New experimental storage system

Dear cluster users,

due to the recent performance problems with the storage system, we are currently testing a distributed file system for user data storage. Please note that the nodes c2-0 to c2-4 are now used as storage servers.

I moved all user data to a new storage. There should be no changes for you, except that the current experimental storage system does NOT provide any redundancy whatsoever and WILL loose all data in case of a hardware failure. As before, it is YOUR obligation to back up any data
that is critical to you.

I hope the dreadful performance of majorana will improve with the new system.

If you experience any problems with the new distributed storage system or have any questions, feel free to contact me. If you do not have any questions, feel free to buy me a beer.

Arne

Read More

Upgraded/Installed some packages, most notably R 2.8.1

List of upgraded packages:

  • R-2.8.1-3
  • R-devel-2.8.1-3
  • R-core-2.8.1-3
  • fftw3-3.1.2-5
  • fftw3-devel-3.1.2-5
  • gsl-1.10-10
  • gsl-devel-1.10-10
  • hdf5-1.6.8-1
  • hdf5-devel-1.6.8-1
  • icu-3.6-4
  • libicu-3.6-4
  • libicu-devel-3.6-4
  • libRmath-2.8.1-3
  • libRmath-devel-2.8.1-3
  • suitesparse-3.1.0-1
  • suitesparse-devel-3.1.0-1

List of newly installed packages (installed mostly because of the ever growing complexity of R):

  • perl-File-Copy-Recursive-0.35-1
  • bzip2-devel-1.0.2-13
  • libjpeg-devel-6b-33
  • libtermcap-devel-2.0.8-39
  • libtiff-devel-3.6.1-12
  • readline-devel-4.3-13
  • tcl-devel-8.4.7-2
  • tk-devel-8.4.7-2
  • netpbm-10.25-2
  • netpbm-progs-10.25-2
  • tetex-2.0.2-22
  • tetex-latex-2.0.2-22
  • tetex-dvips-2.0.2-22
  • tetex-fonts-2.0.2-22
  • psutils-1.17-23
  • dialog-1.0.20040731-3

Read More

How do I get mails about my jobs status?

From the manual:

The qsub -m command requests email to be sent to the user who submitted a job or to the email addresses specified by the -M flag if certain events occur. See the qsub(1) man page for a description of the flags. An argument to the -m option specifies the events. The following arguments are available:

  • b – Send email at the beginning of the job.
  • e – Send email at the end of the job.
  • a – Send email when the job is rescheduled or aborted (for example, by using the qdel command).
  • s – Send email when the job is suspended.
  • n – Do not send email. n is the default.

Use a string made up of one or more of the letter arguments to specify several of these options with a single -m option. For example, -m be sends email at the beginning and at the end of a job.

Thus, you can simply change your submit script like this:

#!/bin/bash
#$ -N NAME_OF_JOB
#$ -m a,s
#$ -M user1@ulb.ac.be,user2@gmail.com
#$ -cwd

This sends a mail to the two specified email addresses if the job is rescheduled or aborted. If you omit -M, the mail is send to the user who submitted the job.

Read More

Rules for the cluster room

Please observe these simple rules when going into the cluster room:

  1. THE DOOR OF THE CLUSTER ROOM HAS TO REMAIN LOCKED AT ALL TIMES
    Always lock the door. If you are the last one in the lab, check
    the door before leaving. If you changed the water tank, lock the
    door afterwards. I found the door open most of the time during
    the last weeks.
  2. NEVER LEAVE LIGHT OR THIN OBJECTS (F)LYING AROUND
    Light objects will be sucked in by the airconditioner or, even
    worse, by the computing nodes. This will overheat and eventually
    kill the nodes. Be careful if you take sheets of paper, plastic
    bags or anything of the like into the room.

Of course, the usual rules (switch of light and monitors etc…) apply as well.

Please make sure that you remember these rules whenever going into the cluster room.

Read More