24 posts tagged with "supercomputer"

If you haven't heard, we recently launched an on-campus S3-compatible object storage service called Kopah [docs] that is available to the research community at the University of Washington. Kopah is built on top of Ceph and is designed to be a low-cost, high-performance storage solution for data-intensive research.

While the deployment of Kopah was welcome news to those who are comfortable working with S3-compatible cloud solutions, we recognize some folks may be hesitant to give up their familiarity with POSIX file systems. If that sounds like you, we explored the use of JuiceFS, a distributed file system that provides a POSIX interface on top of object storage, as a potential solution.

Installation#

JuiceFS isn't installed by default so you will need to compile it yourself or download the pre-compiled binary from their release page.

As of January 2025 the latest version is 1.2.3 and you want the amd64 version if using from Klone. The command below will download and extract the binary to your current working directory.

I have to move it to a folder in my $PATH so I can run it from anywhere by just calling the binary. Your personal environment varies here.

Verify you can run JuiceFS.

Cool, now we can start using JuiceFS!

Standalone Mode#

There are two ways to run JuiceFS, standalone or distributed mode. This blog post explores the former. Standalone mode is meant to only present Kopah via POSIX on Klone. The key points being:

1. There is an active juicefs process required to run while you want to access it.
2. It is intended for you to run it only on the node you are running the process from.

If you wanted to run JuiceFS on multiple nodes or with multiple users then we will have another proof-of-concept with distributed mode in the future.

Create Filesystem#

JuiceFS separates the data (placed into S3 object storage) and the metadata, which is kept locally in a database. The command below will create the myjfs filesystem and store the metadata in a SQLite database called myjfs.db in the directory where the command is run. It puts the data itself into a Kopah bucket called npho-project.

You can rename the metadata file and the filesystem name to whatever you want (they don't have to match). The same goes for the bucket name on Kopah. However, I would strongly recommend having unique metadata file names that match the file system names for ease of tracking alongside the bucket name itself.

You can verify there is now a myjfs.db file in your current working directory. It's a SQLite database file that will store your file system meta data.

We can also verify the npho-project bucket was created on Kopah to store the data itself.

You should run juicefs format --help to view the full range of options and customize the parameters of your file system to your unique needs but just briefly:

• Encryption: When you create the file system and format it you can see it has encryption by default using AES256. You can over ride this using the --encrypt-algo flag if you prefer chacha20-rsa or you can use key file based encryption and provide your private key using the --encrypt-rsa-key flag.
• Compression: This is not enabled by default and there is a computational penalty for doing so if you want to access your files since it needs to be de or re encrypted on the fly.
• Quota: By default there is no block (set with --capacity in GiB units) or inode (set with --inodes files) quota enforced at the file system level. If you do not explicitly set this, it will be matched to whatever you get from Kopah. This is still useful for setting explicitly if you wanted to have multiple projects or file systems in JuiceFS that use the same Kopah account and have some level of separation.
• Trash: By default, files are not deleted immediately but moved to a trash folder similar to most desktop systems. This is set with the --trash-days flag and you can set it to 0 if you want files to be deleted immediately. The default here is 1 day after which the file is permanently deleted.

Mount Filesystem#

Running the command below will mount your newly created file system to the myproject folder in your home directory. It does not need to previously exist.

This process occurs in the background.

Use Filesystem#

Now with the file system mounted (at ~/myproject) you can use it like any other POSIX file system.

Remember, you won't be able to see it in the bucket because it is encrypted before being stored there.

Recover Deleted Files#

If you enabled the trash can option then you can recover files up until the permanent delete date.

First delete a file on the file system.

Verify the file is deleted. Go to recover it from the trash bin.

As you can see, we can recover files that are tracked by their delete date. You would need to copy the file back out to recover it.

Unmount Filesystem#

When you are done using the file system you can unmount it with the command below.

Remember, the file system is only accessible in standalone mode so long as a juicefs process is running. Since we ran it in the background you will need to explicitly unmount it.

Questions?#

Hopefully you found this proof-of-concept useful. If you have any questions for us, please reach out to the team by emailing help@uw.edu with Hyak somewhere in the subject or body. Thanks!

Our January maintenance is complete, and Klone is back in operation. The next maintenance is scheduled for Tuesday February 11, 2025 (AKA the 2nd Tuesday of the month).

Notable Updates#

• Slurm update: We updated to version 24.11.0. If any user software was built against the older Slurm libraries (e.g., openmpi), then users may experience errors, and it may be necessary to rebuild their software against the newer Slurm libraries.
• Compute node updates: The compute node OS images were updated to address any security patches in underlying core packages.
• Mathematica module installed using the license maintained by the UW Physics Department. Learn how to launch Mathematica on Hyak.
• MATLAB application has been added to Hyak Open OnDemand Beta. Learn how to launch MATLAB with Open OnDemand.

Upcoming Training#

Hyak: Open OnDemand – Open OnDemand (OOD) is an open-source web portal for HPC centers to provide users with an easy-to-use web interface to HPC clusters. For the last year, the Hyak team has been adding features to OOD. This workshop will demonstrate OOD's main features such as exploring the filesystem, composing jobs, and launching interactive applications like Jupyter. This workshop will be held in person 10am - 11:30am on Friday January 31, 2025 in CSE2 (Gates Center) Room 371 (3800 E Stevens Way NE, Seattle, WA 98195). Click here to learn more and register for this event.

Winter 2025 Office Hours#

• Wednesdays at 2pm on Zoom. Attendees need only register once and can attend any of the occurrences with the Zoom link that will arrive via email. Click here to Register for Zoom Office Hours.
• Thursdays at 2pm in person in eScience. (address: WRF Data Science Studio, UW Physics/Astronomy Tower, 6th Floor, 3910 15th Ave NE, Seattle, WA 98195). Click here to visit the eScience Office Hours page to see additional eScience office hours including AI/ML, R, Earth Data, and Python (not available to help with Homework).

If you would like to request 1 on 1 help, please send an email to help@uw.edu with "Hyak Office Hour" in the subject line to coordinate a meeting.

Opportunities#

The eScience Institute offers the annual Winter School to students and lecturers interested in developing basic skills and knowledge of the tools used in data science. Gaining literacy in topics such as Python, R, Jupyter, and reproducible environments can be beneficial beyond STEM, including areas like global or public health, public policy, social sciences, social work, international relations, and business management. Apply by January 24, 2025. Learn more!

Summer Internship Opportunity at Purdue - The Rosen Center for Advanced Computing (RCAC) is seeking students for Research Experience for Undergraduates (REU) paid internships for an 11-week onsite summer REU program. This program aims at developing the next generation workforce in advanced computing and cyberinfrastructure technologies. It offers students from diverse backgrounds the opportunity to gain the knowledge and skills necessary to build and support advanced research computing systems and scientific applications. As part of RCAC's decade long successful student apprentice program, the REU students will learn by doing, working on the National Science Foundation funded Anvil system in a team environment and mentored by cyberinfrastructure professionals. Open to undergraduate students from all backgrounds and undergraduate programs within and beyond Purdue. Each student will present their work to Purdue staff, faculty, students, collaborators, and researchers at the end of the program. Students may present at a national conference as part of the program. This onsite program at Purdue University in West Lafayette, Indiana runs from May through August. Learn more and apply NOW! Interviews start in January!

April 2-3, 2025, for AI Unlocked: Empowering Higher Education through Research and Discovery, a workshop designed for individuals across all disciplines and career stages in higher education. This workshop provides access to cutting-edge computational resources and expertise to researchers, students, faculty, and practitioners, especially those from Minority-Serving Institutions (MSIs). Hosted by the National Artificial Intelligence Research Resource (NAIRR) Pilot User Experience Working Group, this event introduces AI fundamentals, hands-on experience with pre-configured AI tasks, and guidance for tailoring AI models to your projects. Travel support is available for selected participants to the Westin Denver Downtown, and virtual attendance options are also offered. Apply now to advance your AI skills and collaboration in higher education. Apply by January 31, 2025. Acceptances will be notified by the end of February.

The University of Alaska Fairbanks (UAF) Alaska Center for Energy and Power (ACEP) summer internship is a 10-week program for students to gain hands-on research experience and skill development in the energy industry. Our program offers two internship strands: AUSI and REU. Regardless of strand, all interns will receive:

• A specific research project with 1:1 mentorship from an ACEP researcher
• Collaborative workspace at ACEP
• Travel to and from Alaska
• Field trips related to energy in Alaska

Applications due January 24, 2025.

If you have any questions about using Hyak, please start a help request by emailing help@uw.edu with "Hyak" in the subject line.

Happy Computing,

Hyak Team

Our December maintenance is complete. Thank you for your patience. The next maintenance will be Tuesday January 14, 2025.

Notable Updates#

• Updated node images, ensuring the security and behavior you expect from Hyak klone.
• We migrated LMOD modules and other tools and software stored in /mmfs1/sw/ to Solid State Drives (SSDs). Users may notice increased speed of modules.

