8 posts tagged with "storage"

View All Tags

April 2024 Maintenance Details

Kristen Finch

Kristen Finch

HPC Staff Scientist

Hello HYAK Community,

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.

Disk Storage Management with Conda

Kristen Finch

Kristen Finch

HPC Staff Scientist

Hello HYAK Users,

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.

warning post under contruction

We have been made aware that the solutions for disk storage presented here result in additional problems with conda environments, specifically with hardlinks to the install directory for Miniconda3 when envs_dirs and pkgs_dirs are configured to a different storage location. Please see this Issue for detailed information. we hope to have a better solution soon.

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.

note

The following assumes you have already installed Miniconda in your home directory or elsewhere such that conda is in your $PATH. Install Miniconda instructions here.

$ conda info
active environment : None
shell level : 0
user config file : /mmfs1/home/UWNetID/.condarc
populated config files : /mmfs1/home/UWNetID/.condarc
conda version : 4.14.0
conda-build version : not installed
python version : 3.9.5.final.0
virtual packages : __linux=4.18.0=0
__glibc=2.28=0
__unix=0=0
__archspec=1=x86_64
base environment : /mmfs1/home/UWNetID/miniconda3 (writable)
conda av data dir : /mmfs1/home/UWNetID/miniconda3/etc/conda
conda av metadata url : None
channel URLs : https://conda.anaconda.org/conda-forge/linux-64
. . .
package cache : /mmfs1/home/UWNetID/conda_pkgs
envs directories : /mmfs1/home/UWNetID/miniconda3/envs
platform : linux-64
user-agent : conda/4.14.0 requests/2.26.0 CPython/3.9.5 Linux/4.18.0-513.18.1.el8_9.x86_64 rocky/8.9 glibc/2.28
UID:GID : 1209843:226269
netrc file : None
offline mode : False

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.

tip

when you ls your home directory ls /mmfs1/home/UWNetID/ you might not see .condarc listed. It is there! To list all hidden files (files beginning with .) use ls -a /mmfs1/home/UWNetID/.

Configuring your package cache and envs directories#

Edit the highlighted lines in .condarc to designate directories with higher storage quotas for our envs_dirs and pkgs_dirs. Use a hyak preloaded editor like nano or vim to edit .condarc in place. More about nano. More about vim. Your .condarc will look like this:

$ nano ~/.condarc
channels:
- conda-forge
- bioconda
- defaults
auto_activate_base: true
envs_dirs:
- /mmfs1/home/UWNetID/miniconda3/envs
pkgs_dirs:
- /mmfs1/home/UWNetID/conda_pkgs

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. Alternatively, your lab/research group might have another directory in /gscratch/ that can be used.

important

Remember to replace the word UWNetID in the paths below with YOUR username/UWNetID.

Here is what your edited .condarc should look like.

$ cat /mmfs1/home/UWNetID/.condarc
channels:
- conda-forge
- bioconda
- defaults
auto_activate_base: true
envs_dirs:
- /gscratch/scrubbed/UWNetID/envs
pkgs_dirs:
- /gscratch/scrubbed/UWNetID/conda_pkgs
warning

If you don't have a directory under your UWNetID in /gscratch/scrubbed/or whereever you intend to designate these directories you will need to create them now for this to work. Use the mkdir command, for example mkdir /gscratch/scrubbed/UWNetID and replace UWNetID with your username. Then create directories for your package cache and envs directory, for example, mkdir /gscratch/scrubbed/UWNetID/conda_pkgs and mkdir /gscratch/scrubbed/UWNetID/envs.

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

$ conda info |grep cache
/gscratch/scrubbed/UWNetID/conda_pkgs
$ conda info |grep envs
/gscratch/scrubbed/UWNetID/envs

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.

warning

rm -r is permanent. We cannot your recover directory. You were warned.

$ du -h --max-depth=1 /mmfs1/home/UWNetID/
6.7G ./miniconda3/envs
4.0G ./conda_pkgs
. . .
$ rm -r /mmfs1/home/UWNetID/envs
$ du -h --max-depth=1 /mmfs1/home/UWNetID/
2.6G ./miniconda3/
4.0G ./conda_pkgs
. . .
note

The hyakstorage command is not simultaneously updated. Although you have cleaned up your home directory, hyakstorage might not yet show new storages estimates. du -sh will give you the most up to date information.

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.

$ conda activate project1_env
(project1_env) $ pip config list -v
. . .
For variant 'user', will try loading '/mmfs1/home/UWNetID/.pip/pip.conf'
. . .

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.

$ mkdir /mmfs1/home/UWNetID/.pip/
$ touch /mmfs1/home/UWNetID/.pip/pip.conf
$ mkdir /gscratch/scrubbed/UWNetID/pip_cache

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

[global]
cache-dir=/gscratch/scrubbed/UWNetID/pip_cache

Check that your pip cache has been designated.

(project1_env) $ pip config list
/mmfs1/home/UWNetID/.pip/pip.conf
(project1_env) $ pip cache dir
/gscratch/scrubbed/UWNetID/pip_cache

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.

$ cat ~/.Renviron
R_LIBS="/gscratch/scrubbed/UWNetID/R/"

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.

March 2024 Maintenance Details

Kristen Finch

Kristen Finch

HPC Staff Scientist

Hello HYAK Users,

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.

New Documentation

The steps discussed here for responsible use of VS Code have been added to our documentation. Please review the solutions for connecting VS Code to HYAK.

  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
  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.

February 2024 Maintenance Details

Nam Pho

Nam Pho

Director for Research Computing

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!

Update on the hyakstorage command

Nam Pho

Nam Pho

Director for Research Computing

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:

HYAK Team Storage Optimizations

Nam Pho

Nam Pho

Director for Research Computing
note

The HYAK team has taken six concrete steps to stabilize and optimize storage on KLONE over the past few weeks.

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.#

KLONE slurm metrics

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.

KLONE storage policy

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.#

KLONE storage metrics

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.

KLONE storage load

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:

KLONE Users Storage Optimizations

Nam Pho

Nam Pho

Director for Research Computing
note

There are steps you, as a researcher using KLONE, can do to limit the impact of whatever else is happening on the cluster on your individual workflows.

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:

An update on KLONE storage

Nam Pho

Nam Pho

Director for Research Computing
note

KLONE has experienced exponential growth over the first year of its launch, necessitating long-standing storage ugprades to occur. The current estimate is between June and July 2022 for deployment of this hardware.

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.

KLONE storage policy

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: