You must define restraints between the two molecules, not only on the protein side, otherwise the first docking stage (rigid body) will fail and return such an error. Do you have any idea where the protein binds? Major or minor groove? Some specific DNA sequence? etc …
If you do not have any clue about the DNA binding, you can specify the whole DNA sequence as passive. This remains suboptimal regarding HADDOCK’s philosophy but it would fix your error.
Another remark, you probably can make some educated guess on the protein side for the putative DNA binding site (positively charged patch of residues at the surface, exposed arginine(s), conserved residues), and refine CPORT predictions.