Office Hours over Winter Break#

I will continue to hold Zoom office hours on Wednesdays at 2pm throughout December. Attendees need only register once and can attend any of the occurrences with the Zoom link that will arrive via email. Click here to Register for Zoom Office Hours.

December 12 will be the last in-person office hours for Fall term to be held at 2pm at the eScience Institute (address: WRF Data Science Studio, UW Physics/Astronomy Tower, 6th Floor, 3910 15th Ave NE, Seattle, WA 98195). In-person office hours will resume on Thursdays at 2pm starting January 9, 2025.

If you would like to request 1 on 1 help, please send an email to help@uw.edu with "Hyak Office Hour" in the subject line to coordinate a meeting.

Research Computing Club Officer Nominations#

The Research Computing Club (RCC) at UW is looking for nominations for Officer positions. The RCC provides essential computational resources to support students working on a wide range of research projects using high-performance computing through UW’s Hyak system and cloud platforms like AWS, Microsoft Azure, and Google Cloud. The RCC relies on student officers to continue providing these resources to the UW community and to organize community-driven events such as Hackathons and trainings. Officer positions are:

• President
• Director of AI/ML
• Director of Outreach
• Director of Hyak
• Director of Cloud Computing

Please consider nominating yourself if you are interested, or nominating someone you know who is interested, by filling out the form linked HERE.

External Opportunities#

DOE Computational Science Graduate Fellowship- Applications are being accepted through January 16, 2025 for the Department of Energy Computational Science Graduate Fellowship (DOE CSGF). Candidates must be U.S. citizens or lawful permanent residents who plan full-time, uninterrupted study toward a Ph.D. at an accredited U.S. university.

The DOE CSGF is composed of two computational science tracks. Eligible fellowship candidates should carefully review the criteria for both tracks prior to initiating the application process:

• The Science & Engineering Track accepts doctoral students engaged in computational science research with a science or engineering focus.
• The Mathematics/Computer Science Track accepts students pursuing research in broadly applicable methods and technology for high-performance computing (HPC) systems.

Students applying to the Mathematics/Computer Science Track must be pursuing a doctoral degree in applied mathematics, statistics, computer science, computer engineering or computational science — in one of these departments or their academic equivalent. A departmental exception is made for students whose research is focused on algorithms or software for quantum information systems and who are enrolled in a science or engineering field. In all cases, research must contribute to more effective use of emerging HPC systems.

DOE NNSA Stewardship Science Graduate Fellowship- Applications are being accepted through January 14, 2025 for the Department of Energy National Nuclear Security Administration Stewardship Science Graduate Fellowship (DOE NNSA SSGF). Candidates must be U.S. citizens who plan full-time, uninterrupted study toward a Ph.D. at an accredited U.S. university.

The DOE NNSA SSGF provides doctoral students with in-depth training in areas relevant to stewardship of the nation's nuclear stockpile: high energy density physics, nuclear science, or materials under extreme conditions. Senior undergraduates and first- or second-year doctoral students are eligible to apply.

DOE NNSA Laboratory Residency Graduate Fellowship- As part of its science and national security missions, the U.S. Department of Energy National Nuclear Security Administration (DOE NNSA) supports a spectrum of basic and applied research in science and engineering at the agency's national laboratories, at universities and in industry.

Because of its continuing needs, the NNSA seeks candidates who demonstrate the skills and potential to form the next generation of leaders in the following fields via the DOE NNSA LRGF program:

• Engineering and Applied Sciences
• Physics
• Materials
• Mathematics and Computational Science To meet its primary objective of encouraging the training of scientists, the DOE NNSA LRGF program provides financial support to talented students who accompany their study and research with practical work experience at one or more of the following DOE NNSA facilities: Lawrence Livermore National Laboratory, Los Alamos National Laboratory, Sandia National Laboratories or the Nevada National Security Site.

NSF ACCESS Student Training and Engagement Program (STEP) - Applications due February 15.

• STEP 1: 2-week introductory experience - Join a dynamic two-week experience in Miami, FL in May! Travel, accommodations, and stipend included. This program is designed to help you:

  • Become more aware of the many areas of interest within the field: web programming, high performance computing, cybersecurity, networking, real use of AI.
  • Determine where your interests are.
  • Discover what areas of study you should pursue in the future.
  • Experience high levels of direct interaction with diverse people with diverse skill sets.

• STEP-2: Full-time for the summer (in-person + virtual) - follows STEP-1 Travel and accommodations covered for in-person events as well as a stipend for participating plus includes travel to a national conference. This program is designed to help you:

  • Develop an intermediate understanding of one of the areas of interest within the field: web programming, high performance computing, cybersecurity, networking, real use of AI.
  • Provide opportunities for interactions with professionals in the field.

• STEP-3: Part-time during the school year, follows STEP-2. Continue interacting (virtually) with your team on projects on a part-time basis during the academic year. Participants will receive a stipend and travel to the annual Supercomputing conference (SC25 in St. Louis, MO, USA). This program is designed to help you:

  • Develop a deep understanding of one of the areas of interest within the field: web programming, high performance computing, cybersecurity, networking, real use of AI.
  • Provide further opportunities for interactions with professionals in the field.

Understand how a career in cyberinfrastructure could fit with your future plans!

External Trainings#

Looking for FREE training on cutting-edge digital technologies for industry? Find out more about the Training from the Hartree Centre HERE!

Whether you’re looking to get to grips with the basics or searching for new tools and techniques to apply, Hartree Training supports both self-directed online learning as well as face-to-face practical sessions with badge certification available for you to share your new skills with your network. Currently offering courses covering a range of advanced digital technologies including:

• Data Science
• Artificial Intelligence and Modelling
• High Performance and Exascale Computing
• Software Engineering
• Emerging Technologies

To enroll, create a FREE account or log in. Once you have created an account, you can log in and sign up for as many of our courses as you like.

NVIDIA Professional Workshops and Certificates Blog post HERE discusses NVIDIA Professional Workshops ($3000 each) and Certificate Exams ($400 each) in AI Infrastructure and AI Operations. Includes links to workshops and certificate programs in Generative AI LLMs and Generative AI Multimodal.

Happy Computing,

Hyak Team

Our November maintenance is complete. Thank you for your patience while we make package updates to node images, ensuring the security and behavior you expect from Hyak klone.

The next maintenance will be Tuesday December 10, 2024.

Notable Updates#

• We updated the GPU drivers to version 565.57.01 to patch security vulnerabilities CVE-2024-0117 through CVE-2024-0121 and CVE-2024-0126.
• We migrated user Home directories to Solid State Drives (SSDs). Users may notice increased speed of software stored in Home directories.

Upcoming Training#

Hyak: Scheduling jobs with Slurm workshop is on Thursday, November 14th, from 10 a.m. to 12 p.m. in the WRF Data Science Studio (UW Physics/Astronomy Tower, 6th Floor; 3910 15th Ave NE, Seattle, WA 98195), and will cover Hyak’s job scheduler Slurm, interactive jobs, batch jobs, and array jobs. REGISTER HERE!

Hyak: Introduction to Deep Learning workshop is on Tuesday, December 3rd, from 10 to 11:50 a.m. in CSE2 (Gates Center) Room 371. Participants will start with a computer vision example, training a model on a sample dataset, and then learn how to execute the training process on Hyak. REGISTER HERE!

Office Hours#

Zoom office hours will be held on Wednesdays at 2pm. Attendees need only register once and can attend any of the occurrences with the Zoom link that will arrive via email. Click here to Register for Zoom Office Hours

In-person office hours will be held on Thursdays at 2pm at the eScience Institute (address: WRF Data Science Studio, UW Physics/Astronomy Tower, 6th Floor, 3910 15th Ave NE, Seattle, WA 98195).

The Research Computing Club will be holding office hours fall term. In-person office hours will be held at the eScience Institute, WRF Data Science Studio.

If you would like to request 1 on 1 help, please send a ticket to help@uw.edu with "Hyak Office Hour" in the subject line to coordinate a meeting.

Our October maintenance is complete. Thank you for your patience while we make package updates to node images, ensuring the security and behavior you expect from Hyak klone.

The next maintenance will be Tuesday November 12, 2024.

New Tools Documentation#

Our research computing interns have been hard at work adding documentation for new user tools that might help optimize your research computing. Click the links below to review the docs for these tools.

Squash Fuse - SquashFS packages multiple small files into a single read-only, compressed filesystem, reducing metadata calls and improving performance. This minimizes server load, enhancing throughput and efficiency in handling storage requests.

Use case on Hyak: In HPC, datasets often consist of numerous small files, which can lead to performance bottlenecks due to excessive metadata operations. By utilizing SquashFS, HPC applications can significantly reduce metadata overhead, improving data access speeds and enhancing overall system performance, particularly in large-scale distributed storage systems.

Checkpointing with DMTCP - DMTCP is a tool to transparently checkpoint and restart jobs, saving it to disk to be resumed at a later time. It requires no changes to application code, allowing easy use. Using DMTCP with your code allows checkpointing at regular intervals so if your job is pre-empted or reaches the time limit, it will resume from its last checkpoint.

