Thank perl monks for your time and input. I have myself written a logic for it.

my @arr;
foreach my $key ( keys %hash ) {
    if ( ref $hash{$key} ne 'HASH' && ref $hash{$key} ne 'ARRAY' ) {
        if ( ( defined $key ) && ( $key eq 'script' ) ) {
            push( @arr, $hash{$key} );
        }
    }
    if ( ref $hash{$key} eq 'HASH' ) {
        foreach my $k ( keys %{ $hash{$key} } ) {
            my $str = $k . "=" . $hash{$key}{$k};
            push( @arr, $str );
        }
    }
}
print Dumper( \@arr );
my %freq = ();
my @unique = grep { !$freq{$_}++ } @arr;
print Dumper( \@unique );

Output : 

$VAR1 = [
          'FILE=fileA',
          'err=99',
          'foo.pm',
          'FILE_READ=fileB',
          'err=99'
        ];
$VAR1 = [
          'FILE=fileA',
          'err=99',
          'foo.pm',
          'FILE_READ=fileB'
        ];
[download]

my %freq = ();
my @unique = grep { !$freq{$_}++ } @arr;


You may be interested in List::MoreUtils::uniq:
    use List::MoreUtils qw(uniq);
    my @unique = uniq @arr;


---

Give a man a fish:  <%-(-(-(-<

> My only point was to explain why the OP's code did not work as expected

Look what the OP wrote (in a badly formatted way) is

Output from my code 
FILE=fileA
err=99
script:foo.pm
foo.pm
FILE_READ=fileB
err=99
err=20
testset:test1 

Output required
FILE=fileA
err=99
script:foo.pm
foo.pm
FILE_READ=fileB
err=20
testset:test1
[download]

I have the biggest problems seeing a code who could possibly produce err=99 and err=20 out of a hash.

He never gave us the full picture, we had no possibility to know what his real problem is and "what didn't work as expected."

And for completeness the "solution" he posted is IMHO still buggy, b/c it'll filter ALL duplicates, not only for key 'err'.

Cheers Rolf
_{(addicted to the Perl Programming Language and ☆☆☆☆ :)
Je suis Charlie!}