|
|
Data Analysis Cluster Frequently Asked QuestionsTable of Contents1. What hosts are part of the cluster?2. Are there any differences between the hosts? 3. Do I have to move any of my files if I change which host I'm working on? 4. Why can't I get to the cluster from roy? 5. What compiler suite do I use to build my Fortran, C and C++ APIs? 6. What compiler suite do I use to build MPI applications? 7. May I run long-running jobs on the cluster? 8. Does the cluster support /contrib for users? 9. What file systems are available? 10. What commercial applications are available? 11. How do I check my quotas? Data Analysis Cluster Frequently Answered Questions1. What hosts are part of the cluster?The cluster currently contains five hosts: squall.scd.ucar.edu, gust.scd.ucar.edu, hail.scd.ucar.edu. breeze.scd.ucar.edu and gale.scd.ucar.edu 2. Are there any differences between the hosts?Functionally there is no difference between the systems. All of them have access to the same data areas, and the architectures are compatible. This means that code compiled on one node will run on any of the nodes. For a complete hardware description, see the Environment Overview document. 3. Do I have to move any of my files if I change which host I'm working on?No. All of the systems have access to a high-performance shared filesystem that serves all the data. You can log in to any node and your environment will be the same. 4. Why can't I get to the cluster from roy?The data analysis cluster is in a different domain than 'roy'. You can get to the system directly from the main NCAR network or through gate.ucar.edu if coming in from an outside network. See the Quick Start Guide for more information. 5. What compiler suite do I use to build my Fortran, C and C++ APIs?We recommend that you use the Intel compiler suite: ifort (Fortran 77 & 90), and icc (C and C++). There are setup files available for each of these. See the Compiler Hints & Tips document for more information. Like many other common utilities, you'll find these compilers in /fs/local/bin. 6. What compiler suite do I use to build MPI applications?We do not currently support MPI applications on the cluster. Most data analysis and visualization applications are not dependent on MPI. If you wish to download and install MPICH in your own directory, you may do so. If you need to run an MPI based model, please ask about using lightning for this purpose. 7. May I run long-running jobs on the cluster?You may run data analysis jobs of any length on the cluster. If you have special requirements, please submit a trouble ticket using the CISL Customer Support system detailing your requirements. Please do remember that these machines are not designed for model runs. If you need to run a model on a Linux based platform, please ask about using lighting. 8. Does the cluster support /contrib for users?Yes. /contrib hosts software that we don't officially support on the systems, but that users are willing to support. If you have questions about how to make software available in /contrib, please submit a trouble ticket using the CISL Customer Support system. 9. What file systems are available?You can find a list of the file systems available on the cluster on the File System Layouts & Policy page. Some of these file systems are only mounted when they are in use, so a "df" command will show you ones that are currently in use, but will not show you all file systems that are available. 10. What commercial applications are available?We maintain a list of currently supported applications on the Supported Applications page. You will find these applications in /fs/local/bin. 11. How do I check my quotas?You can find the different quota commands used to check your quota usage on the Checking Your Quota Usage page. |
|
|