Use case on Hyak: DMTCP offers a solution for folks who would like to use Hyak's ckpt partitions, but have jobs that exceed the ckpt time limits of 5 hours for CPU-noly jobs and 8 hours for GPU-only jobs. Checkpointing with DMTCP facilitates efficient use of ckpt resources, allowing higher throughput for your jobs.

Tools for Kopah Storage Users - We have installed Command Line Interface tools like s3cmd and s5cmd on klone and provide insctructions for using Python library boto3 for Kopah interaction and retreival to build Kopah S3 storage usage into your research computing applciations on Hyak.

If you have any issue using these tools, please open a ticket by emailing help@uw.edu with "Hyak" in the subject line. We appreciate any feedback about how to improve ease of use for tools presented in our documentation.

Upcoming Training#

We have planned 2 Hyak-specific trainings for this Fall (more to come, stay tuned). These trainings will be held in person and will not be recorded since recorded materials are already publicly accessible. Capacity is limited, follow the links below to register today to guarantee your spot.

Hyak: Containers are your friend - Monday October 28 10am - 12pm.

Hyak: Scheduling Jobs with Slurm - Thursday November 14 10am - 12pm

In the first hour and a half, we will go over content. The last 30 minutes will be reserved for questions.

Location: eScience Classroom; WRF Data Science Studio, UW Physics/Astronomy Tower, 6th Floor; 3910 15th Ave NE, Seattle, WA 98195

Keep an eye on your inbox for updates about additional trainings this fall.

Fall Office Hours#

Hyak HPC Staff Scientist and Facilitator, Kristen Finch, will be holding office hours fall term.

Zoom office hours will be held on Wednesdays at 2pm. Attendees need only register once and can attend any of the occurrences with the Zoom link that will arrive via email.

Click here to Register for Zoom Office Hours

In-person office hours will be held on Thursdays at 2pm at the eScience Institute (address: WRF Data Science Studio, UW Physics/Astronomy Tower, 6th Floor, 3910 15th Ave NE, Seattle, WA 98195). Click here to RSVP for in-person Office Hours.

Click here to visit the eScience Office Hours page to see additional eScience office hours including AI/ML, R, Earth Data, and Python (not available to help with Homework).

The Research Computing Club will be holding office hours fall term. In-person office hours will be held at the eScience Institute (address: WRF Data Science Studio, UW Physics/Astronomy Tower, 6th Floor, 3910 15th Ave NE, Seattle, WA 98195).

If you would like to request 1 on 1 help, please send a ticket to help@uw.edu with "Hyak Office Hour" in the subject line to coordinate a meeting with Kristen.

Please don't hesitate to reach out to the Hyak team with issues and feedback by opening a tickey by emailing help@uw.edu with "Hyak" in the subject.

Have a great October!

Happy computing,

Hyak Team

Thanks again for your patience with our monthly scheduled maintenance. During this maintenance session, we were able to provide package updates to node images to ensure compliance with the latest operating system level security fixes and performance optimizations. Of note, the Nvidia GPU driver on Klone has been updated to the latest production datacenter release, version 560.35.03.

The next maintenance will be Tuesday October 8, 2024.

AMD Math libraries#

In June we announced the addition of AMD Nodes and Slices to klone which make up our generation 2 or g2 collection of resources. Click here to read more about the difference between our g1 and g2 resources. During August, we installed a new AMD compiler suite, AOCC, along with specialized math libraries like AOCL, OpenBLAS, and ScaLAPACK as modules to make the most of this upgade. The new modules are useful on all partitions. The AOCC and AOCL modules are particularly relevant for partitions cpu-g2, cpu-g2-mem2x, and ckpt-g2. These tools are designed to optimize performance on AMD processors, speeding up complex mathematical computations. Whether you're working on simulations, data analysis, or any number-crunching tasks, these libraries may help ensure you get faster, more efficient results. If you're looking to boost your workflow, it's worth exploring how these libraries can benefit your projects. Here are the names of the new modules:

In our benchmarking tests, performance of these libraries was similar on g1 and g2 CPUs for each math library, regardless of architecture. The best library performer on AMD CPUs is AOCC+AOCL, and for Intel CPUs it’s OpenBLAS+ScaLAPACK:

Fall Office Hours#

Hyak HPC Staff Scientist and Facilitator, Kristen Finch, will be holding office hours fall term. Zoom office hours will be held on Wednesdays at 2pm. Attendees need only register once and can attend any of the occurrences with the Zoom link that will arrive via email.

Click here to Register

In-person office hours will be held on Thursdays at 2pm at the eScience Institute (address: WRF Data Science Studio, UW Physics/Astronomy Tower, 6th Floor, 3910 15th Ave NE, Seattle, WA 98195). Click here to RSVP for in-person Office Hours.

Click here to visit the eScience Office Hours page to see additional eScience office hours including AI/ML, R, Earth Data, and Python (not available to help with Homework).

If you would like to request 1 on 1 help, please send a ticket to help@uw.edu with "Hyak Office Hour" in the subject line to coordinate a meeting with Kristen.

August 2024 Training Videos#

In case you missed it, we recorded the August 2024 Wednesday training sessions and posted them on the UW-IT YouTube channel under the playlist, "Hyak Training." Here are the links:

• Hyak: R and Rstudio (August 28, 2024) recording.
• Hyak: Containers are your friend (August 21, 2024) recording
• Hyak: Slurm and Advanced Slurm (August 14, 2024) recording
• Hyak Basics: Linux CLI workshop (August 7, 2024) recording

Keep an eye on your indox for updates about our Fall training schedule; training sessions are currently TBA. Trainings will be announced via the Hyak mailing list, click here to join the mailing list.

Thanks again for your patience with our monthly scheduled maintenance. During this maintenance session, we were able to provide package updates to node images to ensure compliance with the latest operating system level security fixes and performance optimizations.

The next maintenance will be Tuesday September 10, 2024.

New self-hosted S3 storage option: KOPAH#

We are happy to announce the preview launch of our self-hosted S3 storage called KOPAH. S3 storage is a solution for securely storing and managing large amounts of data, whether for personal use or research computing. It works like an online storage locker where you store can files of any size, accessible from anywhere with an internet connection. For researchers and those involved in data-driven studies, it provides a reliable and scalable platform to store, access, and analyze large datasets, supporting high-performance computing tasks and complex workflows.

S3 uses buckets as containers to store data, where each bucket can hold 100,000,000 objects, which are the actual files or data you store. Each object within a bucket is identified by a unique key, making it easy to organize and retrieve your data efficiently. Public links can be generated for KOPAH objects so that users can share buckets and objects with collaborators.

Click here to learn more about KOPAH S3.

Who should use KOPAH?#

KOPAH is a storage solution for anyone. Just like other storage options out there, you can upload, download, and view your storage bucket with specialized tools and share your data via the internet. For Hyak users, KOPAH provides another storage option for research computing. It is more affordable than /gscratch storage and can be used for active research computing with a few added steps for retreiving stored data prior to a job.

Test Users Wanted#

Prior to September, we are inviting test users to try KOPAH and provide feedback about their experience. If you are interested in becoming a KOPAH test user, please each help@uw.edu with Hyak or KOPAH in the subject line.

Requirements:

1. While we will not charge for the service until September 1, to sign up as a test user, we require a budget number and worktag. If the service doesn't work for you, you can cancel before September.
2. We will ask for a name for the account. If your groups has an existing account on Hyak, klone /gscratch, it makes sense of the names to match across services.
3. Please be ready to respond with your feedback about the service.

Opportunities#

PhD students should check out this opportunity for funding from NVIDIA: Graduate Research Fellowship Program

Questions? If you have any questions for us, please reach out to the team by emailing help@uw.edu with Hyak in the subject line.

Thanks again for your patience with our monthly scheduled maintenance. During this maintenance session, we were able to provide package updates to node images to ensure compliance with the latest operating system level security fixes and performance optimizations.

The next maintenance will be Tuesday August 13, 2024.

Questions? If you have any questions for us, please reach out to the team by emailing help@uw.edu with Hyak in the subject line.

Thanks again for your patience with our monthly scheduled maintenance. This month, we deployed new node resources that were purchased by various UW Researchers from across campus. These nodes are a little different, so we wanted to bring your attention to them and provide guidance on their use when they are idle with the checkpoint partition.

New G2 Nodes#

A new class of nodes have been deployed on klone which we are calling g2 because they are the second generation of nodes, and we will retroactively refer to the first generation nodes as g1. g2 CPU nodes feature AMD EPYC 9000-series 'Genoa' processors, and new GPU nodes featuring either NVIDIA L40 or L40S GPUs (with H100 GPUs possibly becoming available in the future). These nodes will join our community resources that can be used when idle (ckpt) under the new partitions:

• ckpt-g2 for scheduling jobs on g2 nodes only.

