NFS default behaviour is blocking IOs, that means the program accessing the share may eventually be frozen and can't be killed. I guess to circumvent this you need forking, accessing data from child, and waiting from father:

sub checkfile {
    my $file = shift;
    my $duration = shift || 10;

    my $pid = fork();

    unless ($pid) {
        # this is run from child
        die "cannot access $file\n" unless -e $file;
  } else {
      # this is run from parent
    sleep $duration; 
    if ( kill 0, $pid) {
        warn "$pid still running after $duration seconds!";
        kill 9, $pid;
        return;
    }
    return 1
  }
}
[download]