Hi Jerry-
Off the top of my head, I don't have an answer for any of your questions. However, the start_over script is an "at your own risk" utility. I recommend starting with a fresh install if you want OSCAR 1.3. OSCAR 1.4 is in it's final beta though, and should be released any day now. I recommend waiting for that.
Jeremy
At 09:04 AM 10/7/2002 -0400, [EMAIL PROTECTED] wrote:
We upgraded from oscar-1.2 to oscar-1.3 a couple of week ago and we added two more nodes to the cluster for a total of 14 nodes. We ran start_over for oscar-1.2 to removed most of the oscar files and then re-booted the server. Can somebody try to answer these 4 questions:1. test_cluster_as_root comes up with error message: ROOT TESTS ---------- sshd service:7[60G[ 7[1;32mOK7[0;39m ] pbs_server service:7[60G[ 7[1;32mOK7[0;39m ] maui service:7[60G[ 7[1;32mOK7[0;39m ] /home mounted on 14 free nodes, need 14:7[60G[ 7[1;32mOK7[0;39m ] Preparing user tests...33333333333333333333333 33333333333333333333333generating ssh file /home/oscartst/.ssh/id_dsa ... Generating public/private dsa key pair. Saving the key failed: /home/oscartst/.ssh/id_dsa: No such file or directory. generating ssh file /home/oscartst/.ssh/identity ... Generating public/private rsa1 key pair. Saving the key failed: /home/oscartst/.ssh/identity: No such file or directory. generating ssh file /home/oscartst/.ssh/id_rsa ... Generating public/private rsa key pair. Saving the key failed: /home/oscartst/.ssh/id_rsa: No such file or directory. adding id to ssh file /home/oscartst/.ssh/authorized_keys2 adding id to ssh file /home/oscartst/.ssh/authorized_keys adding id to ssh file /home/oscartst/.ssh/authorized_keys2 USER TESTS ---------- Checking PBS default queue definition:7[60G[ 7[1;32mOK7[0;39m ] server -> node SSH:7[60G[7[1;31mFAILED7[0;39m] 2. What is the different between MPI-1.2.1 and MPI-1.2.4? Is the 2 versions of MPICH conflicting with each other? 3. Some of the system Oscar test fail with a message "no user oscartst". Oscar did set up a group called "oscartst". Simply MPICH = failed 4. Running the same ScalaPack test with MPIRUN (1.2.4) took over 35 hours (instead of 6 hour for a 3000 x 3000 matrix) and with deceasing results (453 Mflops down to 236 Mflops). WHY? Jerry Waterman U.S. EPA [EMAIL PROTECTED]
------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ Oscar-users mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/oscar-users </x-flowed>
