To all HCPF Users,

Given interest in a new hardware purchase to extend the taki HPC, we would like to convene a meeting of those faculty interested in contributing to a community purchase in the coming weeks. The goal of this meeting would be to quickly review HW options and discuss topics ranging from desired machine specifications, purchase timeline, desired number of machines, and language we would like to distribute to departments and colleges to find other potential contributors ahead of the purchase.

Please email me at roy.prouty@umbc.edu by Wednesday February 7 at 1200 ET if you are interested in attending such a meeting.

Dear Colleagues and all users of HPCF,

TL;DR:

(i) Recording available from user meeting Fri 01/26/2024

(ii) Some notes of user meeting.

(iii) send publication info!

(i) We got the HyFlex technology in Webex to work and I recorded the meeting, which lasted just about exactly 1 hour. The recording is available in Panopto at

(ii) We had a great attendance both in-person and online, namely listed below.  I introduced myself, in case anyone did not know, and the two HPCF RAs Mostafa Cham and Ehsan Shakeri, see hpcf.umbc.edu under Contact Us.  We also appreciate that DoIT was represented by several staff, Roy Prouty, Andy Leeds, Todd Haddaway, and Damian Doyle.  My aim was to collect questions and concerns, depending on audience, and provide any information.  Damian Doyle very graciously actually came with two agenda items that essentially covered these goals. First, he provided an overview of the RCA Computing Taskforce from 2023, whose co-chair Zhibo Zhang was also present; the other co-chair was Karuna Joshi.  We thank all of them for their service. The ending point was that the leading VPs for Research Karl Steiner and for IT Jack Suess are in the process of implementing the new governance structure recommended and adopted by the taskforce.  Second, Damian shared that a purchase is intended in very near future, such as even within February; any interested faculty should contact Damian Doyle by e-mail (damian@umbc.edu).

Attendees: Matthias Gobbert, Mostafa Cham, Ehsan Shakeri, Roy Prouty, Damian Doyle, Zhibo Zhang, Tyler Josephson, Daniel Lobo, Joseph Bennett, Enrique Barbosa, Vijay Sharma, Andy Leeds, Todd Haddaway, Curtis Menyuk, Tim Finin, Lei Ren, Akash Vartak, Xuezhi Cang, Dimitry Sushon.

(iii) And please remember again, since it is really the most important, please send me the citation of publications that we can list on the HPCF webpage!

Matthias Gobbert, Math & Stat, gobbert@umbc.edu

Dear Colleagues and all users of HPCF,

TL;DR:

Reminder for user meeting with Webex link

(i) user meeting Fri 01/26/2024 10:00 in MP102/103,

(ii) send publication info!

Publications:

To start with item (ii), since it is really the most important, please always remember to send me the citation of publications that we can list on the HPCF webpage!

User meeting:

The purpose of this -- perfectly voluntary -- user meeting is just to offer help if anyone can profit from in-person communication or to get to know other users. It's just a personal offer from me.

The meeting is Friday 01/26/2024 at 10:00, which is the Friday before semester start. I will be in the physical location Math/Pysc 102 or 103, depending on crowd size, but the meeting will also be in Webex and recorded.

The Webex link is, and I hope I did this correctly,

If you want to contact me, e-mail gobbert@umbc.edu (do not just hit reply if you get this via mail.

Matthias Gobbert, Math & Stat, gobbert@umbc.edu

Dear Colleagues and all users of HPCF,

TL;DR:

(i) user meeting Fri 01/26/2024 10:00 in MP102/103,

(ii) send publication info!

(iii) hopefully MRI proposal this year

Publications:

To start with item (ii), since it is really the most important, please always remember to send me the citation of publications that we can list on the HPCF webpage!  This is invaluable to demonstrate our community's weight for our university administration -- all new by now really! -- and will be used in future NSF (or other) proposals. Include student theses (undergrad, MS, PhD), as well, please.

User meeting:

Item (i) calls for a general user meeting next Friday 01/26/2024 at 10:00, which is the Friday before semester start. I will be in the physical location Math/Pysc 102 or 103, depending on crowd size, but the meeting will also be in Webex and recorded.  The Webex link will be in the reminder posting on Friday morning.  If you plan to attend, maybe reply to this mail -- with indication of in-person or online -- so I can plan (and make doubly sure you receive the web link). In particular if you have to RSVP negative, feel free to include notes (see below).

