2007 CHPC News Announcements

CHPC DOWNTIME: Thursday January 3, 2008

Posted: December 19, 2007

Event date: January 3, 2008

Duration: Downtime starts at 3pm and will last until sometime early morning on January 4, 2008

Systems affected:

All of Arches and CHPC/INSCC Network

After this downtime all users will be using the campus uNID and password for authentication on all HPC systems (and other Linux systems admined by CHPC). Windows users will use the uNID and current password for authentication.

Arches:

All clusters will be down from 3pm to allow for updates to the OS and for the other changes outlined below. The Batch Queues will be drained of all running jobs. Reservations are in place so that jobs will not be started if they will not finish before the start of the downtime. Jobs that are queued but not running will be started after the downtime ends. The one exception to this is if you are being moved to using your unid for authentication during this downtime (see below); in this case any queued jobs you have will need to be deleted. The clusters will down until sometime the following morning.

**MIGRATION TO NEW FILESERVER: Some CHPC Users, those on the CHPC owned home directory filesystems i.e., those with home directories /uufs/inscc.utah.edu/common/home/USERID - will be migrated to a new, larger fileserver during this downtime. If you are one of these users your new home directory path will be /uufs/chpc.utah.edu/common/home/UNID

**CHANGE TO UNID: All CHPC users that are not already using their UNID as the CHPC login will be changed to doing so. If you do not have a UNID you will need to get one BEFORE this downtime. All University of Utah students and employees automatically have a UNID. But if you are a not a part of the University of Utah, you need to fill out a Person of Interest (PoI) form to get assigned a UNID. This form can be found at http://www.hr.utah.edu/forms/lib/u-affiliate-poi-form.pdf.

Network Outage:

All networking in CHPC/INSCC will be down from about 5-7pm


CHPC DOWNTIME: Thursday January 3, 2008

Posted: November 29, 2007

Duration: Times and Scope to be determined

Systems affected:

After this downtime all users will be using the campus uNID and password for authentication on all HPC systems (and other Linux systems admined by CHPC). Windows users will use the uNID and current password for authentication.

Arches:

To Be Determined. (Queues will be drained at a minimum.)

Network Outage:

To Be Determined


CHPC DOWNTIME: September 13, 2007

Posted: September 4, 2007

Systems Affected/Downtime Timelines: *****/scratch/serial-old will be flushed and rebuilt********** Network outage, including wireless, from 5:30 p. m. until approx 7:00 p. m.

Arches Downtime Duration: Sanddunearch down from 11 a.m. until approx noon on September 14th Remainder of arches down from 5 p.m. until approx noon on September 14th

Systems affected: All of arches, chpc network

Arches: CHPC will be taking down the sanddunearch cluster on Thursday September 13th at 11a.m. to continue the maintenance on the Infiniband fabric. The remainder of arches will be taken down on at 5 p.m. Jobs idle the queue will remain in the queue, but a system reservation is in place so that jobs which cannot complete before the beginning of the downtime will not start.

Network Outage: This portion will affect all of arches as well as desktop access to filesystems and to rest of world. Wireless will also be affected. Connectivity will be intermittent during this time period.

This downtime is scheduled to perform maintenance on the KOMAS cooling system. OS updates will be applied to arches. /scratch/serial-old will be rebuilt to increase the file space available. All data on the /scratch/serial-old will be deleted. Please remove any necessary data from /scratch/serial-old before the start of the downtime.


CHPC Fall 2007 Presentation Series

Posted: August 23, 2007

CHPC Fall 2007 Presentation Series begins August 30th, 2007

CHPC is kicking off its Fall 2007 Presentations beginning next week! Mark your calendars!

