# Obtaining the dual variables of an optimisation problem with second order cone constraints solved using Gurobi

I'm solving an optimisation problem with linear and second order cone constraints, in other words the problem is convex and should have dual variables. I'm using Julia with JuMP to formulate the optimisation problem and I'm solving the problem with Gurobi. From the JuMP documentation, it should be possible to obtain the dual variables using the function dual. However I obtain NaNs when I do this while a linear version of my problem returns the duals as expected. Can anyone explain to me why this is and possible work arounds?