The 2nd iteration of the High Performance Cluster, codename “Yeti”, will be retired on November 1st.
The Social Sciences Computing Committee has purchased a number of nodes on the new cluster for its affiliates, which includes Economics, ISERP, and CPRC. Any faculty or graduate student in those departments can get an account on Habanero. Please contact me if you would like a Habanero account, and I will request on on your behalf.
To get started quickly, login using ssh UNI@habanero.rcs.columbia.edu
Documentation for Habanero is available here. Most of the software available on Yeti is also available on Habanero, although note that the process to load software is slightly different, you have to load the specific software module before you invoke it.
Using Stata Example:
1) Start an interactive session first:
$ srun –pty -t 0-2:00 -A sscc /bin/bash
2) Then load Stata module and start Stata
$ module load stata
Storage on Yeti was limited by quota. Habanero does not have quotas, each user can store as much as they need. Please note that 10TB of scratch space and 4 million inodes (files & folders) is shared between all users of the SSCC group, including Economics, ISERP, and CPRC. Please be mindful of your used storage. When you are finished using a particular dataset, please delete it or copy it to your local computer. If we see that you use a disproportionate amount of space (both in total space and in number of inodes), we will ask you to clean up.
You can access storage from Yeti and Habanero from the habanero login. “Old” Yeti storage is available on /vega/sscc/ with the subdirectories of collab, tmp1, and work. “New” Habanero storage is available on /rigel/sscc/ with subdirectories projects (for shared projects) and users (for individuals). You can copy or move data from one location to the other.
Please note that by default, data copied to /rigel/sscc/ and their subfolders are readable by everyone in the SSCC group – if you want to limit it, please change the permissions accordingly.
For general help or overview questions, you can ask me for help.
For more specific cluster/software help, please email email@example.com.