Where: INSCC Auditorium (room 110 INSCC) (http://www.chpc.utah.edu/chpc/office/map.php)

When: Thursdays at 1:00 p.m. (most last about an hour)

For more details please go to http://www.chpc.utah.edu/docs/presentations/

  • 8/30 - Overview of CHPC (by Anita Orendt)
  • 9/6 - Statistical resources at CHPC (Byron Davis)
  • 9/13 - Introduction to Parallel Computing (Martin Cuma)
  • 10/4 - Introduction to programming with MPI (Martin Cuma)
  • 10/18 - Debugging with Totalview (Martin Cuma)
  • 10/25 - Mathematical Libraries at CHPC (Martin Cuma)
  • 11/1 - Parallel performance with Tau (Martin Cuma)
  • 11/8 - Chemistry Packages at CHPC (Anita Orendt)
  • 11/15 - Gaussian03 & Gaussview (Anita Orendt)
  • 11/29 - Fast parallel I/O (Martin Cuma)
  • 12/6 - Intro to programming with OpenMP (Martin Cuma)

CHPC Downtime: Arches and Network 16th, 2007 beginning at 5 p.m.

Posted: August 14, 2007

CHPC Downtime: Arches and Network 16th, 2007 beginning at 5 p.m.

Systems affected:

All arches clusters will go down at 5 p.m. to replace critical hardware which is failing. The duration of the arches downtime is unknown. We apologize for the short notice. Any running jobs will be killed but jobs waiting in the queue should ride out this downtime.

Connectivity to the INSCC building, including wireless, will be intermittant from 5-7 p.m.

Duration:Thursday August 16th:

  • Network from 5:00 p.m. - approximately 7:00 p.m.
  • Arches from 5:00 p.m. - until repairs are completed

CHPC Allocation Requests Due September 1st, 2007

Posted: August 8, 2007

CHPC Allocation Requests Due September 1st, 2007

This is a reminder that proposals and requests for allocation on the CHPC ARCHES cluster are due by September 1, 2007. We must have this information if you wish to be considered for an allocation of time for the Fall 2007 (Oct 1 - Dec. 31) calendar quarter. You may apply for up to four quarters at a time.

  1. Information on the allocation process and relevant forms: Allocation Policy and Instructions Only those requests following these instructions will be considered.
  2. Fall quarter allocations will go into effect on October 1, 2007
  3. Send your proposal and relevant information to the attention of: Janet Ellingson, INSCC Room 405

CHPC to change usernames to unid during August 2007

Posted: July 27, 2007

CHPC to change usernames to unid during August 2007

CHPC will be migrating user authentication to the campus kerberos system for our unix systems (including the clusters) over the next month or so. This will position us to take advantage of any federation relationships the campus may participate in. Also, it is simpler for users to have only one campus username (unid) and password to remember and the kerberos security is much stronger that what we are currently using.

We expect to have this project completed by the end of August. This means that your username on our systems will change to be the campus unid. For unix systems, your password will also change to the one you use for campus systems such as Umail or Kronos. All new accounts will be created using unids.

Windows/Mac desktop access: If you use CHPC for your windows/mac support, your username for access will be your unid, but for the time being, the password will be the same one it is today. We will move to the campus password on those platforms sometime after we complete moving everyone's username to unid.

I will be contacting the PI for each group when we are about a week out from scheduling this for your group. In order for us to make these changes the users must be logged off of all systems and must not have any running jobs on any CHPC system. If there is a better time than another to schedule this, please let me know.

If there are members of your group who do not have unids on campus, then your department will need to get one assigned. To do this, you will need to use the "U Affiliate/POI Form" Please see: http://www.hr.utah.edu/forms/lib/u-affiliate-poi-form.pdf. For example if you have a user who is a researcher at another university doing collaborative work with you, you can get a unid assigned using this form, so they may continue to have access to the CHPC systems. In this case you would use the "10017 Other Faculty" designation on the form. Once you have the unid, just send it to us and we'll get the user migrated over.

We have already migrated the CHPC staff and have had few problems. Please let me know if you have any questions or concerns about this process.


CHPC Downtime: June 21, 2007

Posted: June 14, 2007

Duration: Sand Dune Arch down from 4 p.m. until approx 7 p.m.

Network outage from 5:30 p. m. until approx 7:30 p. m.

Systems affected: The first part of this outage will only affect only sanddune arch

CHPC will be taking down the sanddunearch cluster on Thursday evening, June 21st at 4:00 p.m. to continue the maintenance on the Infiniband fabric. The downtime is expected to last approximately 2 to 3 hours. Jobs in the queue will remain in the queue, but a system reservation is in place so that jobs which cannot complete before the beginning of the downtime will not start.

