Daily Bulletin Archive

Dec. 4, 2017

No Downtime: Yellowstone, Cheyenne, Geyser_Caldera, GLADE and HPSS

Dec. 1, 2017

The CISL User Services Section will present an in-person and online tutorial at 1:30 p.m. MST Monday, December 4, and again at 9 a.m. MST Tuesday, December 19, on how to run jobs on the Geyser and Caldera clusters from the Cheyenne HPC system. The Yellowstone system is being decommissioned December 31, but Geyser and Caldera will remain in production throughout 2018.

The 45-minute tutorial will cover submission and monitoring of jobs with the Slurm scheduler, running an interactive job, performing graphically intensive tasks on Geyser, and considerations when using Cheyenne and Geyser/Caldera together in a workflow. The session will include time for questions.

Register to attend in person at NCAR’s Mesa Lab in Boulder or via webcast by selecting a link below:

Register for December 19 tutorial

Attend in person (Chapman Room, ML245, space is limited)

Attend webcast

Nov. 30, 2017

The Yellowstone supercomputing cluster will be decommissioned on December 31, 2017. The system will be powered down on that date and no longer available to users. The Geyser and Caldera clusters will remain in production until the end of 2018.

Yellowstone’s maintenance contract expires December 20. CISL will continue to operate the system as a courtesy to users but should any hardware components fail before December 31, no effort will be made to repair them.

Please note that it is not necessary to copy or move files to HPSS or to another disk space as a result of Yellowstone’s decommissioning. All files in the GLADE system file spaces—home, scratch, work, and project—are accessible via Cheyenne. Copying or archiving files unnecessarily will waste expensive tapes and valuable disk space while increasing use of your allocation.

Users who have not already done so are strongly encouraged to migrate their work to Cheyenne as soon as possible to avoid unnecessary interruptions. Documentation available to help Cheyenne users make the transition includes this Quick start overview as well as procedures and scripts for running jobs. For help in migrating job scripts or rebuilding executables, please contact cislhelp@ucar.edu.

See our allocations documentation if you do not already have a Cheyenne allocation. University users who want to convert an existing Yellowstone allocation to a Cheyenne allocation should mail that request to cislhelp@ucar.edu.

Nov. 29, 2017

Registration is open for a workshop for new NCL users February 6-9, 2018, at NCAR’s Center Green Campus in Boulder. The workshop will include introductory lectures on NCL, scientific data formats, and graphics, and daily hands-on labs. During the labs, instructors will work with students to help them write NCL programs for analyzing their own data.

Registration is limited to 16 students. The registration deadline is January 15, 2018. Travel support is available on a first-come, first-served basis for qualified students from EPSCoR states or minority-serving institutions. The deadline for applying for funds is December 11, 2017.

Get full details and a link to the registration form here.

Nov. 24, 2017

A recording of the November 9 webcast, “Moving from Yellowstone to Cheyenne,” is now available on the CISL web site. See the course page to review the presentation and download the slides.

The tutorial was presented to help Yellowstone users migrate their work to the new Cheyenne HPC system before Yellowstone is decommissioned at the end of December. It covers considerations when configuring your Cheyenne account for the first time, how to load supported software on Cheyenne, scheduling and running jobs using the PBS scheduler, and using MPI libraries for parallel jobs.

Additional training for how to run jobs on Geyser and Caldera from the Cheyenne system is being planned. Details will be announced soon.

Nov. 17, 2017

The Cheyenne cluster was released to users at approximately 5:30 p.m. on Thursday.

CISL staff and HPE engineers identified several issues that are believed to have caused the system’s instability and they replaced problematic hardware components. The system was tested extensively before it was released and as of this morning is meeting or exceeding established performance metrics.

CISL again thanks you for your patience throughout this downtime.

Nov. 16, 2017

The Cheyenne cluster is expected to released to users later today, Thursday, Nov 16.  

CISL staff have been working closely with HPE to identify sources of the system’s instability and a number of hardware components have been replaced or repaired.  Currently we are focused on verifying system performance and the system will be released when expectations have been met or exceed.

CISL will inform users through the Notifier service when the system is restored and again thanks you for your patience throughout this extended outage.

Nov. 15, 2017

The Cheyenne cluster is expected to remain unavailable today, Wednesday, Nov 15.  At this time there is no estimate for when the system will be restored to service.

CISL staff are continuing to track down the source of the system’s instability and several hardware components have been replaced or repaired. The vendors HPE and Mellanox are working closely with the system administrators to resolve the problems as quickly as possible.

CISL will inform users through the Notifier service on any significant updates and again thanks you for your patience.

Nov. 14, 2017

CISL staff are testing new procedures for starting Geyser and Caldera jobs from the Cheyenne system and will make related user training and documentation available soon. While the Yellowstone cluster will be decommissioned on December 31, 2017, the Geyser and Caldera clusters will remain in production until the end of 2018 as announced previously.

Keep an eye out for training date announcements and other updates in the Daily Bulletin.

Nov. 6, 2017

The CISL User Services Section will present an in-person and online tutorial at 9 a.m. MST Thursday, November 9, at the NCAR Mesa Lab in Boulder to help Yellowstone users migrate their work to the new Cheyenne HPC system. Yellowstone will be decommissioned December 31.

The 60-minute tutorial will cover considerations when configuring your Cheyenne account for the first time, how to load supported software on Cheyenne, scheduling and running jobs using the PBS scheduler, and using MPI libraries for parallel jobs. The session will include time for questions.

Register to attend in person—in the Chapman Room in NCAR’s Mesa Lab (ML245)—or via webcast by selecting a link below:

Pages