• ckpt-all for scheduling jobs on either g1 or g2 nodes.

• ckpt will now schedule jobs on g1 nodes only.

  Please review our documentation HERE for specific instructions for accessing these resources. Additionally, please see the blog post HERE where we discuss additional considerations for their usage.

To accompany the new g2 node deployments, we are providing a new Open MPI module (ompi/4.1.6-2), which is now the default module when module load ompi is executed. Previous OpenMPI modules will cause errors if used with the AMD processors on the new g2 nodes due to how the software was compiled. ompi/4.1.6-2 (and any openmpi module versions we provide in the future) are compiled to support both Intel and AMD processors. If your MPI jobs are submitted to a partition that includes g2 nodes, you should use module load ompi to use the new module by default, or explicitly load ompi/4.1.6-2 (or a newer version in the future) via module load ompi/4.1.6-2.

If you have compiled software on g1 nodes, you should test them on g2 nodes before bulk submitting jobs to partitions with g2 nodes (i.e., ckpt-g2 and ckpt-all), as they may or may not function properly depending on exactly how they were compiled.

Student Opportunities#

In addition, we have two student opportunities to bring to your attention.

Job Opportunity: The Research Computing (RC) team at the University of Washington (UW) is looking for a student intern to spearhead projects that could involve: (1) the development of new tools and software, (2) research computing documentation and user tutorials, or (3) improvements to public-facing service catalog descriptions and service requests. Hyak is an ecosystem of high-performance computing (HPC) resources and supporting infrastructure available to UW researchers, students, and associated members of the UW community. Our team administers and maintains Hyak as well as provides support to Hyak users. Our intern will be given the choice of projects that fit their interest and experience while filling a need for the UW RC community. This role will provide students with valuable hands-on experiences that enhance academic and professional growth.

The position pays $19.97-21.50 per hour depending on experience with a maximum of 20 hours per week (Fall, Winter, and Spring) and a maximum of 40 hours allowed during summer quarter. How to apply: Please apply by emailing: 1) a current resume and 2) a cover letter detailing your qualifications, experience, and interest in the position to me, Kristen Finch (UWNetID: finchkn). Due to the volume of applications, we regret that we are unable to respond to every applicant or provide feedback.

Minimum Qualifications:

• Student interns must hold at least a 2nd year standing if an undergraduate student.
• Student interns must be able to access the internet.
• Student interns must be able to demonstrate an ability to work independently on their selected project/s and expect to handle challenges by consulting software manuals and publicly available resources.

We encourage applicants that:

• meet the minimum qualifications.
• have an interest in website accessibility and curation.
• have experience in research computing and Hyak specifically. This could include experience in any of the following: 1) command-line interface in a Linux environment, 2) Slurm job scheduler, 3) python, 4) shell scripting.
• have an interest in computing systems administration.
• have an interest in developing accessible computing-focused tutorials for the Hyak user community.

Conference Opportunity: The 2024 NSF Cybersecurity Summit program committee is now accepting applications to the Student Program. This year’s summit will be held October 7th-10th at Carnegie Mellon University in Pittsburgh, PA. Both undergraduate and graduate students may apply. No specific major or course of study is required, as long as the student is interested in learning and applying cybersecurity innovations to scientific endeavors. Selected applicants will receive invitations from the Program Committee to attend the Summit in-person. Attendance includes your participation in a poster session. The deadline for applications is Friday June 28th at 12 am CDT, with notification of acceptance to be sent by Monday July 29th. Click Here to Apply

Our next scheduled maintenance will be Tuesday July, 9, 2024.

Questions? If you have any questions for us, please reach out to the team by emailing help@uw.edu with Hyak in the subject line. Student intern applications sent to help@uw.edu will not be considered. Email applications to finchkn at uw.edu

Hello Hyak Community,

Thanks again for your patience with our monthly scheduled maintenance, there are some notable improvements we implemented today.

klone node image: Over the past few weeks, you may have noticed some klone instability. This was a result of some behind the scenes storage upgrades that inadvertently introduced wider impacts to the existing cluster automation. At the time, we introduced a temporary fix to get the cluster back online but with today’s maintenance we implemented a more comprehensive fix.

Infiniband firmware: The klone cluster is built on the infiniband HPC interconnect for node-to-node communication. While klone originally launched with the HDR generation of infiniband, we have since upgraded mid-klone to have a HDR-NDR hybrid interconnect. NDR infiniband is required to support the latest compute slices we offer. We updated the firmware on our NDR switches following vendor recommendations for increased stability.

Apptainer on MOX: Apptainer (formerly Singularity) is the root-less containerization solution we provide on both Hyak clusters. Apptainer version 1.3.1 was deployed on both klone and MOX. As a reminder, on klone Apptainer is accessed through a module and is only available on compute nodes after module load apptainer. On MOX, Apptainer is default software and can be accessed with Apptainer commands directly after starting an interactive job for example, apptainer --version.

Training Opportunities: COMPLECS (San Diego Supercomputer) is hosting an Intermediate Linux Shell Scripting online workshop on Thursday May, 16 at 11:00 am Pacific Time. Register here.

Our next scheduled maintenance will be Tuesday June, 11, 2024. Stay informed by joining our mailing list. Sign up here.

Questions? If you have any questions for us, please reach out to the team by emailing help@uw.edu with Hyak in the subject line.

Thank you for your patience this month while there was more scheduled downtime than usual to allow for electrical reconfiguration work in the UW Tower data center. We appreciate how disruptive this work has been in recent weeks. Please keep in mind that this work by the data center team has been critical in allowing the facility to increase available power to the cluster to provide future growth capacity, which was limiting deployment of new equipment in recent months.

The Hyak team was able to use the interruption to implement the following changes:

• Increase in checkpoint (--partition=ckpt) runtime for GPU jobs from 4-5 hours to 8-9 hours (pre-emption for requeuing will still occur subject to cluster utilization). Please see the updated documentation page for information about using idle resources.
• The NVIDIA driver has been updated for all GPUs.

Our next scheduled maintenance will be Tuesday May 14, 2024.

Training Opportunities#

Follow NSF ACCESS Training and Events posting HERE to find online webinars about containers, parallel computing, using GPUs, and more from HPC providers around the USA.

Questions? If you have any questions for us, please reach out to the team by emailing help@uw.edu with Hyak in the subject line.

The Research Working Group of the UW AI Task Force would like faculty and research staff input on the needs and challenges of using AI in research at UW across a broad spectrum of disciplines. Please help by responding to a short survey at: https://forms.gle/mZrV3aCgJYNNBV6j8. Responses by April 25 would be most helpful, but the survey will remain open until April 30.

Thank you in advance for your time,

Hyak Team

It has come to our attention that the default configuration of Miniconda and conda environments in the user's home directory leads to hitting storage limitations and the dreaded error Disk quota exceeded. We thought we would take some time to guide users in configuring their conda environment directories and package caches to avoid this error and proceed with their research computing.

Conda's config#

Software is usually accompanied by a configuration file (aka "config file") or a text file used to store configuration data for software applications. It typically contains parameters and settings that dictate how the software behaves and interacts it's environment. Familiarity with config files allows for efficient troubleshooting, optimization, and adaptation of software to specific environments, like Hyak's shared HPC environment, enhancing overall usability and performance. Conda's config file .condarc, is customizable and lets you determine where packages and environments are stored by conda.

Understanding your Conda#

First let's take a look at your conda settings. The conda info command provides information about the current conda installation and its configuration.

The output shoudl look something like this if you have installed Miniconda3.

The paths shown above will show your username in place of UWNetID. Notice the highlighted lines above showing the absolute path to your config file in your home directory (e.g., /mmfs1/home/UWNetID/.condarc), the directory designated for your package cache (e.g., /mmfs1/home/UWNetID/conda_pkgs), and the directory/ies designated for your environments (e.g., /mmfs1/home/UWNetID/miniconda3/envs). Conda designates directories for your package cache and your environments by default, but under Hyak, your home directory has a 10G storage limit, which can quickly be maxed out by package tarballs and their contents. We can change the location for your package cache and your environments to avoid this.

Configuring your package cache and envs directories#

If you don't have a .condarc in your home directory, you can create and edit it with a hyak preloaded editor like nano or vim. Here we will used nano.

Edit OR ADD the highlighted lines to your .condarc to designate directories with higher storage quotas for our envs_dirs and pkgs_dirs. In this exercise, we will assign our envs_dirs and pkgs_dirs directories to directories in /gscratch/scrubbed/ where we have more storage, although remember scrubbed storage is temporary and files are deleted automatically after 21 days if the timestamps are not updated. Alternatively, your lab/research group might have another directory in /gscratch/ that can be used.

Here is what your edited .condarc should look like.

In addition to designating the directories, please include always_copy: true, which is required on the Hyak filesystem for configuring your conda in this way.

After .condarc is edited, we can use conda info with grep to see if our changes have been incorporated.

