From krishna.chivukula at navy.mil Tue Oct 9 14:07:56 2007 From: krishna.chivukula at navy.mil (Chivukula, Krishna CIV 470000D 476200D) Date: Tue Nov 9 01:14:29 2010 Subject: [scyld-users] NASTRAN: Compute Node question Message-ID: <3DFD6049DDA3D44DA6908353EE6D5C100BD612@nawechlkez03.nadsuswe.nads.navy.mil> Hello: We are trying to run NASTRAN on a Scyld beowulf cluster. The license authorization is provided by a Windows server on the same subnet. - No problem if the job runs on the master node. Problem is I can only run it on the master node... - Job will not run on a compute node. I use the following command. mach 37% bpsh 5 nast_perl (nast_perl is a perl script that runs nastran by executing system "nastran scr=yes nastran_example.bdf" inside the script. The compute node is unable to get license authorization. The message looks this: ==================================================================== MSC.Nastran V2005.0 (AMD Linux 2.4.21-40.EL.301u0030smp) Tue Oct 9 13:36:59 2007 *** USER FATAL MESSAGE (pgm: nastran, fn: validate_authorize) authorize=1700@xxx.xx.xxx.xx (Env: MSC_LICENSE_FILE) Host "xxx.xx.xxx.xx" is unknown in this network. ==================================================================== The xxx.xx.xxx.xx is the license server on the network. The idea is not to run Nastran in parallel mode (or a multi threaded version of it), just run a single instance of it on a compute node. The master node has no problem. I am not much of an admin. But it probably useful to mention the /etc/nsswitch.conf is configured as: bootparams: nisplus [NOTFOUND=return] files ethers: files bproc beo netnasks: files networks: files protocols:files bproc beo rpc: files services: files bproc beo ... Any pointers on what I should look for? -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.scyld.com/pipermail/scyld-users/attachments/20071009/a03a9630/attachment.html From jbernstein at penguincomputing.com Wed Oct 10 13:26:32 2007 From: jbernstein at penguincomputing.com (Joshua Bernstein) Date: Tue Nov 9 01:14:29 2010 Subject: [scyld-users] NASTRAN: Compute Node question Message-ID: <470D3578.2010108@penguincomputing.com> > Hello: > > We are trying to run NASTRAN on a Scyld beowulf cluster. The license > authorization is provided by a Windows server on the same subnet. > > - No problem if the job runs on the master node. Problem is I can only > run it on the master node... Ah! Stop right there. You'll need to enable IP forwarding on the compute nodes, so that the compute nodes reach can reach the license server through the headnode. If you are running Scyld 29cz-5 or later you can easily enable this by uncommenting the line in /etc/beowulf/config that reads: #ipforward yes so that it reads: ipforward yes Then, you'll need to quickly reboot the compute nodes and restart the "beowulf" service. It is important that the next two commands be run in quick succession. The first command reboots the nodes, the second, restarts the beowulf service. Be sure to restart the beowulf service before any of the nodes have come up. # bpctl -S all -R # /etc/init.d/beowulf restart You'll know it worked correctly, because at the end of the restart message you will see a line that says something like ""Setting up IP Forwarding:" -Joshua Bernstein Software Engineer Penguin Computing From vgregorio at penguincomputing.com Wed Oct 31 09:08:59 2007 From: vgregorio at penguincomputing.com (Victor Gregorio) Date: Tue Nov 9 01:14:29 2010 Subject: [scyld-users] Scyld ClusterWare 4.1.4 Release Announcement Message-ID: <4728A89B.9080707@penguincomputing.com> Penguin Computing is proud to announce the release of Scyld ClusterWare 4.1.4, the latest update to Scyld ClusterWare 4.1. Release Notes are available on MasterLink: http://www.penguincomputing.com/go/masterlink. Notable changes between ClusterWare 4.1.3 and ClusterWare 4.1.4 include: (1) The base kernel is upgraded to 2.6.9-55.0.9 The upgraded Kernel addresses these RedHat Security advisories: RHSA-2007:0937-2 https://rhn.redhat.com/errata/RHSA-2007-0937.html RHSA-2007:0774-4 https://rhn.redhat.com/errata/RHSA-2007-0774.html (2) TORQUE is upgraded to 2.1.9 TaskMaster 2.0.1 requires configuration changes to work with TORQUE 2.1.9. Scyld TaskMaster customers should contact Customer Support for details: http://www.penguincomputing.com/go/support. (3) Improved TORQUE integration with mpirun (4) Bproc enhanced to support the full range of process ID values (5) Updated Scyld ClusterWare Documentation The Scyld ClusterWare Installation Guide, Administration Guide, Programmers Guide, Reference Guide and Users Guide are published on MasterLink: http://www.penguincomputing.com/go/masterlink. For the most up-to-date product documentation and other helpful information about Scyld ClusterWare, visit the Scyld MasterLink Site. Penguin Computing 1-888-PENGUIN http://www.penguincomputing.com Scyld MasterLink: http://www.penguincomputing.com/go/masterlink Scyld Customer Support: http://www.penguincomputing.com/go/support