Random sequential adsorption (RSA) on a triangular lattice with defects is studied by Monte Carlo simulations. The lattice is initially randomly covered by point-like impurities at a certain concentration p. The deposited objects are formed by self-avoiding random walks on the lattice. Jamming coverage and percolation threshold are determined for a wide range of impurity concentrations p for various object shapes. Rapidity of the approach to the jamming state is found to be independent on the impurity concentration. The jamming coverage decreases with the impurity concentration p and this decrease is more prominent for objects of larger size. For a certain defect concentration, decrease of the jamming coverage with the length of the walk making the object is found to obey an exponential law, . The results for RSA of polydisperse mixtures of objects of various sizes suggest that, in the presence of impurities, partial jamming coverage of small objects can have even larger values than in the case of an ideal lattice. Percolation in the presence of impurities is also studied and it is found that the percolation threshold is practically insensitive to the concentration of point defects p. Percolation can be reached at highest impurity concentrations with angled objects, and the critical defect concentration pc is lowest for the most compact objects.