Problem interpolating variables from CFSV2 data

I recently downloaded the global CFSV2 data (DS094.0). The wget script helped me download many files. Using three types of files (*pgrbh.grb2, *sfluxgrbf.grb2 and *splgrbf.grb2) and a vtable, which is set to Vtable.CFSR included in the WPS/ungrib directory, ungrid and metgrid went successfully. But upon examination, I found that some variables in met*nc files (PMSL, PSFC, RH, etc) are not correctly interpolated from the CFSV2 data. This is very strange, because I had gone through almost exactly same steps to successfully run a different case a few months ago. Have anyone encountered this problem recently? thanks.

Do you have a plot (or some other way) to show the differences in values for these variables between the original CFSV2 data and the met_em* values? Can you also let me know which version of WPS you are using? Thanks!