Systems affected by Network Outage: This portion will affect all of arches as well as desktop access to filesystems and to rest of world

CHPC will be replacing the link, upgrading it from 1G to 10G, that connects the CHPC network to campus and the path out of campus. This will require that all running jobs on arches be drained. Jobs idle in the queue at the start of the downtime will remain in the queue and start after the downtime.


CHPC Downtime: Sand Dune Arch down June 11th, 2007 from 9 a.m. until 5 p.m.

Posted: June 7, 2007

CHPC Downtime: Sand Dune Arch down June 11th, 2007 from 9 a.m. until 5 p.m.

Systems affected:

CHPC will be taking down the sanddunearch cluster on Monday morning, June 11th at 9:00 a.m. to continue the maintenance on the Infiniband fabric. The downtime is expected to last approximately 8 hours. Jobs in the queue will remain in the queue, but jobs which cannot complete before the beginning of the downtime will not start.

Duration:Monday June 6th from 9:00 a.m. - approximately 5:00 p.m.

Scope: This downtime will only affect sanddunearch.


CHPC Downtime: Sand Dune Arch down June 4th, 2007 from 9 a.m. until 9 p.m.

Posted: June 6, 2007

CHPC Downtime: Sand Dune Arch down June 4th, 2007 from 9 a.m. until 9 p.m.

Systems affected:

CHPC will be taking down the sanddunearch cluster on Monday morning, June 4th at 9:00 a.m. to perform maintenance on the Infiniband fabric. The downtime is expected to last until about 9:00 p.m. Jobs in the queue will remain in the queue, but jobs which cannot complete before the beginning of the downtime will not start.

Duration:Monday June 6th from 9:00 a.m. - approximately 9:00 p.m.

Scope: This downtime will only affect sanddunearch.


CHPC Downtime: Sand Dune Arch down June 4th, 2007 from 9 a.m. until 9 p.m.

Posted: June 4, 2007

CHPC Downtime: Sand Dune Arch down June 6th, 2007 from 9 a.m. until 9 p.m.

Systems affected:

CHPC will be taking down the sanddunearch cluster on Monday morning, June 4th at 9:00 a.m. to perform maintenance on the Infiniband fabric. The downtime is expected to last until about 9:00 p.m. Jobs in the queue will remain in the queue, but jobs which cannot complete before the beginning of the downtime will not start.

Duration:Monday June 6th from 9:00 a.m. - approximately 9:00 p.m.

Scope: This downtime will only affect sanddunearch.


/scratch/serial status

Posted: May 21, 2007

/scratch/serial status

The /scratch/serial file server is mounted in read-only mode. We are waiting to receive a new set of hard drives that we hope will alleviate the problems we have had with this file system. The anticipated arrival of these drives in 2 weeks, that is, in the week of June 4th. After that, we will shut the file server down and replace the drives. If the drives arrive earlier, we may shut the file server down earlier.

As a result of the drives replacement, all the data on /scratch/serial will be lost. Those who have data that they want to keep on /scratch/serial, please, copy these files somewhere else as soon as possible. If you don't have enough disk space to copy these files to, please, contact problems@chpc.utah.edu. We will work with you to find a solution.

In the following few weeks, please, continue using /scratch/parallel and /scratch/serial-old as scratch servers on all Arches clusters, /scratch/da on Delicatearch and /scratch/mm on Marchingmen.

We encourage users to use /scratch/da on Delicatearch and /scratch/mm on Marchingmen to reduce the load on /scratch/parallel. We hope that putting /scratch/serial back online will improve the /scratch/parallel performance problems that we are currently experiencing.


Major CHPC Downtime: Thursday May 3, 2007, 5:00 p.m. - about Midnight

Posted: April 28, 2007

Major CHPC Downtime: Thursday May 3, 2007, 5:00 p.m. - about Midnight

Systems affected: All arches clusters down for maintenance for the entire downtime. Network connectivity to all servers will be unavailable from about 5pm to 7pm. After 7pm home directory space, except for the IGrid space, should be available if you have that space mounted on your desktop.

