Regarding the PCA bi-plot, I see no major issue, assuming that you have generated this PCA bi-plot in an unbiased ('unsupervised') way using all genes. Can you share the code that you used? Your 2 groups (Control + Sample) are almost exclusively segregated along PC1. The sample at the bottom-right is behaving differently, but it is still not grouping with Control.
Then, in your second figure generated with pheatmap(), it seems that —yes— your groups are segregated perfectly via hierarchical clustering, and the heatmap colour shade also indicates this.
Regarding the gene naming issue, which species is this? Can you confirm how the read count quantification was performed and with which reference GTF? Generally, to help, please explain your broader analysis pipeline so that we can begin to try to solve this.