An 60Fe peak in a deep-sea FeMn crust has been interpreted as due to the signature left by the ejecta of a supernova explosion close to the solar system 2.8 +/- 0.4 Myr ago [Knie et al., Phys. Rev. Lett. 93, 171103 (2004)]. To confirm this interpretation with better time resolution and obtain a more direct flux estimate, we measured 60Fe concentrations along a dated marine sediment. We find no 60Fe peak at the expected level from 1.7 to 3.2 Myr ago. However, applying the same chemistry used for the sediment, we confirm the 60Fe signal in the FeMn crust. The cause of the discrepancy is discussed.