User Guide

User Guide

The information below is specific to our cluster.

General Information and Practices

• UMD Tier 3 computing cluster provides a login to Linux platform. To log in use SSH. More details on how to ssh to the cluster are given below.

• Do not try logging into individual machines. You should simply ssh to username@hepcms.umd.edu.

• Your working area on the cluster is intended for interactive work, so please refrain from running CPU intense jobs on these machines. Do not run any jobs interactively for more than 15 minutes. Any interactive jobs running longer than that period and/or taking up more than 70% of the memory will be automatically killed.

• Consider using CONDOR or CRAB batch services for longer and CPU intensive needs.

• Do not download any personal or entertainment material on the cluster.

• Your home areas are not for the storage of long term data. Ask for NFS or HADOOP disk space for such needs.

Where to get user help

To prevent spambots from harvesting emails, the phrase nospam has been inserted into email addresses. Simply remove it before sending email. If you've found a listserv or guide of general use and would like to see it here or if any links have expired, please email the System Administrators with your suggestion.

User help

1 Where to get user help?

New user

1 Request an account

2 Your first connection & password changes

3 The first time you login

4 If you need to change your password,

5 use the yppasswd command:

6 Enable your web directory

7 Troubleshooting your connection

Connect to the cluster

1 Connect to the cluster

1.1 From a Linux machine (Mac will need -Y option as well):

1.2 From a Windows machine:

2 Get Kerberos tickets for FNAL & CERN

Submit jobs to the cluster

1 Submit jobs to the cluster using Condor

2 Get your grid certificate & proxy, add CAs to your browser

2.1 Get your grid certificate

2.2 Get your proxy:

2.3 Add CAs to your browser:

3 Submit jobs to the grid using CRAB

3.1 Setup your environment:

3.2 Edit crab.cfg:

3.3 Edit crab.cfg for SE output:

3.3.1 To stage your data back to the hepcms SE:

3.3.2 To stage your data back to your user area in the FNAL SE:

3.4 Create, submit, watch, and retrieve your jobs:

3.5 Example:

3.6 Submit a CRAB job to the UMD hepcms cluster

Transfer data to/from the cluster

1 Transfer files to the cluster

1.1 From a Linux machine:

1.2 From a Windows machine:

2 Run CMSSW

3 More: CMSSW environment

4 Transfer data via PhEDEx

5 Transfer data via FileMover

6 Transfer data via the lcg-cp protocol

6.1 General srm-copy syntax:

6.2 Examples:

6.3 T3_US_UMD hadoop examples:

General Information

• • CMSSW:

    • Tutorials

    • Workbook

  • Root:

    • User's guide

    • Tutorials

    • Class index

    • Forum

• Condor

  • CRAB

    • Tutorial, generic

    • Tutorial @ FNAL

    • Documentation

    • Hypernews

    • CMS dashboard

    • Troubleshooting