While using exo2nek to process a Tetra+Wedge mesh generated by ICEM.

I found that some part of the boundary surface covered by the prism mesh (wedge for boundary layer), can not be read, i.e. when exo2nek write out the involved boundary numbers, the part covered by prism mesh (wedge) can be missing. If you ignore this can just run the case using that mesh. The results will just be a lot of zeros.

I can go around by manually converting all element to tetra in ICEM. Yet just in case of any further problem. I put an issue here first.