After this downtime the /scratch/serial-old, /scratch/da, and /scratch/mm systems will be mounted as READ ONLY and only on the interactive nodes. They will remain available in this manner for a few weeks. During this time users are encouraged to move data off of these systems such as /scratch/serial or /scratch/parallel.

All arches batch queues will be drained of running jobs leading up to the start of the downtime. Jobs waiting to be run will be kept and started after the downtime; however, if these waiting jobs write to /scratch/serial-old, /scratch/mm, or /scratch/da they will fail when started due to the change of these scratch file systems back to READ ONLY.

Duration:Thursday May 3rd from 5:00 p.m. - approximately Midnight

Scope: CHPC will be doing system maintenance on the arches clusters and on the CHPC networks.

Please note that /scratch is not backed up and is not intended for the storage of important or permanent data there. CHPC reserves the right to cleanup old files from all scratch space. The user is responsible to move important data to a permanent location such as a home department file server.


Major CHPC Downtime: Thursday May 3, 2007, 5:00 p.m. - about Midnight

Posted: April 28, 2007

Major CHPC Downtime: Thursday May 3, 2007, 5:00 p.m. - about Midnight

Systems affected: All arches clusters down for maintenance for the entire downtime. Network connectivity to all servers will be unavailable from about 5pm to 7pm. After 7pm home directory space, except for the IGrid space, should be available if you have that space mounted on your desktop.

After this downtime the /scratch/serial-old, /scratch/da, and /scratch/mm systems will be mounted as READ ONLY and only on the interactive nodes. They will remain available in this manner for a few weeks. During this time users are encouraged to move data off of these systems such as /scratch/serial or /scratch/parallel.

All arches batch queues will be drained of running jobs leading up to the start of the downtime. Jobs waiting to be run will be kept and started after the downtime; however, if these waiting jobs write to /scratch/serial-old, /scratch/mm, or /scratch/da they will die when started due to the change of these scratch file systems back to READ ONLY.

Duration:Thursday May 3rd from 5:00 p.m. - approximately Midnight

Scope: CHPC will be doing system maintenance on the arches clusters and on the CHPC networks.

Please note that /scratch is not backed up and is not intended for the storage of important or permanent data there. CHPC reserves the right to cleanup old files from all scratch space. The user is responsible to move important data to a permanent location such as a home department file server.


/scratch/serial is back up.

Posted: April 11, 2007

/scratch/serial back up.

The new /scratch/serial filesystem is back up. We have determined the cause of the problem and applied appropriate fix. Please, start migrating to this file server since /scratch/serial-old, /scratch/mm and /scratch/da will be made read only during the downtime at the end of the month.


/scratch/serial is currently down. Please use /scratch/parallel, /scratch/serial-old, /scratch/mm or /scratch/da.

Posted: April 5, 2007

/scratch/serial is currently down. Please use /scratch/parallel, /scratch/serial-old, /scratch/mm or /scratch/da.

The new /scratch/serial filesystem is down for troubleshooting by the vendor. We expect it to be available on Monday April 9th. Please use the PVFS space (/scratch/parallel). We have also temporarily mounted the "old /scratch/serial" at /scratch/serial-old temporarily on all of the clusters. We have also mounted /scratch/mm temporarily on marchingmen and /scratch/da temporarily on delicatearch. When we bring up the new /scratch/serial space we will be returning the old /scratch spaces to read-only. Please let us know if you have questions.


/scratch/serial is currently down. Please use /scratch/parallel, /scratch/serial-old, /scratch/mm or /scratch/da.

Posted: April 5, 2007

/scratch/serial is currently down. Please use /scratch/parallel, /scratch/serial-old, /scratch/mm or /scratch/da.

The new /scratch/serial filesystem is down for troubleshooting by the vendor. We expect it to be available on Monday April 9th. Please use the PVFS space (/scratch/parallel). We have also temporarily mounted the "old /scratch/serial" at /scratch/serial-old temporarily on all of the clusters. We have also mounted /scratch/mm temporarily on marchingmen and /scratch/da temporarily on delicatearch. When we bring up the new /scratch/serial space we will be returning the old /scratch spaces to read-only. Please let us know if you have questions.


