I'm using Albany w/ Trilinos to solve an elasticity problem with thermal expansion mismatch. I'm using block GMRES with MueLu preconditioning. It works for problem size of several million dofs, but when the problem size gets bigger, the true residual won't converge even though the preconditioned residual converges. I've tried different parameters for MueLu (sym Gauss-Seidel, Jacobi, ...), but it won't work. Any suggestions on the convergence problem?
For the problem size that converges with Block GMRES with MueLu, I tried Block GMRES with ILU preconditioner, and the true residual norm stays 1. Does that mean I have a singular preconditioner with ILU? Why did that happen?