|
I installed oscar2.3 on 3 HP 360G2 server(redhat linux8.0), I made
my new partition rule file cause the smart array 5i of HP give a
/dev/cciss/c0d0 to the scsi disk. so I think you should give a
$mount command to see what is the right name for the disks on your clients
and make your own partition rule file. when clients booting and got
the "clientnodename.sh", you may find it halts on errors like "can't find
/dev/xxxxxxx", so you mast check the /dev/ directory on your client to see
the file name of your hard disk again, it may be different from what you
see on the server. I am not familiar with searching and replacement in vi
editor, so I make some symbolic links to these device names in the
"clientnodename.sh", then, it works.
The clients are installed automatically by the server via PXE(I
don't see X installed on them), if PXE failed after clients get their IP
addresses and the kernels and initrds (clients hang up after displayed :
READY), you can try boot your clients by floppy disk(generated from
server) or try to connect client to server with a peer-to-peer
rj45 cable, for me, these two ways work well. I think the reason is my
D-link switch is not good enough to support tftp. after the tests, I'll
change it.
good luck!
below is my confusion, I compiled and run the tests in
/opt/mpich-1.2.5.10-ch_p4-gcc/examples,
# ./mpirun pi3
Process 0 of 1 is alive
Enter
the number of intervals: (0 quits)
5555555
pi is approximately: 3.1415926535898766 Error
is: 0.0000000000000835
Enter the number of intervals: (0 quits)
0
#
I have one server and 2 clients, why it says: Process 0 of
1 is alive, not Process 0 of 2 is alive? process 0 is on
the server, the processes 1 and 2 should on
clients? ?
I know this question seems stupid, but I
am new on Oscar, please let me know the reason.
webman | 
