Caitlin’s Back! (Temporarily)
The SSCC Help Desk has been short-staffed since August and, by happy coincidence, Caitlin Tefft is temporarily available. Caitlin worked for the SSCC from 2009 to 2021 before she left to complete her Master’s degree in Library and Information Studies with a concentration in Data/Information Management & Analytics. She’ll be working half-time for the SSCC, and dividing that time between staffing the Help Desk and putting her new expertise to use by helping SSCC’s qualitative researchers–more on that in a moment.
Once Caitlin starts, on December 2nd, the Help Desk will again be available for walk-in visits (4226 Sewell Social Sciences Building) and phone calls (262-9917). Of course you can continue to send email (helpdesk@ssc.wisc.edu) and make appointments for video chats (https://sscc.wisc.edu/get-help/).
SSCC Now Offering Consulting on Qualitative Analysis
Caitlin’s primary interest has always been qualitative analysis, and some of you will recall that she was assisting researchers with both Nvivo issues and methodological questions even before she finished her master’s degree. As long as she’s here, the SSCC will be able to offer consulting on qualitative analysis similar to the consulting on quantitative analysis we’ve offered for many years. Please take advantage of it! We hope to use this opportunity to make the case that qualitative consulting should be permanently available on campus.
This includes training, so watch for additions to the training schedule in early December.
Spring Training
Our “Introduction to…” and “Data Wrangling in…” workshops for R and Stata are scheduled for the week of January 13th, and will be online so you don’t have to be back in Madison. We highly recommend them for new grad students, grad students who are ready to start doing research, or anyone who wants to learn a new language. Data Wrangling will teach you how to take real-world data and put it in a form you can actually analyze, a critical but often neglected skill. We’ll also teach a variety of topics later in the semester.
For details and to register visit the training page–and watch for workshops on qualitative analysis to be added soon!
Self-Service Portal Now Includes Project Permissions
The SSCC Self Service Portal (under Accounts on the SSCC web site) now allows you to give people access to your project directory or take it away without needing to contact the Help Desk. Of course you’re welcome to continue sending these kinds of things to the Help Desk.
New Research Database Server Available
The SSCC has a new database server for research data, ResearchDB. It runs two flavors of SQL: Postgres and MariaDB, which is a variant of MySQL. Contact the Help Desk if you’d like to use it.
Slurm Tips
Slurm has been busy this semester. Some tips to both get your jobs running sooner and make sure you aren’t unnecessarily delaying other peoples’ jobs:
Only reserve the resources your job needs. We see too many jobs reserving 128 cores and 200+GB of memory and then using one core and maybe 10GB of memory. The Guide to Research Computing at the SSCC has a section on identifying your job’s needs before you submit it; the email you get when your job finishes will tell you what it actually used so you can adjust your reservations in the future.
Check what’s available. It’s not unusual for the short job server (<6 hours) to be available when most of the rest of the cluster is busy, or the servers that only have 44 cores. Maybe your job would run faster with 64 cores, but using 44 cores and starting right away might get it done sooner. (To see what’s available, visit the Slurm Status page.)
If you’re in Econ, use the econ partitions when others do. Jobs submitted to the econ, econ-grad, or econ-fac partitions will preempt jobs submitted to the sscc partition on the servers the Economics Department helped pay for. It may feel collegial to just use the sscc partition even if you’re in Economics, and usually it is! But if a small number of heavy Econ users have filled the econ partitions, all sscc partition jobs must run on the non-Econ servers, which can make for long queues. When this happens (and it happened recently for several weeks) we’d appreciate it if Econ users would submit their jobs to the econ partitions, and let the non-economists use the sscc partition. That will get us closer to the ideal of everyone getting a fair share of the cluster’s compute time independent of the partition they chose to use. (To see what’s currently in the Slurm queue, run “squeue -a”.)
Changes to MFA Requirements for Linstat
We’ve expanded the list of networks where you do not need multifactor authentication (MFA) to log onto Linstat to include all wired networks on campus and eduroam. You still need to use MFA to connect to Linstat from UWNet, the default wireless network. Note that all students, staff, and faculty can set up their computer or other device to use eduroam over the campus wireless network, which has the added advantage of always using encryption. You can also bypass Linstat MFA by using VPN (since it already requires MFA).
Check Out Linstat3 to Test Alma Linux 9
During January’s tech update, we will upgrade all of our Linux servers to Alma Linux 9 (from Alma Linux 8). We do not expect any problems due to this upgrade–in fact you may not notice any differences–but if you’d like to verify that your jobs will run properly under Alma Linux 9, Linstat3 is running it now. To try it, just log into Linstat normally and then type “ssh -Y linstat3”.