X-ray spectra of the O-type star ζ Puppis exhibit an anomaly in the 2p - 3s lines of Fe XVII, in which the (3G + M2) / (3F) line intensity ratio is approximately 1.4 for ζ Puppis, in comparison with ~2.4 for almost all other collisionally excited astrophysical spectra. The electron density and UV field dependence of the 2p - 3s line ratios were studied by Mauche et al (2001); based on their work, we expect that the strong UV field of ζ Puppis is responsible for the observed line ratio, and that the ratio can therefore be used as an independent diagnostic of plasma formation radius. We used the Flexible Atomic Code (FAC) collisional-radiative model to model the effect of UV photoexcitation from O-type stars on the Fe XVII transition lines, and we incorporate the results of these calculations into existing line profile models for O star X-ray spectra. We compared our model calculations to archival observations of coronal and hot stars taken by the Chandra HETG spectrometer, in order to test the validity of our calculations in the limit of low electron density and a range of UV field strengths. We examined the M2 / (3G + 3F + M2) line intensity ratio as a diagnostic of the quality of our line intensity calculations, and found a ~20% model-data discrepancy, which is consistent with discrepancies found in similar previous studies. However, our calculations do not accurately reproduce the 3F / (3G + 3F + M2) intensity ratio observed in coronal stars, and the dynamic range of the model ratio as a function of UV field intensity is not large enough to explain the observed difference in this ratio between the coronal stars and ζ Puppis. The large mismatch between the observed and modeled dynamic ranges of the 3F line strength is surprising, even given the known uncertainties in the model line intensities.