>SCF that converges or fails to converge is for the ground state. CIS procedure generates the excited states. Their energies should be no less than the ground state energy (variational principle). In your output, the energy of the first excited state decreased, while the ground state energy increased. In the step before SCF failed to converge, the energies of the ground and first excited states became close, so you might expect problems. Sooner or later, they would appear.
>As for root tracking in CIS, this can be necessary for the excited states, which can easily flip. The manual says: The state-tracking feature of the PC GAMESS' CIS code can be activated by selecting negative value of istate in the $cis group. It is intended for geometry optimization of the excited states in the case of root flipping. This is not necessary for assumed flipping of the ground and excited state, because they never flip, they can only become degenerate, and this means that you should use MCSCF for your system.
Thanks, that's much clearer. Believe it or not with DIIS enabled and SOSCF disabled I finished geometry optimizations successfully. I will have to examine the outputs for anything suspicious WRT first excited and ground state energies being close. I used state tracking so I assume it actually converged to first excited. Is there a way to check for that?