The result should be something like

And for the environments directory

Result

Cleaning up disk storage#

After you have reset the package cache and environment directories with your conda config file, you can delete the previous directories to free up storage. Before doing that, you can monitor how much storage was being occupied by each item in your home directory with the command du -h --max-depth=1. Remove directories previously used as cache and envs_dir recursively with rm -r. The following is an example of monitoring storage and removing directories.

Below is an example output from the du -h --max-depth 1 command

Storage can also be managed by cleaning up package cache periodically. Get rid of the large-storage tar archives after your conda packages have been installed with conda clean --all.

Lastly, regular maintenance of conda environments is crucial for keeping disk usage in check. Review you list of conda environments with conda env list and remove unused environments using the conda remove --name ENV_NAME --all command. Consider creating lightweight environments by installing only necessary packages to conserve disk space. For example, create an environment for each project (project1_env) rather than an environment for all projects combined (myenv).

Disk quota STILL exceeded#

Be aware that many software packages are configured similarly to conda. Explore the documentation of your software to locate the configuration file and anticipate where storage limitations might become an issue. In some cases, you may need to edit or create a config file for the software to use. pip and R are two other common offenders ballooning the disk storage in your home directory.

Configuring PIP#

If you are installing with pip, you might have a pip cache in ~/.cache/pip. Let's locate your the pip config file location under variant "global." You might have to activate a previously built conda environment to do this. For this exercise we will use an environment called project1_env.

The message "will try loading" rather than listing the config file pip.conf means that a pip config file has not been created. We will create our config file and set our pip cache. Create a directory in your home directory (e.g.,/mmfs1/home/UWNetID/.pip) to hold your pip config file and create a file called pip.conf with the touch command. Remember to also create the new directory for your new pip cache if you haven't yet.

Open pip.conf with nano or vim and add the following lines to designate the location of your pip cache.

Check that your pip cache has been designated.

Configuring R#

We previously covered this in our documentation. Edit or create a config file called .Renviron in your home directory. Use nano or vim to designate the location of your R package libraries. The contents of the file should be something like the following example.

The directory designated by R_LIBS will be where R installs your package libraries.

I'm still stuck#

Please reach out to us by emailing help@uw.edu with "hyak" in the subject line to open a help ticket.

For our March maintenance we had some notable changes we wanted to share with the community.

Login Node#

Over the last several months the login node has been crashing on occasion. We have been monitoring and dissecting the kernel dumps from each crash and this behavior seems to be highly correlated with VS Code Remote-SSH extension activity. To prevent node instability, we have upgraded the storage drivers to the latest version. If you are a VS Code user and connect to klone via Remote-SSH, we have some recommendations to help limit the possibility that your work would cause system instability on the login node.

Responsible Usage of VS Code Extension Remote-SSH#

While developing your code with connectivity to the server is a great usage of our services, connecting directly to the login node via the Remote-SSH extension will result in VS Code server processes running silently in the background and leading to node instability. As a reminder, we prohibit users running processes on the login node.

1. Check which processes are running on the login node, especially if you have been receiving klone usage violations when you are not aware of jobs running. Look for vscode-server among the listed processes.

   $ ps aux | grep UWNetID

   Copy

2. If you need to develop your code with connectivity to VS Code, use a ProxyJump to open a connection directly to a compute node. Step 1 documentation. and then use the Remote-SSH extension to connect to that node through VS Code on your local machine, preserving the login node for the rest of the community. Step 2 documentation.

3. Lastly, VS Code’s high usage is due to it silently installing its built in features into the user's home directory ~/.vscode on klone enabling intelligent autocomplete features. This is a well known issue, and there is a solution that involves disabling the @builtin TypeScript plugin from the VS Code on your local machine. Here is a link to a blog post about the issue and the super-easy solution. Disabling @builtin TypeScript will reduce your usage of the shared resources and avoid problems.

In addition to the upgrade of the storage driver, we performed updates to security packages.

Training Opportunities#

We wanted to make you aware of two training opportunities with the San Diego Supercomputer Center. If you are interested in picking up some additional skills and experience in HPC, check this blog post.

Questions?#

If you have any questions for us, please reach out to the team by emailing help@uw.edu with Hyak in the subject line.

Hello Hyak community! We have a few notable announcements regarding this month’s maintenance. If the hyak-users mailing list e-mail didn’t fully satisfy your curiosity, hopefully this expanded version will answer any lingering questions.

GPUs#

• Software: The GPU driver was upgraded to the latest stable version (545.29.06). The latest CUDA 12.3.2 is also now provided as a module. You are also encouraged to explore the use of container (i.e., Apptainer) based workflows, which bundle various versions of CUDA with your software of interest (e.g., PyTorch) over at NGC. NOTE: Be sure to pass the --nv flag to Apptainer when working with GPUs.

• Hardware: The Hyak team has also begun the early deployments of our first Genoa-Ada GPU nodes. These are cutting-edge NVIDIA L40-based GPUs (code named “Ada”) running on the latest AMD processors (code named “Genoa”) with 64 GPUs released to their groups two weeks ago and an additional 16 GPUs to be released later this week. These new resources are not currently part of the checkpoint partition but we will be releasing guidance on making use of idle resources here over the coming weeks directly to the Hyak user documentation as we receive feedback from these initial researchers.

Storage#

• Performance Upgrade: In recent weeks, AI/ML workloads have been increasingly stressing the primary storage on klone (i.e., "gscratch"). Part of this was attributed to the run up to the International Conference for Machine Learning (ICML) 2024 full paper deadline on Friday, February 2. However, it also reflects a broader trend in the increasing demands of data-intensive research. The IO profile was so heavy at times that our systems automation throttled the checkpoint capacity to near 0 in order to keep storage performance up and prioritize general cluster navigation and contributed resources. We have an internal tool called iopsaver that automatically reduces IOPS by intelligently requeuing checkpoint jobs generating the highest IOPS while concurrently limiting the number of total active checkpoint jobs until the overall storage is within its operating capacity. At times over the past few weeks you may have noticed that iopsaver had reduced the checkpoint job capacity to near 0 to maintain overall storage usability.

  During today’s maintenance, we have upgraded the memory on existing storage servers so that we could enable Local Read-Only Cache (LROC) although we don’t anticipate it will be live until tomorrow. Once enabled, LROC allows the storage cluster to make use of a previously idle SSD capacity to cache frequently accessed files on this more performant storage tier medium. We expect LROC to make a big difference as during this period of the last several weeks, the majority of the recent IO bottlenecking was attributed to a high volume of read operations. As always, we will continue to monitor developments and adjust our policies and solutions accordingly to benefit the most researchers and users of Hyak.

• Scrubbed Policy: In the recent past this space has filled up. As a reminder, this is a free-for-all space and a communal resource for when you have data you only need to temporarily burst out into past your usual allocations from your other group affiliations. To ensure greater equity among its use, we have instituted a 10TB and 10M files limit for each user in scrubbed. This impacts <1% of users as only a handful of users were using an amount of quota from scrubbed >10TB.

Questions?#

Hopefully you found these extra details informative. If you have any questions for us, please reach out to the team by emailing help@uw.edu with Hyak somewhere in the subject or body. Thanks!

We’ve made an update to our storage accounting tool, hyakstorage, and with this update we are also phasing out usage_report.txt. That text file contained minimally-parsed internal metrics of the storage cluster, and we found it caused as many questions as it answered. Moving forward, the hyakstorage tool will display only the four relevant pieces of information for each fileset you query: storage space used vs. the storage space limit, and current amount of files (inodes) vs. maximum number of files.

The default operation–running hyakstorage with no arguments–will show your home directory & the gscratch directories you have access to, and it will only show the fileset totals & your contributions.

You can also specify which filesets you want to view, in a few different ways: you can use the flag --home to show your home directory, --gscratch to show your gscratch directories, and --contrib to show your group’s contrib directories. You can also specify an exact gscratch directory with the group name (e.g. hyakstorage stf), contrib directory (e.g. hyakstorage stf-src), or full path to a fileset (e.g. hyakstorage /mmfs1/gscratch/stf).

If you want more detailed metrics, you can use the flags --show-user or --show-group to break down the fileset totals by individual users or groups. Those detailed metrics can be sorted by space with --by-disk (the default) or by files with --by-files.

See also:

• gscratch documentation and
• hyakstorage documentation.

The Hyak community has grown substantially over the past year, including the administrative teams that work with us to support the service. Some terms (e.g., nodes, servers) have been loosely used in communication, but have specific meanings for different backend teams. Beginning today, we are harmonizing all the terms to alleviate any confusion between the different teams supporting Hyak and our end users. This is only a clarification of language: there is no change to how Hyak operates.

At the physical layer we have nodes or servers: the smallest individual physical units of the HPC cluster. These are what we, the Hyak engineering team, purchase from our vendor partners. Historically, a physical node or server was what a lab would purchase to join the cluster. However, since Hyak’s inception, resource density has increased to such an extent–servers with hundreds of CPU cores, hundreds of gigabytes of RAM, multiple graphics cards, etc.–it no longer made sense to require labs to purchase an entire physical node.

