.

    Firefly's fiber network provider will be running diagnostics. During this period, Firefly's internet connection may be intermittently unavailable. Submitted jobs should continue to run.


    A temporary fiber outage for Firefly is scheduled from 1:00 PM to 2:00 PM July 30, 2010.

    Firefly's fiber network provider will be running diagnostics. During this period, Firefly's internet connection may be intermittently unavailable. Submitted jobs should continue to run.

  • 2010-07-19:  Firefly planned outage July 19, 2010
    Category:  Maintenance

    We will schedule this outage Monday, July 19, 2010, starting 11:30 AM. This outage may potentially last until late in the day. We will send out an announcement once this patch has been applied.


    On July 2, 2010, we discovered a bug with the Panasas storage on Firefly. We sent out a message not to delete files larger than 2.8GB on /home or 7GB on /work as the storage blades on our Panasas storage would crash.

    Panasas has now made a fix available for this issue. We will need to schedule an outage on Monday, July 19, 2010 to apply this patch. While we understand that this means yet another outage for Firefly, this is a critical issue that needs to be solved to prevent potential data loss on /home and /work.

    We will schedule this outage Monday, July 19, 2010, starting 11:30 AM. This outage may potentially last until late in the day. We will send out an announcement once this patch has been applied.


    We apologize for the inconvenience caused by this outage.

Pages