You cannot order hash keys, by definition, their order is arbitrary.

What I understand by first pass means that for one key $host from %nodes, $rrd takes on the values from %rrd in thje way you expect, but it does not for the other keys in %nodes. I doubt that is what you mean ... or if it is, you are not showing the relevant coded, since %rrd is not modified in the code you present.

How about putting a print statement in the inner loop, to see what is happening?

for my $host(keys %nodes) {
  for my $rrd(keys %rrd) {
    printf "host: %s\trrd: %s\n", $host, $rrd;
#    my $avg = &get_latest_avg($host,$rrd,'free');
#    print $avg. "\n";
  }
}
[download]

--
TTTATCGGTCGTTATATAGATGTTTGCA

I understand what you're saying about the random order of hash keys. However, in my situation, its always the first one that succeeds, regardless of which keys its using. The rest always return -1 as the result of the return. I've tried adding hosts to the %nodes hash and always, the first host works, and the rest fail, no matter which host is selected in the first iteration of the for loop. Hosts that failed in a previous run, run properly if they are the first in line.

To plug in some values for the hashes I use (both are declared outside of the for loops), I'm adding this:

my %nodes = (
        'host1' => {
                'hardware' => 'servers',
                'community' => "$serverCommunity",
                },
        'host2' => {
                'hardware' => 'servers',
                'community' => "$serverCommunity",
                },
);
my %rrd = (
        'real_trend' => {
                'free' => $memAvailReal,
                'desc' => 'Physical Memory',
                },
        'swap_trend' => {
                'free' => $memAvailSwap,
                'desc' => 'Swap Memory',
                },
);
[download]

I just don't think these values matter that much since their order seems unimportant.

I somehow think the issue may be in the snmp session somehow however I have no clue as to why.

Thanks very much for your time on this -c

my ( $session, $error ) = Net::SNMP->session(
         -hostname       => $host,
         -version        => '1',
         -community      => $nodes{$host}->{community}, );

if ( !defined($session) ) {
        printf("ERROR: sub snmp session undef\n");
        }

my $num = $snmp->get_request( -varbindlist =>   [$rrd{$rrd}{$type}] )

if ( !defined($num) ) {
        printf( "ERROR: %s.\n", $session->error );
        }
[download]

For clarity i would suggest not using a global %rrd and a local $rrd in the same sub. You could rename one or the other to make it clear. At first i thought you where using a hash as a key in a hash, seemed strange. :-)

I take completely the opposite view. I see it as quite natural to use the same name for a hash or array and for individual elements. Generally, I try to make the collective object be plural, while the individial are sometimes localized with 'the_rrd' or 'a_rrd', but often the sigils are sufficient differentiation, especially if the scope is small.

--
TTTATCGGTCGTTATATAGATGTTTGCA

I'm in between you two (is ambivalence a problem for me? Maybe...). I'll end up using the plural and singular so I can say for my $host ( @hosts ) { code }, but I think using the exact same name could be confusing. Admittedly anyone who knows perl will know that $host and $host{key} are compeltely different variables, but why court confusion? Also, if you keep them separate, 'strict' and 'warning' can save you from typos or brainfarts.

--Bob Niederman, http://bob-n.com