Once we crossed a certain threshold, we began to offer labs an amount of computing resources–a specific number of CPU cores, amount of RAM, number of GPUs, etc.–rather than discrete servers, but we kept the node nomenclature. For a while, when labs purchased a node, it no longer meant they were purchasing a server, even though those words are identical in many computing contexts. From today forward, we are restoring those terms to their original meanings for Hyak: one node is one physical server.

When labs join Hyak, they will not purchase a physical node, they will purchase a slice. A slice represents an amount of on-demand compute capacity–CPUs, RAM, GPUs, etc. Again, this is only a terminology clarification: Hyak has operated this way for a while. One of the benefits of this model is that slices–representing resources, not specific, physical pieces of hardware–make resource scheduling considerably easier for our cluster's scheduling software, Slurm. This efficiency is returned to the entire community both as depth of the checkpoint partition’s resources, and as faster scheduling for non-checkpoint jobs.

We've seen some users refer to this as "virtualization", and that is a misnomer. We want to emphasize that there is no hardware virtualization taking place here: your job will run on the bare-metal, physical resources you have requested from Slurm.

While this may seem like a minor change in language, it will greatly ease the coordination among many groups working behind the scenes to support the Hyak service. As always, we appreciate your understanding and patience as we continue to refine and improve the support provided.

See also:

• Glossary of Hyak specific terms.

While the storage on klone (i.e., mmfs1 or gscratch) may appear to be a monolithic device, it is an extremely complex cluster in its own right. This storage cluster is mounted on every klone node: so despite appearing as "on the node", gscratch physically resides on specialized storage hardware separated from the compute resources of klone. The storage is accessed across a high-speed, ultra low-latency HDR Infiniband network, and is designed to be scalable independent of KLONE’s compute resources.

As mentioned in an earlier blog post today, our incoming hardware expansion will drastically increase the amount of demand the storage cluster can handle. In the meantime, the Hyak team has taken measures to help maintain a usable level of storage performance for users and jobs:

1. Improved internal storage metrics gathering and visibility.#

The Hyak team improved storage-cluster metric gathering and visibility, allowing us to correlate those metrics to reports of poor user experience, and to make data-driven tuning and storage policy decisions.

In the figure above we have visibility into if an abnormally high number of jobs have errors that might suggest underlying storage or other user experience issues.

2. Created custom filesystem migration policies to optimize the use of the NVMe layer.#

The bulk of the storage capacity on klone is stored on rotary hard disk drives totalling approximately 1.7 Petabytes (PB) of raw storage. In addition to the hard disk storage, there is a much smaller, extremely fast–and expensive–pool of NVMe "flash" storage that functions both as a write buffer for new files written to the filesystem, and also as a read-cache-like layer where files can be read without causing load on the rotary disks.

The Hyak team has also optimized the file placement policy: files most likely to generate heavy load reside in the limited space of the NVMe layer, ensuring that no storage load is generated on the hard disk layer when those files are repeatedly accessed.

In the figure above you can see that the flash tier (green line) is allowed to fill up to 80% capacity due to job writes then the migration policy begins until the flash tier is down to 65% full. For the majority of the past few several weeks we can see things worked as expected. However, there were a few events recently where jobs were producing so much data that the flash tier was able to get to 100% full faster than the storage system could move data off the flash tier. Giving the migration process too high of a priority results in "slowness" in the user experience. We have since been tuning the aggressiveness of this migration process to reduce the likelihood of it occuring again.

3. Added QoS policies to improve worst-case filesystem responsiveness.#

The klone filesystem has a coarse Quality-of-Service (QoS) tuning facility that allows the filesystem to cap the rate of storage operations for various types of storage input-output (IO). The Hyak team has used this facility in two different ways:

1. First, to limit the storage load impact when the NVMe layer, described above, needs to free up space by moving files to the hard drive layer.

2. Secondly, to moderate the amount of storage load that can be generated by any single compute node in the cluster. This way, outlier jobs in terms of storage load generation are less likely to have an outsized performance impact on the storage.

4. Manually identifying jobs causing a disproportionate impact on storage performance.#

Utilizing metrics and old-fashioned sleuthing, we have been manually tracking down individual jobs that appear to be having a disproportionate and/or unnecessary impact on storage performance, and working with users to address the storage performance impact of these jobs.

In the above figure we can see job IO follows a power law dynamic, a small handful of jobs are often responsible for the majority of load. In this case a single job on a single node is responsible. When users report storage "slowness" this disrepancy can be even more pronounced but we are able to quickly narrow down which specific nodes are responsible and address these corner cases.

5. Dynamically reducing the number of running checkpoint partition jobs.#

As of April 19th, 2022, we have implemented data-driven automation to moderate storage load by dynamically managing the number of running checkpoint (ckpt) partition jobs. When the number of running ckpt jobs is being limited, pending jobs will show AssocGrpJobsLimit as the REASON for not starting.

Please note that non-ckpt jobs (i.e., jobs submitted to nodes your lab contributed to the cluster) are not limited in any way. The social contract when joining the Hyak community is that you get access to the nodes your lab contributes on-demand, and–if and when they are idle–access to other labs’ resources on the cluster. However, access to other labs’ resources isn’t and hasn’t ever been guaranteed: it’s just that there’s often a steady state idle capacity for users to "burst" into by submitting ckpt jobs.

In aggregate, 'Storage Load' is a consumable resource just like CPU cores or memory, albeit one that impacts the whole cluster when it is over-consumed. The Slurm cluster scheduler cannot directly consider storage load availability when evaluating resources for starting ckpt jobs, hence our need to automate. Our new tooling limits the storage performance impact from ckpt jobs in order to improve storage stability for everyone.

The red and blue lines represent two storage servers that we have most closely tied to the user experience and 50% load being the threshold we aim to remain at or under by dynamically reducing the number of running ckpt jobs when it exceeds that limit.

So far, this appears to be very effective at moderating the overall storage load, preventing the storage cluster from becoming unusably slow and avoiding other storage-performance issues. We will continue to tune it in search of the best balance between idle resource utilization via ckpt and storage performance.

6. Expanding the team#

Acknowledging that the storage sub-system is a complicated machine in its own right, it needs much more care and attention and the current Hyak team is stretched incredibly thin as is. We have started the process of hiring a dedicated research data storage systems engineer to focus on optimizing storage going forward.

See also:

• A summary of the state of the union on klone storage.
• Things you as a researcher using klone can do to optimize your storage use.

While some of what precipitated this conversation is the current state of the storage (i.e., mmfs1 or gscratch), there are several things you can do as a researcher to both reduce the load on gscratch as well as help insulate your jobs from cluster-wide storage slowdowns.

1. Use local node SSDs.#

Each node on the cluster has a local SSD drive with 350+ GB of space available for use by user jobs. This space is available only to jobs running on that node and all contents are purged when the users’ last job running on the node completes. It is mounted as /scr and /tmp (both paths go to the same place) on all the compute nodes.

If input data, Apptainer (Singularity) images, or other files used by your job will fit, copying those files to the SSD (via cp, rsync, etc.) once at the beginning of your job and reading them from there during the remainder of the job run results in less load on the central storage, helps insulate your job from any instances of central storage slowness, and can often result in better overall job performance.

Slurm has a command called sbcast [www] that is useful for efficiently copying files to all nodes used in a multi-node job as part of an sbatch script.

For files being written that need to be kept after the job run, it is generally best to write these directly to the central storage. Because new files are written directly to the very fast NVMe layer, such writes are less likely to impact overall storage performance. That said, it is still beneficial to write intermediate job files to the local SSD whenever possible.

2. Code for efficient file IO.#

While this can be a very complicated topic, a great deal of overall job performance can be gained by thoughtful and judicious use of file input-output (IO). Some general tips:

• Keep in mind that file access is orders of magnitude slower than memory access, and processes often have to completely "stop and wait" for disk IO operations to complete. Minimizing file IO operations, especially inside "inner loops" of programs can greatly speed up job completion, and helps to reduce load on the cluster central storage.
• Fewer, larger file IO operations are generally more efficient than multiple smaller file operations accessing the same data.
• When possible, store data in an efficient format such as HDF5 instead of many small files.
• "Open/read once, access many times" if job memory permits.

3. Containerize your environment.#

As mentioned above, minimizing the number of files you need to access can help reduce the number of input / output operations per second (IOPS) happening on the cluster. For example, a Python miniconda environment can create hundreds or even thousands of small files when you install different library dependencies. While Python is a common compute environment, this can be generalized to most other programs you may need. When you containerize your environment, this gets reduced to a single file. A brief introduction to Singularity (now called Apptainer) can be found here. As a side benefit, containerizing your environment–making it a single file–makes it much easier to move it around (see #1 above).

4. Stay under quota.#

Constantly hitting your inode (e.g., file) or block (e.g., number of GBs or TBs) quotas can cause extra storage slowness. If you need a bump on either please reach out to discuss your options. As a reminder you can us the hyakstorage command on klone to display current quota usage for all of your filesets as well as your home directory. Please note that this output is updated once an hour so it will take time to reflect any overages.

