The message can be correct, but not-all inclusive. Can you run a static (linear) analysis? Usually that can help isolate the problem. Alternatively, if the structure is buckling, it could be at the limit and you'd have to try the nonlinear solution using Displacement Control or Arc-Length control. If you can plot a response graph, it may help show what is happening to a point on the structure. If there are no-penetration contacts in the model, it can be very difficult to trouble-shoot, and you may have to restrict the areas where contact surfaces are coming together or moving apart (use the Split Line command to split surfaces and redefine the contact surfaces using the smaller areas). If you have friction turned-on, you might turn it off as a start. I have read from other analysts, that friction can be so problematic (due to more iterations and extra tangent, shear, or traction force calculations required), they just turn it off (usually that gives a conservative solution anyway).