Major CHPC Downtime: Thursday March 29, 2007, 4:00 p.m. - Midnight

Posted: March 14, 2007

Major CHPC Downtime: Thursday March 29, 2007, 4:00 p.m. - Midnight

Systems affected: All arches clusters down for maintenance. Network connectivity to all servers and from INSCC to campus will be intermittent from 5 -7. Access to home directories will be intermittent throughout the downtime. All jobs will be drained and queued jobs will be flushed.

Duration:Thursday March 29th from 4:00 p.m. - Midnight

Scope: CHPC will be doing system maintenance on the arches clusters, and on the CHPC networks. The /scratch filesystems will be arranged as follows:

  1. /scratch/serial-beta will be moved to /scratch/serial
  2. The "old" (current) /scratch/serial will be moved to /scratch/serial-old and will only be mounted READ ONLY on the interactive nodes and available at that path for a few weeks.
  3. /scratch/da, /scratch/mm and /scratch/serial-pio will also be mounted READ ONLY on the interactive nodes for a few weeks.
  4. Please note that data in /scratch are never backed up and you should not store important or permanent data there. CHPC reserves the right to cleanup old files from this space. Please move important data to a permanent location such as a home department file server. We plan to take the space recovered from the "old" /scratch/serial, /scratch/da, /scratch/mm and /scratch/serial-pio and add it to PVFS (/scratch/parallel) .


Major CHPC Downtime: Thursday March 29, 2007, 4:00 p.m. - Midnight

Posted: March 14, 2007

Major CHPC Downtime: Thursday March 29, 2007, 4:00 p.m. - Midnight

Systems affected: All arches clusters down for maintenance. Network connectivity to all servers and from INSCC to campus will be intermittent from 5 -7. Access to home directories will be intermittent throughout the downtime. All jobs will be drained and queued jobs will be flushed.

Duration:Thursday March 29th from 4:00 p.m. - Midnight

Scope: CHPC will be doing system maintenance on the arches clusters, and on the CHPC networks. The /scratch filesystems will be arranged as follows:

  1. /scratch/serial-beta will be moved to /scratch/serial
  2. The "old" (current) /scratch/serial will be moved to /scratch/serial-old and will only be mounted READ ONLY on the interactive nodes and available at that path for a few weeks.
  3. /scratch/da, /scratch/mm and /scratch/serial-pio will also be mounted READ ONLY on the interactive nodes for a few weeks.
  4. Please note that data in /scratch are never backed up and you should not store important or permanent data there. CHPC reserves the right to cleanup old files from this space. Please move important data to a permanent location such as a home department file server. We plan to take the space recovered from the "old" /scratch/serial, /scratch/da, /scratch/mm and /scratch/serial-pio and add it to PVFS (/scratch/parallel) .


Coming Soon - new arches cluster at CHPC

Posted: February 6, 2007

Coming Soon - new arches cluster at CHPC

CHPC is happy to announce that we will be adding an additional cluster to Arches, significantly increasing our current capacity. We started taking delivery of the new system February 1st and hope to have it operational by late February and open to users sometime in March. We plan to place it under allocation controls next calendar quarter (beginning April 1st, 2007.)

The cluster purchased from Dell, to be called "sanddunearch", will have 156 compute nodes (each node is an AMD dual/dual 2.4 GHZ, 8 GB memory) connected with InfiniBand high speed interconnect. The configuration and usage of the new system will be similar to the other Arches clusters, with the same scratch spaces etc.

This cluster is 100% University purchased. Therefore, the NIH block grants will not be honored on this cluster, and those PI's with NIH block grants only, who wish to run on this cluster, will need to submit an allocation proposal. If your allocation was authorized through the committee, you WILL be able to use it on the new cluster. New or renewal allocation requests are due March 1st and the form is available online at http://www.chpc.utah.edu/docs/forms/allocation.html.

Also beginning April 1st, 2007, CHPC will be changing the Service Unit (SU) metric from 1Ghz to 2Ghz. Details on Service Units and instructions for the allocation procedure are available at: http://www.chpc.utah.edu/docs/policies/allocation.html.

Please let us know if you have any questions on whether you will want to submit an allocation request or of course if you have any other questions.


