Using action-based distribution function for the dynamical model of the Milky Way (MW) we have estimated its total mass and its density profile. Constraints are coming from the globular cluster proper motions from Gaia EDR3, from the rotation curve based on Gaia DR2 data, and from the vertical force data. We use Bayesian Markov chain Monte Carlo method to explore the parameters, for which the globular cluster distribution function and the Galactic potential are fully constrained. Numerical simulations are used to study the uncertainties on the potential constraint if considering a possible massive Large Magellanic Cloud (LMC). We found that a massive LMC (1.5 × 10¹¹ M_⊙) will affect the MW mass measurement at large radius, which includes both the MW and the LMC. We also use the FIRE2 Latte cosmological hydrodynamic simulations to make mock data set from an MW-like galaxy that includes many unrelaxed substructures. We test the effect of these unrelaxed substructures on the final results, and found that the measured rotation curve fluctuated around input value within 5 per cent. By keeping a large freedom in choosing a priori mass profile for both baryonic and dark matter leads a total mass of the MW that ranges from $5.36_{-0.68}^{+0.81}\times 10^{11}$ M_⊙ to $7.84_{-1.97}^{+3.08} \times 10^{11}$ M_⊙. This includes the contribution of a putative massive LMC and significantly narrows the MW total mass range published earlier. Such total mass leads to dark matter density at solar position of $0.34_{-0.02}^{+0.02}$ GeV cm^-3.