Center for Quantitative Life Sciences HPC Documentation¶
-
Quick Links
-
Get Support
-
Handy Resources
-
Questions or Comments?
- Email the Bioinformatics group for research design and analysis questions
- Questions for the Lab?
- Contact Ed for HPC software support, if you'd like to contribute a tips or BUG talk, or if you have feedback about the HPC docs site.
- Contact Ken for HPC hardware support or VM support (RStudio, etc)
-
RELMS
Announcements¶
2025-01-16 - Updating git repo links to the new gitlab¶
Since we successfully upgraded our hosted gitlab, you may have had trouble using your previously-cloned git repositories. A simple way to update your old repos is to navigate to the repo page on the new gitlab web interface, click the Code -> clone with SSH button, and then run
Making sure to paste in your new url. You'll note that the port number is now
included in the address so you are no longer required to add the number to your
~/.ssh/config
.
If you prefer to clone with https, you can also do that instead and use your ONID password.
You can check that your repo was updated with the git remote -v
command.
2025-01-16 - Updated info about using the VPN with WSL¶
We added a note about using the VPN with WSL. See here
2024 Announcements¶
2024-10-22 - Using pixi in place of conda¶
We added more details about using pixi for managing conda environments.
Please let us know successes and pain points of using pixi going forward.
2024-10-22 - Fall Conference 2024 Registration is Open!¶
Register here and see the conference website for more information.
Message from Brent Kronmiller, Interim Director, Center for Quantitative Life Sciences
Conference registration is free. While registering, please sign up to present your poster and indicate if you would like to be considered for a lightning talk.
Friday, November 22nd, 2024 at OSU's CH2M HILL Alumni Center.
Please visit the conference website (https://cqls.oregonstate.edu/2024-cqls-fall-conference) for the in-progress agenda and more information.
We look forward to seeing you there,
Brent
2024-10-21 - Memory settings update¶
We noticed some machines quickly running out of memory to allocate (~50% CPUs reserved, yet 100% Memory reserved). We adjusted the default memory request to 4GB of memory per CPU. Please let us know if you notice machines being unable to accept new jobs due to overallocation, yet seemingly have available CPU and Memory free.
Improvements to hqavail
are on the wait to make it easier for folks to manage CPU and Memory requests and allocations.
2024-10-17 - Memory settings, Slurm, and Draining State¶
As more researchers migrate to the Wildwood cluster, the CQLS and CEOAS computing teams noticed an increase in node getting stuck in a 'Draining' state, usually due to running out of memory resources. The nodes enter a draining state to get those nodes in a partition cleared of jobs. Once the jobs are finished, the node remains in a draining state indefinitely, until it is manually reset by staff.
In order to reduce the number of nodes entering the draining state, the CQLS and CEOAS teams added a Slurm setting to ensure jobs to remain under the requested memory values. In this way, the nodes will not use all of the memory available on them, and nodes will not enter the draining state. The downside is that all jobs need to have some memory setting, and it can be difficult for users to know appropriate choices for their memory use.
Slurm has been configured to automatically reserve 8GB of memory per CPU requested. Most jobs will be able to
complete with this setting, and most users should not have to think about manually changing the setting. Some machines,
especially those with less than (Number of CPUs) * 8GB
Memory will need to have memory settings adjusted. You can
manually request a different amount of (total) memory using the hqsub -f
flag.
The default 4GB setting in hqsub
has been removed such that the Slurm default will now be used.
Please let us know if this causes additional headaches for you, and if you notice nodes going into the draining state.
Tip
You can check nodes in the draining state by running this command: sinfo -t DRAINING
2024-10-17 - CQLS Gitlab down time Thursday Oct, 17 - UPDATE¶
Gitlab downtime will be extended today until 7:00pm this evening. We apologize for the added inconvenience.
Gitlab upgrade has been postponed due to technical issues. See the updated message here.
2024-10-16 - CQLS Gitlab down time Thursday Oct, 17¶
The CQLS Gitlab will be migrating to the Wildwood cluster and will experience downtime from 12:00-5:00pm on Thursday October 17 for security and feature updates.
See the full announcement here.
Command-line git
will still be available during this time, but you will be unable to git push
or git pull
, etc,
with respect to interacting with the CQLS-managed repositories. Please plan accordingly.
You can find updated instructions for using Gitlab over ssh that will be necessary when downtime is over. Please plan to update your config to use gitlab after downtime is over.
2024-10-10 - Transition to RELMS ordering system¶
As a reminder, we will be transitioning to RELMS for billing starting this October (for Bioinformatics) and November (for Biocomputing). The core lab billing as already transitioned. See the RELMS card above for more useful links related to RELMS.
2024-10-09 - New landing page¶
Announcements have been moved below the link grid above. Access those highest impact pages faster!
2024-10-08 - Singularity information is now available¶
We are using a module system (Lmod) to manage our singularity images. See here for more information.
2024-10-02 - Added queue use and priority queue resources¶
See accounts and partitions and priority queueing information.
2024-09-30 - Updated information about checking out GPUs¶
See here and here for some updated information. More changes forthcoming to make using GPUs easier.
2024-09-27 - ssh configuration clarification about username¶
Take a look at the updated ssh config information about usernames and ssh. When you set up
your config, you can use ssh $ONID_USERNAME@hpc
or you can add the User $ONID_USERNAME
to the ssh config for the
shell
machine.
2024-09-21 - hpcman documentation migrated¶
You will find the hpcman documentation updated and now included on this website. Some of the tips posts have also been migrated.
2024-09-20 - Website reorganization¶
The website has been reorganized to facilitate the addition of other groups/colleges (e.g. CEOAS, COE, COB, DRI) to the documentation site. Updates and other announcements will be posted here as well.