Cluster Shutdown
When
Thu, Feb 22, 2024 @ 8:00am to Fri, Feb 23, 2024 @ 8:00pm
Reason
Maintenance and upgrade of the operating system and central storage system.
During the shutdown users will not be able to access the cluster via SSH or web-based connections, including Rstudio Server, JupyterHub and OnDemand. Web-based file access will also not be possible. Prior to the shutdown please save your work. Updates on the status of the shutdown will be posted on HPCC’s alerts page. If you need additional information or help, you can reach us on HPCC’s Slack workspace (https://ucr-hpcc.slack.com/) or outside the shutdown also via email (support@hpcc.ucr.edu).
Note that if you submit a Slurm (or OnDemand) job prior to the shutdown that extends into the maintenance window, your job will be queued with the reason “ReqNodeNotAvail”, indicating that your process cannot run due to the scheduled shutdown. If this happens please cancel the affected job and resubmit with a shorter run time that does not overlap with the shutdown time. As of this email, the maximum possible run time for jobs is just over 27 days.
We apologize for the inconvenience. Occasional maintenance events are necessary to keep our systems up-to-date, which is to the benefit of everyone using our resources.
Release of Open OnDemand
What is it?
Recently we deployed Open OnDemand, which enables accessing our HPC resources directly from a web browser with no additional client software required. It allows users to start their own JupyterHub, RStudio (Posit), and Desktop Sessions. This approach scales much better to large numbers of users by running these instances on individual computer nodes. This means, users no longer need to share hardware resources with other users on a single server instance as this was necessary on our old web-based Jupyter and RStudio servers.
Where
The new OnDemand service is available here:
https://ondemand.hpcc.ucr.edu/
Additional information about using Open OnDemand on the HPCC is here.
Future
If OnDemand works well for our users then it may become the default way for accessing Jupyter and RStudio on the HPCC.