The Argon HPC /nfsscratch filesystem is performing slowly for some jobs. The Research Services team is investigating and monitoring the situation.

The Argon High Performance Computing (HPC) system will be down for maintenance.

Any jobs running at the beginning of the maintenance window will be stopped. Please plan your work accordingly for minimal disruption.

The retention period of files on the Argon HPC system-wide /nfsscratch filesystem has been reduced from 60 days to 40 days after they are created.

The retention period was temporarily reduced to 40 days this fall, when usage of /nfsscratch increased significantly. Since restoring the retention period to 60 days, /nfsscratch has consistently reached capacity, which prevents jobs from running.

The retention policy for node-specific /localscratch filesystems is not affected by this change.

All scratch filesystems policies are available as part of the Argon Scratch Filesystems documentation.

Some users may be unable to mount their home directory and receive an error message similar to "cannot access /Users/HawkID: No such file or directory". Support staff are investigating.

Users may experience degraded performance as the Argon HPC /nfsscratch filesystem has reached capacity.

To alleviate performance issues and avoid impacting running jobs further, we are now reducing the deletion of data to retention periods from 40 days instead of 50 days on the /nfsscratch system.

You can help by deleting data which is no longer needed from /nfsscratch, or contact research-computing@uiowa.edu to have your entire /nfsscratch directory deleted.

Please note that job launches will resume automatically once the share drops below the 99% threshold. However, if the share fills back up again job launches will once again be paused.

We are continuing to monitor the situation and may suspend job launches if it does not improve. We will communicate if additional actions are necessary.

Free space on /nfsscratch is decreasing rapidly, and at the current pace we expect it to fill in the next 1-2 days. If /nfsscratch fills, it will cause issues including HPC job failures and no new data will be written.

To help avoid filling /nfsscratch, please ensure you are removing data in /nfsscratch as you finish with it rather than waiting for the normal automated cleaning mechanisms.

In the event that /nfsscratch becomes full, you can use the /scratch file system.

ITS-Research Services is taking steps to slow the filling:

• HPC users writing to /nfsscratch with large data consumption have been contacted.
• Data is actively being removed on the backend.

ITF lost utility power today around 9:08 AM. This has impacted ITF HPC compute nodes. When the power switched over to the generator those rows/systems lost power during this brief period. This resulted in jobs running on any ITF HPC nodes to fail. These systems will run while ITF is using generator power, but do not restart automatically.  Research Services is in the process of bringing the computer nodes back up.

Quarterly maintenance for the Argon HPC system.

See Fall 2022 HPC Maintenance for more information.

Quarterly maintenance for High Performance Computing and related services.

See Summer 2022 HPC maintenance for more information.

Some users are currently unable to access their home drives on Argon. Affected users may receive an error message like “Could not chdir to home directory /Users/hawkid: No such file or directory” when trying to access their Argon home directory. 

IDAS users might receive errors starting an IDAS session when requesting their hpchome be mounted.

ITS is investigating this issue and updates will be posted on the HPC System News support page.

Some users are currently unable to access their home drives on Argon. Affected users may receive an error message like “Could not chdir to home directory /Users/hawkid: No such file or directory” when trying to access their Argon home directory.

ITS is investigating this issue and updates will be posted on the HPC System News support page.

System updates will be installed during this maintenance window. Please refer to https://hpc.uiowa.edu/system-news/spring-2022-hpc-maintenance for more information. 

Argon HPC is experiencing a service degradation which may cause jobs to fail with errors indicating inability to create a lock for an input or output file. It may also cause apps to freeze or become non-responsive.

For timely updates, please see: https://hpc.uiowa.edu/recent-news

See Winter 2022 HPC maintenance | High Performance Computing for more information and updates.

Argon and IDAS are currently unavailable. Users are seeing failed or stalled logins when attempting to connect to Argon. IDAS will still continue to function if Argon Homes or LSS are not being used. IDAS class instances are down if they have shared libraries or shared storage. 

Research Services is investigating this issue. Further update information can be found at https://hpc.uiowa.edu/system-news/argon-login-issues

Logins to the Argon HPC system may periodically fail due to an unresponsive login node. ITS is currently investigating the issue.

Users experiencing issues connecting to argon.hpc.uiowa.edu should connect to a login node directly instead:

• argon-login-1.hpc.uiowa.edu
• argon-login-3.hpc.uiowa.edu
• argon-login-4.hpc.uiowa.edu

See Fall 2021 HPC maintenance | High Performance Computing for more information.

The Argon HPC system will be in a regularly scheduled quarterly maintenance. For more details, see: HPC Summer 2021 Maintenance

One of the servers for the Argon Home directories is having an issue with file locks. If your home account is on this server, you may experience processes that get stuck shortly after launching. The problem is being investigated and will be mitigated as soon as possible.