--DocumentContent--
NAVIGATION MENU
-------------------
User Menu
Home
FAQ
News
Known Problems
Message Board
The Project
The Status of Replication Server
Globus wrappers
General Scheme
Job Status panel (authorization required)
PHENIX Grid Status
Mailing Lists
Data disk space status
Backed UP data history
 
Manager Menu
Changes LOG
Cron Menu
Manage Board
Manage Mailing Lists

RAMData FAQ

Jérôme LAURET & Andrei SHEVEL

  1. What is RAMData ?
    RAMData is the name of a computing installation consisting of a range of peripheral machines (mix: Intel PIII/500MHz, 0.5GB main memory, dual processor, Intel P4/3.06GHz, 1 GB mainmemory) under Linux Scientific Linux 3.03
    The main server name is ram11.i2net.sunysb.edu.The main machine is running under OS Fedora 3.
    Peripheral machines have the names in the form ram??.i2net.sunysb.edu where ?? stands for a number between ranges [11,18] [21,28] [31,38] and [41,48].

    Due to security reason ram11 has some limitation for logging. You may log from the subnetwork .chem.sunysb.edu. Also it is possible to log to ram11 from BNL. All other sites have to be discussed separately.

  2. What was RAMData created for ?
    RAMData was created mainly for physics analysis in collaboration with the RHIC/Phenix group (BNL experiment). The facility was designed though to be a Data processing facility ...

  3. How to be registered as user on RAMData ?
    To become a user you have to send an official request for registration. To do this, log on ram11 under the account registry with password chem.sunysb.
    Remark 1. The dot at the end of password is part of password.
    After you should answer to all the questions and your request will be sent automatically. One, two days later, the administrator will notify you about the registration.
    Remark 2. You need to log to be registered from SUNYSB campus.

  4. How to connect to RAMData ?
    You MUST first connect to the main server ram11.i2net.sunysb.edu using the Secure Shell command (ssh). telnet connections from outside are not allowed. Connecting directly to any individual Linux PC is not allowed. Along the same line, you cannot use ssh, rcp or rsh from outside to the main server but can use ssh, scp or sftp.

  5. Why this ? is there anything else I should know ?
    We require that all users go through the main server for security reasons. telnet connections are not allowed from outside in but allowed from inside out. This also applies to services such as ftp, rcp etc.

  6. I tried passwd ; it did not work ! How Do I change my password ?
    Our facility uses NIS (Network Information System) to consistently spread the passwords on all machines. Therefore, you must use the yppasswd command to modify your NIS password.

  7. What is the volume and organization for disk space ?
    We have several directories which are located on the main server ram11 where a RAID array (about 4TB) is available. The RAID array is currently divided in three parts :
    • /Ram directory tree is devoted to general application software and is user readable only.
    • /Users directory tree to keep the users home subdirectories
    • /Data directory tree to keep the experimental data.

  8. Does the facility have access to afs ?
    Yes. Open afs is installed on all peripheral machines. ram11 has no Open afs installed.

  9. I am Phenix User. How do I get my phenix-login loaded ?
    The usual way. You can either
    • source /opt/phenix/bin/phenix_setup.csh for the afs copy of the software (linux only). Beware !! Connection between our facility and BNL is not currently the best ever. Nothing we can do about this.
    • source /Ram/afs/rhic/phenix/login/phenix_setup.csh will setup your environment to use the local copy of the RHIC/Phenix software.

  10. When/where do I have to do this ?
    After registering, template login files will be installed in your directory named /Users/$username. Look in your .login file : there are some comments and the above instruction are commented. Uncomment what is more suitable for you.

  11. But how often is the local Phenix software copy updated ?
    The local copy is currently updated during the week-end.

  12. Is there a place where I can put my data ?
    • As a Phenix user, you have full access to the tree /Data/data01.
    If you belong to another group not listed here, please, contact the system administrator root@ram11.i2net.sunysb.edu with information on your requirements or leave a message on our mailing list ramdata-news@ram3.chem.sunysb.edu. The archive of the list is available at
        https://ram3.chem.sunysb.edu/majordomo/ramdata-news
    
        and when asked please use
    
         username = newslist
         pass     = Wil_Sbu;   (the ';' is part of the password)
    

  13. Is there some backup scheme which saves my files ?
    Yes. The /Ram and /Users are saved every night either as an incremental backup or as a full copy. /Data is not backed-up at the moment.

  14. Then, what would prevent me from stacking files on the /Users directory
    Simple, quota are enabled on /Users (user by user basis). This is not the case on /Data (this partition may be later group restricted but there are no plans to limit users).

  15. Is there a batch system available on the facility ?
    Yes. Now we use torque (clone of well known PBS) and maui (both from http://supercluster.org). All the useful commands are described in the Frequently Used PBS Commands. You can also get lots of help using the local manpages. For example, man qsub or man qstat (to, respectively, have information on how to submit a job and how to check the current queue status).
    Additional information about PBS, Torque, other batch systems, and many related topics concerning computer cluster technology is available at www.ClusterGate.RU.

  16. Great ! But I don't want to read the doc. So, how do I start a simple job ?
    qsub job_script_name for queing job_script_name (a shell script written in any language you want).

  17. I would like to start a job quickly and interactively. Can I do that ?
    Yes. You could do that by logging to the peripheral machine and start the script you like
    ssh -X ram41 for example.

    However if you need to get really not loaded machine it is better to use different algorithm:

    • qsub -I (you use this command only to reserve the CPU)
    • ssh -X ramxx (where ramxx is machine name you see from previous command).

  18. What is the main scheduling algorithm for PBS ?
    On any machine may not be more than two jobs at the same time. Also the less time you require for job run the more priority your job will have with all other equal parameters.

  19. I don't want to use the batch system, I would like to run interactively. Can I do that ?
    Yes you can but for the sake of all user's and to make sure that the resources are fairly shared, we suggest that you use the batch system. Interactive jobs can though be started easily. Just log onto ram11, use the installed term command (installed under your account when created) to open as many windows as required and then, ssh to any of the Linux-ram nodes (or better, use qsub -I to log onto the least used node in the facility).

  20. What should I do if I want to delete all submitted PBS jobs
    On all of our machine, we have added some scripts to manage and make PBS use easier. TO delete all of your submitted jobs, use the qdelall command.

  21. How to find help if I am in trouble with something ?
    You can send an Email to root at root@ram11.i2net.sunysb.edu. A more general way may be to leave a message on our mailing list.