5. Report issues.#

While the Hyak team has an extensive monitoring and alerting framework in place to help us to proactively determine when things may be going wrong, not all causes of slow user experience are currently correlated to metrics. Furthermore, our team generally interfaces with the cluster in different ways than our users, so we may not be as equally exposed to any pains until it is reported to us. If you’ve run into a performance issue, please submit a ticket by emailing help@uw.edu. Please provide any symptoms you are observing, along with the date, timeframe, job IDs (if applicable), commands you are running with their full output, etc. If you don’t need or want a reply from us it is still helpful for us to hear from you, feel free to say "no response needed" or something along these lines so we know how to respond.

See also:

• A summary of the state of the union on klone storage.
• Things the Hyak team has done (and currently doing) to optimize the storage environment.

The 3rd generation Hyak cluster, klone, launched in spring 2021 with 144 HPC nodes and 192 GPUs. In just a single year, we’ve grown to over 384 HPC nodes (a 166% increase) and 448 GPUs (a 133% increase). klone has more than doubled in size, and while some of this growth comes from long-standing Hyak members migrating to the new cluster, much of our increased capacity comes from hundreds of new researchers joining the Hyak community. We’ve seen existing sponsors such as the College of Engineering increase their already substantial footprints by 60%, we’ve welcomed new sponsors such as UW Bothell, UW Tacoma, and the Puget Sound Institute, and seen over 1000% growth–seriously–in our new self-sponsored tier for investigators and faculty without an existing Hyak sponsor affiliation. As with any large project, during KLONE’s initial planning stages we made assumptions about our growth rate & the types of research we would be supporting: assumptions that have been shattered by our growth over the past year. It was never a question of if we would need to upgrade our support infrastructure–like storage–but when, and our rapid growth significantly accelerated our upgrade timeline.

Monitoring – and developing more monitoring for – the Hyak clusters is a central responsibility of our team. The status quo at the beginning of 2022 was to track down errant jobs or workflows when storage issues came up. In almost every instance, we were able to pinpoint the problematic job and work with the researcher to shape their code into a normal IO profile. Pausing jobs and providing best practices was sufficient to keep the storage performance solid for everyone. However, starting around the last week of March 2022, we started having trouble finding an obvious job, or even a set of jobs, impacting storage performance.

The truth is that our baseline load had shifted. Due to our tremendous growth, things researchers had previously been doing without issue were now causing problems. We also noticed an evolution of the types of research happening on klone. The Hyak community diversified from traditional HPC workflows (e.g., simulations) into more data-intensive areas like data science (e.g., R jobs), deep learning, and artificial intelligence research. We accelerated our discussions with storage vendors: in a few short months, an expansion went from an eventuality to an immediate and pressing need. Still, we tried several last-minute optimizations to see if we could prevent spending all that money. We are serious about our fiduciary duty, as stewards of this research platform, to provide the most value for the Hyak community with the dollars we are entrusted with. We knew a storage upgrade for klone would cost hundreds of thousands of dollars and we needed absolute certainty that we couldn’t engineer a way around that expense.

The storage on klone (i.e., mmfs1 or gscratch) might pretend to be a mere folder or directory, but in truth it’s an abstraction of a highly complex system. To provide cost-effective, high-performance storage, a small high-speed NVMe "flash" layer acts both as a write buffer for the slower spinning disks–which make up the vast majority of cluster’s capacity–and as a high-speed "cache" for recently & frequently accessed small files. While presented as a single folder to the researcher, behind the scenes the storage cluster moves data between these tiers to balance performance. As seen in the figure above, when the flash layer reaches 80% capacity, a process begins to drain it by moving less frequently used files to the spinning-disk layer until the flash layer reaches 65% capacity. You might also notice that despite our precautions and monitoring, as of April 9, 2022, we were no longer able to migrate data from flash to spinning disks faster than our users were writing. This was the final deciding factor for us, and we initiated our long-standing plan to upgrade the storage for klone.

This necessary investment to upgrade storage will double both the maximum input-output operations-per-second (IOPS) and throughput (storage bandwidth), providing much needed overhead for current workflows as well as accommodating future growth. We are excited for this upgrade – and are doing everything we can to expedite its deployment – but due to the sheer amount of hardware we’re purchasing, we’ve been swept up in the pandemic-induced global supply chain crunch. Our vendors have predicted that the end of July is the worst-case scenario, but that a June delivery is also possible. We will update the Hyak community as we know more. As always, we welcome any questions: if you want to speak with us about something, send an email to the Hyak team via help@uw.edu and we’ll follow up with you.

See also:

• Things the Hyak team has done (and currently doing) to optimize the storage environment.
• Things you as a researcher using klone can do to optimize your storage use.

Background#

In late 2020, while building the current-generation cluster, klone, our previous-generation cluster, MOX, was running CentOS 7 – which was nearing end-of-life support. We used the transition to klone as an opportunity to deploy CentOS 8, the world’s most popular OS in academic research computing environments. Unfortunately, around the time we were wrapping up KLONE’s software stack, the CentOS project announced [1, 2] a transition of their own: Red Hat unilaterally terminated the development of CentOS as an open-source version of Red Hat Enterprise Linux (RHEL). CentOS would become an upstream version of RHEL – in other words, more experimental and ultimately less stable.

As the dust from this announcement settled, a consensus emerged: Rocky Linux, led by the initial founder of the CentOS project, Greg Kurtzer, would become the CentOS successor.

The Transition#

Fast-forward to late 2021: after our summer ‘21 launch of klone, and our fall ‘21 cluster capacity expansion, we were finally able to turn our attention to the CentOS to Rocky migration. And just in time, too, because CentOS 8–the operating system we deployed just months earlier–would be officially unsupported after December 21, 2021.

Our goal was to make this OS transition as smooth and unnoticeable to our users as possible. After all, this is our mission: we take care of the tech so that you can take care of the science. Rocky, like CentOS, is intended to be a bug-for-bug, open-source version of RHEL, and with its talented, globe-spanning team of developers, we were confident that the impact of this transition would be minimal.

We began the transition with our backend during the December ‘21 maintenance: the klone head node, our Slurm scheduler, was successfully migrated to Rocky 8. So far so good! During our next maintenance, January ‘22, we migrated all the compute node images to Rocky. A handful of users reported code-compiling issues, which we were able to resolve, but otherwise it was uneventful. We took extra care on the final piece of the Rocky migration–the login nodes–due to their accessibility from the wider internet. And, as of today’s maintenance, we are excited and relieved to report that klone is now a 100% Rocky cluster! 🥳

Summary#

The Hyak team was forced to revisit a major OS migration, mere months after the initial launch of klone. This is highly unusual–and no small feat–but we have prevailed. We deployed a widely-supported, open-source OS with enterprise-level stability, while remaining cost-effective to the research community at the University of Washington. With this work behind us, we’ve arrived at a sustainable platform for the life of the klone cluster. We’re excited for the future of klone, and excited to redirect our time back to feature development.

We want to give a huge thank-you to our users for their patience during this migration period. Spoiler alert: Rocky won and it’s a good thing!

History#

In fall 2019 (almost two years ago to the day) the Hyak team received our first Turing generation GPU node. Hyak has had a modest GPU footprint in the past as far back as a decade ago with the first generation cluster (called "IKT") and its pre-Pascal generation cards. In 2015 we acquired a smaller test bed of Pascal generation GPUs for the second generation cluster (called "MOX"). There were never more than a dozen GPUs in either the IKT or MOX clusters, but the introduction of Turing GPUs marked a resurgence of interest in these accelerators among the UW research community. In the last two years, we've substantially expanded our capabilities to over 300 GPUs.

Background#

Hyak clusters work on a "condo" model: labs are able to utilize their contributed hardware on-demand as well as take advantage of idle capacity from other groups' hardware via the checkpoint (ckpt) partition. Your checkpoint priority — or "fairshare" in Slurm scheduler parlance — is weighted such that your fairshare is directly proportional to your lab’s contribution to the cluster. In the MOX days, GPU users tended to stay within their contributed hardware partitions and rarely made use of checkpoint. We attributed this to a mental shift: students were used to using a single resource, like a desktop computer, rather than a shared cluster of computing resources. However, with the migration to the third generation Hyak cluster (called "klone") and its new QoS scheduling system and the increasing comfort of students using a shared platform, GPU utilization in the checkpoint partition has increased as well. This is a good thing: we want groups to benefit from their Hyak membership in the cluster and take advantage of idle cluster resources beyond their initial hardware contributions. This is a primary tenet of our social contract with the Hyak community: as a node contributor to the cluster, you have access to idle resources of the whole cluster.

