There are attempts in the literature to theoretically explain the large breaking of isotopic invariance in the decay $\eta(1405)$\,$\to$\,$f_0(980)\pi^0$\,$\to$\,$ 3\pi$ by the mechanism containing the logarithmic (triangle) singularity, i.e., as being due to the transition $\eta(1405)\to(K^*\bar K+\bar K^*K)\to(K^+K^-+K^0\bar K^0)\pi^0\to f_0(980)\pi^0\to3\pi$. The corresponding calculations were fulfilled for a hypothetic case of the stable $K^*$ meson. Here, we show that the account of the finite width of the $K^*$ ($\Gamma_{K^*\to K\pi}\approx50$ MeV) smoothes the logarithmic singularities in the amplitude and results in the suppression of the calculated decay width $\eta(1405)\to f_0(980) \pi^0\to3\pi$ by the factor of $6-8$ as compared with the case of $\Gamma_{K^*\to K\pi}$\,=\,0. We also analyze the difficulties related with the assumption of the dominance of the $\eta(1405)\to(K^*\bar K+\bar K^*K)\to K\bar K\pi$ decay mechanism and discuss the possible dynamics of the decay $\eta(1405)\to\eta\pi \pi$. The decisive improvement of the experimental data on the $K\bar K$, $ K\pi$, $\eta\pi$, and $\pi\pi$ mass spectra in the decay of the resonance structure $\eta(1405/1475)$ to $K\bar K\pi$ and $\eta\pi\pi$, and on the shape of the resonance peaks themselves in the $K\bar K\pi$ and $\eta\pi\pi $ decay channels is necessary for the further establishing the $\eta(1405)\to3 \pi$ decay mechanism.