HPCF proposals / purchases:

I really thank a user who reminded me by asking about purchase (and MRI) that I should inform you all better. Some of you may recall me saying -- and certainly sharing with the governance committee -- that we should participate in the November 2023 NSF MRI competition.  Since the NSF changed the time frame, this was the second MRI competition in the year after one in January 2023. Future ones will be in November of each year. In fact, I had been suggesting throughout the preceding year that we should hold back on any purchase, until after that MRI was submitted (so we do not look rich at the time of submission). Given I am aware of 10+ faculty who were at some point ready to contribute funding, my suggestion was not crazy, but did lose out on the energy of the willingness to contribute at the time.

To all taki users,

Starting at around 0345 this morning, portions of the taki cluster environment became unresponsive. System administrators have been able to bring the cluster back online. Any jobs running this morning were unfortunately lost.

Users can resume normal activity on the cluster. Please let us know if you notice any issues or if you have any questions/concerns by submitting a descriptive help request form found at the following link.

https://doit.umbc.edu/request-tracker-rt/doit-research-computing/ 

Roy Prouty

UMBC HPCF System Administrator

To all taki users,

Starting at around 0345 this morning, portions of the taki cluster environment became unresponsive. System administrators have been working to bring the cluster back online. Any jobs running this morning were unfortunately lost.

I will send another message when the cluster is again available.

Roy Prouty

UMBC HPCF System Administrator

To all taki users,

The taki downtime scheduled for today from 0800-1700 has completed a little behind schedule. We have updated the workload manager, SLURM, from major version 19 to major version 20. In the near future, we will be scheduling additional downtimes to iteratively upgrade through major version 23.

We have run some small jobs to confirm normal operating conditions, but will otherwise rely on user reports of unforeseen issues potentially related to these upgrades. Therefore, please let us know if you notice any issues or if you have any questions/concerns by submitting a descriptive help request form found at the following link.

https://doit.umbc.edu/request-tracker-rt/doit-research-computing/ 

The next scheduled downtime for the HPCF are listed below:

Roy Prouty

HPCF System Administrator, UMBC DoIT

Director, UMBC Observatory

PhD Candidate, CSEE

Hello all.

The Campus experienced a short power outage earlier tonight that unfortunately took the main data center offline for a short period of time.  

The Ada cluster is located in this data center and was affected by that outage.  We expect to have the cluster back online this morning (9/13/2023), and appreciate your patience while we work on doing so.

Thank you!

Tim Champ

Asst. Director of Research and Enterprise Computing

DoIT - EIS

To all taki users:

System Administrators will be taking the taki cluster environment offline Wednesday September 13 from 0800-1700. This is to upgrade the workload management software, SLURM. We expect (and hope) that the downtime will not require the entire nine hour duration. I will send an update toward the end of our downtime window alerting users to normal operation or the unlikely extension of the downtime window. 

As always, let us know if you have any questions or concerns by submitting a descriptive help request form found at the following link.

https://doit.umbc.edu/request-tracker-rt/doit-research-computing/ 

Roy Prouty

UMBC HPCF System Administrator

To all taki users,

The taki downtime scheduled for today from 0800-1700 has completed. We have updated the operating system and cluster management software stack. We have run some small jobs to confirm normal operating conditions, but will otherwise rely on user reports of unforeseen issues potentially related to the downtime.

Therefore, please let us know if you notice any issues or if you have any questions or concerns by submitting a descriptive help request form found at the following link.

https://doit.umbc.edu/request-tracker-rt/doit-research-computing/ 

The next few scheduled downtimes for the HPCF are listed below:

Tuesday August 29, 0800-1700 for upgrade ada OS and management software stack

*Wednesday September 13, 0800-1700 for upgrading taki SLURM

Tuesday September 19, 0800-1700 for upgrading ada SLURM

* In order to support classes currently running on the taki cluster, we’ve moved this downtime from Tuesday to Wednesday

Roy Prouty

HPCF System Administrator, UMBC DoIT

Director, UMBC Observatory

PhD Candidate, CSEE