Major CHPC Network Downtime: Thursday February 15th, 2007 5-8 p.m.

Posted: January 30, 2007

Major CHPC Network Downtime: Thursday February 15th, 2007 5-8 p.m.

Systems affected: All CHPC Network connectivity. There will be a reservation on ALL Arches Clusters preventing jobs from running during this timeframe.

Duration: Thursday February 15, 2007 from 5-8 p.m.

Scope: CHPC will updating to a new router. This should fix some of the networking problems we've had the past several weeks.


Major CHPC Network Downtime: Thursday February 15th, 2007 5-8 p.m.

Posted: January 30, 2007

Major CHPC Network Downtime: Thursday February 15th, 2007 5-8 p.m.

Systems affected: All CHPC Network connectivity. There will be a reservation on ALL Arches Clusters preventing jobs from running during this timeframe.

Duration: Thursday February 15, 2007 from 5-8 p.m.

Scope: CHPC will updating to a new router. This should fix some of the networking problems we've had the past several weeks.


Major CHPC Downtime: Thursday January 25th, 2007 5 p.m. - duration to be determined

Posted: January 16, 2007

Major CHPC Downtime: Thursday January 25th, 2007 5 p.m. - duration to be determined

Systems affected: All arches clusters down for maintenance. Network connectivity to all servers in INSCC will be off-line for part of the downtime.

Duration: Thursday January 25th, 2007 from 5:00 p.m. - duration not yet determined but will be posted when we have an estimate.

Scope: CHPC will be doing system maintenance on the arches clusters. PVFS will have maintenance performed requiring all files in /scratch/parallel to be purged. PLEASE migrate important data to another file system such as a home department file server prior to this downtime!!! Maintenance will also be performed on several of the CHPC switches.

More information will be posted as details are confirmed.


Scientific Supercomputing with Cells, GPUs, FPGAs, and Multi-Core CPUs: Wednesday, January 31, 2007

Posted: January 16, 2007


Scientific Supercomputing with Cells, GPUs, FPGAs, and Multi-Core CPUs: Wednesday, January 31, 2007


Location:
Center for High Performance Computing
INSCC Auditorium
University of Utah

About

Non-experts will gain an understanding of the fundamentals of computing with Cells, GPUs, and FPGAs, as well as an understanding of their potential benefits and limitations.

We hope to answer the following questions: Are any of these technologies ready to solve exploration geophysical and scientific problems? Does their potential warrant close monitoring of their progress in solving scientific problems?

For more information and registration: http://www.chpc.utah.edu/hybrid_computing/


Major CHPC Downtime: Thursday January 25th, 2007 5 p.m. - duration to be determined

Posted: January 16, 2007

Major CHPC Downtime: Thursday January 25th, 2007 5 p.m. - duration to be determined

Systems affected: All arches clusters down for maintenance. Network connectivity to all servers in INSCC will be off-line for part of the downtime.

Duration: Thursday January 25th, 2007 from 5:00 p.m. - duration not yet determined but will be posted when we have an estimate.

Scope: CHPC will be doing system maintenance on the arches clusters. PVFS will have maintenance performed requiring all files in /scratch/parallel to be purged. PLEASE migrate important data to another file system such as a home department file server prior to this downtime!!! Maintenance will also be performed on several of the CHPC switches.

More information will be posted as details are confirmed.


Utah State University hold first annual Agricultural Computing Research and Education Symposium: March 26-27, 2007.

Posted: January 9, 2007

News from USU: USU's Center for High Performance Computing (HPC@USU) will hold its first annual Agricultural Computing Research and Education Symposium on March 26-27, 2007.

We encourage all faculty, researchers, staff and students interested in high performance computing to attend. The symposium.s focus, .Challenges and Opportunities for High Performance Computing in Agriculture,. will feature national HPC leaders and keynote speaker Dr. Colien Hefferan, administrator of the Cooperative State Research, Education and Extension Service. Successful applications of high performance computing to agricultural research will be presented, and tutorials will be offered on scientific computing, visualization, and parallel computing.

For more information about the symposium, visit www.hpc.usu.edu

Sincerely,

Brent C. Miller, Ph.D.
Vice President for Research
Utah State University