I am trying to debug performance issues with a perl file that runs a snippet of code similar to psuedo code like below. This script is running on a Unix machine with a db connection to a local Oracle 11.2.0 database.

    my $sth1 = $localDB->prepare("< Some select query that returns mor
+e 100000 records (1 column) >");
    my $sth2 = $localDB->prepare("< Inner select query that uses the v
+alues (id) from the previous sql >");
    
    $sth1->execute();
    my ($id);
    $sth1->bind_columns( undef, \$id);
    
    while ( $sth1->fetch() )
    {
        ...

        $sth2->execute($id);
        my $dblinkname;
        while (($dblinkname) = $sth2->fetchrow_array())
        {
            my $dblink_sql = "<Innermost select query that dynamically
+ creates a sql query with $dblinkname>"); 
E.G. <select * from table@dblinkname;>
            eval {
                my $dblink_sth = $localDB->prepare($dblink_sql);  <-- 
+INV CURSOR HERE
            
                $dblink_sth->execute($id);                        <-- 
+EXECUTE FAILED HERE
                
                ...

                $dblink_sth->finish();
            };
            
            if ($@) {
                <log errors>
                next;
            }        
        }
        ...
        $sth2->finish();
        ...
    }
    $sth1->finish();
[download]

There are some machines that do not exist and executing the dblink query for those sql statements would throw the database not reachable error. However, every time I run the perl file the prepare statement marked above throws an invalid cursor (ORA-01001: invalid cursor) error and some times the execute statement right below throws an execute failed (DBD::Oracle::st fetchrow_array failed: ERROR no statement executing). I am currently at a loss on how to troubleshoot this script. Any pointers would be greatly appreciated.