Problem#

Fairshare was simpler to calculate in the pre-GPU days because our infrastructure was homogenous: one node contributed to the cluster equaled one fairshare unit. During the last two years of exponential GPU adoption on Hyak, the fairshare calculation has not evolved: 1 HPC node was the same as 1 GPU node at 1 fairshare unit. This didn’t hold because a GPU node can cost between 4 to 8 times (or more) than a traditional HPC node. The result was that labs with GPU or other speciality (e.g., high-memory) nodes tended to have smaller fairshares compared to groups with the same dollar investment but only in traditional CPU nodes. In practice, this meant these GPU users often directly competed for resources with non-GPU jobs in the checkpoint partition on a non-level playing field.

Solution#

Taking into consideration all of this information, as well as the fact that you can request as little as 1 GPU or 1 CPU from the scheduler, we have adjusted the fairshare calculations as follows:

• Financially: 1 GPU card is roughly equivalent to 40 CPU cores (on a dollar basis), therefore the cost normalization is 40:1 in favor of GPUs.
• Scarcity: 1 server typically holds 8 GPU cards or 40 CPU cores, therefore the scarcity normalization is 5:1 in favor of GPUs.
• Combining the financial and scarcity considerations in the points above, the final weighting is 200:1 in favor of GPUs. In other words, 1 GPU card is worth 200 times more than a single CPU core in the eyes of the scheduler and factored into your checkpoint fairshare. Please note that this example only applies to the higher GPU memory cards (i.e., gpu-rtx6k) while less expensive GPUs have commensurately less weight.

Summary#

With the October monthly maintenance today we have introduced a new fairshare weighting system on the klone cluster's checkpoint (ckpt) partition that commensurately acknowledges GPU labs for their contributions to the Hyak community. This has no impact on jobs submitted to non-ckpt partitions.

If you were previously a proficient mox user and now find yourself on klone, what's new / different? This is a high-level summary, please consult the documentation [link] for more details.

Login#

• Logging in was previously to mox.hyak.uw.edu now it's klone.hyak.uw.edu.
• As a reminder login nodes are only to connect to the cluster, navigate the cluster file system, and submit jobs. This applies to both klone and mox. Do not compile codes on the login node or run any programs that require significant compute (get a session with Slurm).

Data Transfer#

• Only use the login node to transfer data on klone. On mox you'd have used a build node or could have used the login node if it wasn't very computationally heavy.

Storage#

• The path to lab storage is still /gscratch/mylab on both klone and mox. You'll need to copy over the data from mox to klone you want to continue using.
• Home directories are still 10GB per user, same on both clusters.
• Scrubbed exists on klone just as it did on mox at /gscratch/scrubbed this is a free-for-all space on both clusters where files are automatically deleted after 21 days.
• Some new benefits of the klone storage compared to mox:
  • There are snapshots for gscratch! Look inside the /gscratch/mylab/.snapshots folder for a copy of your lab folder once an hour, every hour, for 24 hours. This is not a backup copy nor a replacement for version management (e.g., git) but useful for retrieving recent versions or something accidentally deleted. This is currently disabled.
  • More storage! Previously you received 500GB or 0.5TB of gscratch quota per node (or pair of GPUs) contributed to mox. Now on klone we've doubled your associated storage quota! For example, 2 nodes on mox would mean 1TB of gscratch but 2 nodes on klone now means 2TB of gscratch. If you had an 8 x GPU node on mox you would have received 2TB of gscratch but an 8 x GPU node on klone now means 4TB of gscratch.
  • It's faster! We've had reports of performance that's averaging a 30% speed up all else being equal, nothing you need to do aside from use klone instead of mox.
  • It's faster than fast! While klone storage is faster than mox storage overall, gscratch on klone is further turbo charged with a NVMe flash based tier. NVMe flash is among the fastest storage mediums you can get and further differentiating benefit if you use gscratch vs scrubbed on klone.

Compute#

1. When submitting a Slurm job, whether interactive (i.e., salloc) or batch (i.e., sbatch) you'll want to first decide which account to use. This is the group you're part of. You can run the command groups to see your affiliated accounts and run hyakalloc to see all the resources (e.g., compute cores, memory, GPUs) used and available associated with each affiliated account.
2. Then decide if you want to run this job to count under your resource allocation by submitting to the compute partition (i.e., -p compute) or if you want this job to use idle resources from other groups across the cluster using the checkpoint partition (i.e., -p ckpt).

• Non-standard partitions. Run sinfo to see the list of all possible partitions, this is only if your group contributed non-standard nodes (e.g., high memory, GPUs) and need to idenitify the appropriate partition names to get immediate use. Otherwise, you'd only be able to get them in a checkpoint capacity. For GPU users this is currently either the gpu-2080ti or the gpu-rtx6k partitions for 11GB and 24GB of GPU memory cards, respectively.
• There is no build node on klone. Get an interactive session (e.g., salloc) under an existing account and partition combination you have access to.
• All nodes have internet now on klone. Do all data transfers to and from klone on the klone login nodes, the login nodes on klone have dual 40 Gbps uplinks to the internet. While the compute nodes on klone have internet routing now, they are bottlenecked at 1 Gbps so not suitable for big data transfers.

Software#

• Singularity containers work the same on both clusters, we encourage this when possible. Refer to our container documentation [link].
• Modules is updated to the latest versions of the most core parts that the Hyak team maintains (e.g., gcc, Intel, Matlab). Refresh yourself about modules [link].
• If neither Singularity nor existing modules works for you, you may have to re-compile your codes on klone. "contrib" modules works different now on klone vs mox, please check out the details [link].

February 25, 2021#

The UW research computing team celebrates the soft launch of project klone, the 3rd generation Hyak supercomputer. Welcome to those researchers invited to participate in the early access program 🥳 🎉

The user documentation [link] has been updated to reflect the changes and new features of klone but this will be an ongoing process.

Compute#

• Soft launch with 1,920 compute cores over 48 nodes:
  • 28 x mem1 nodes (192GB of memory each) in the compute partition,
  • 4 x mem2 nodes (384GB of memory each) in the compute-bigmem partition,
  • 16 x mem3 nodes (768GB of memory each) in the compute-hugemem partition.
• build nodes no longer exist on klone as they did on mox. All instances have the potential to be interactive and all have internet routing by default (even non-interactive jobs).

Storage#

• gscratch on klone is 1.4PB total capacity with a new 500TB NVMe flash tier. Data tiering happens automagically, if you use a file frequently it will be moved to the faster storage.
• Storage quota is still charged back at the same rate ($10 / TB / month). Researchers receive 1TB per node purchased and contributed to klone.

Data#

• gscratch is not backed up that is the responsibility of the researcher (e.g., LOLO, the cloud, external hard drive). Feel free to email us if you have any questions.
• While all nodes have internet access now, transfer data using the login nodes. Login nodes have full 2 x 40 Gbps bandwidth. If you transfer using a compute node interactive session you are limited to 1 x 1 Gbps connection.

Software#

• modules works the same as it did on mox. This is an improved implementation called LMOD on klone compared to environment modules on mox.
• We provide the basic compilers (e.g., GNU, Intel) as modules.
• The Hyak team is encouraging a container first world (i.e., use Singularity).

March 3, 2021#

The updated total is 3,840 cores and 96 nodes on klone.

Compute#

• Compute has doubled by adding another rack to klone, an additional 1,920 compute cores over 48 nodes:
  • 44 x mem1 nodes (192GB of memory each) in the compute partition,
  • 2 x mem2 nodes (384GB of memory each) in the compute-bigmem partition,
  • 2 x mem3 nodes (768GB of memory each) in the compute-hugemem partition.

Software#

• We created a module for cmake.

March 5, 2021#

Storage#

• Implemented usage_report.txt files in the base folder of /gscratch/yourlab/ that is updated once an hour to reflect both your block quota and inode capacity usage. This is similar to the gscratch experience on the MOX cluster.

Website#

• We migrated our site from https://UWrc.github.io to its new home at https://hyak.uw.edu.

March 9, 2021#

Storage#

• Snapshots are here! We are piloting once an hour for 24 hours for every lab storage folder under /gscratch/. Check out the updated documentation here on how to access past snapshots.

Software#

• We created more LMOD software modules:
  • Matlab R2020b [docs]
  • OpenMPI-4.1.0

March 12, 2021#

• LMOD software modules:
  • Intel has bundled their software suite (e.g., compiler, MPI) as oneCLI and we created this module (i.e., module load intel/oneCLI).
  • There is now a "contrib" framework for groups to store their shared codes separately from their /gscratch/labname/ data. You can get 100GB of storage to compile codes at /sw/contrib/labname-src/ and then put your LMOD module file in /sw/contrib/modulefiles/labname/. Your module would appear when anyone runs module avail. This is created upon request so if you'd like to opt-in your group please let us know.

April 13, 2021#

Things have been going steady the past week and changes are coming less frequently. We are now increasing time between maintenance periods on klone from weekly on Tuesdays to monthly and aligning it with the mox maintenance as the 2nd Tuesday of every month.

That wraps up our klone soft launch blog updates here, other updates will appear on our Hyak users mailing list. Don't forget to subscribe, instructions on this page at the bottom.