hi ppl,

i have written this subroutine to check the status of my caster (3 nodes + frontend). so if some application needs to see what other proc are running on the cluste it run's the sub :

##################################################
sub _ganglia_info {
##################################################

my ($self,%arg) = @_;

my ($total_load,$total_proc) = (0,0);
my (%gngl_info,%free_slots);
foreach (@{$arg{ginfo}}){
    chomp(my $who = `whoami`);
    chomp(my $host = `hostname -s`);
    my $pstmt = ($_ eq $host) ? ("cat /proc/cpuinfo | grep processor.*
+: | wc -l") : ("ssh $who\@"."$_ cat /proc/cpuinfo | grep processor.*:
+ | wc -l");
    my $lstmt = ($_ eq $host) ? ("ps -eo pcpu r") : ("ssh $who\@"."$_ 
+ps -eo pcpu r");
    my $usrs = ($_ eq $host) ? ("ps -eo user r") : ("ssh $who\@"."$_ p
+s -eo user r");
    my $rnprc = ($_ eq $host) ? ("ps -eo cmd r") : ("ssh $who\@"."$_ p
+s -eo cmd r");
    chomp(my $proc = qx($pstmt));
    chomp(my $tmp_load = qx($lstmt));
    chomp(my $rhsts = qx($usrs));
    chomp(my $rjob = qx($rnprc));
    my @hsts = split("\n",$rhsts);
    my @job = split("\n",$rjob);
    my %tmp_hsh;
    @tmp_hsh{@hsts} = @job;
    $gngl_info{$_} = \%tmp_hsh;
    my @array = split("\n",$tmp_load);
    my $tmp_load_s = 0;
    foreach (@array){
        $tmp_load_s += $_ if ($_ =~/\d+\.\d+/g);
    }
    $total_proc += $proc;
    $free_slots{$_} = int($proc - ($tmp_load_s / 100 ));
    $total_load += (100 - ($tmp_load_s / $proc ));
}
$total_load = $total_load / @{$arg{ginfo}};
return [$total_proc,$total_load,\%free_slots,\%gngl_info];

}
[download]

retrieves the results :

Total numb of proc:
Total load %:
Number of free(allegedly) cpu's per node:
who is running which proc on which node:
[download]

im not concerned if a application running takes only 10% of my cpu , if that is the case i'll treat this cpu as a taken one.

so the problem is : it seams i can ssh into a nod and run ps and retrieve the right results for avg. load and number of free cpu's per node but i can't get the results for

who is running which proc on which node: -> ps -eo user r and ps -eo cmd r

i am retrieving partial results(example: if there are 3 proc (100%) running on node 1 i'll get for 4CPU machine that 1 CPU is free but it will also report only one-two users running one-two proc's and there should be 3)

can anybody help me with this one ? is there a module that does that?

thnx