I ran into the contradiction when I was trying to explicitly show that $D_6$ and $S_3$ are isomorphic following the outline given in my textbook. I couldn't resolve it, despite spending considerable amount of time. I also couldn't find the answer to my question here.
I am showing the contradiction I have stumbled upon in the following image:
EDIT:
I made a simple mistake in the image above. The first permutation should be $(321)$. Then everything checks out and there is no contradiction.

Actually, you just made a simple mistake. Your first permutation (the rotation) should be $$ \begin{bmatrix}1&2&3\\3&1&2\end{bmatrix} = (132) $$ With this, the equation checks out.
So you shouldn't think what you wrote in a comment: "I incorrectly assumed that every symmetry of equilateral triangle should correspond to a geometrically motivated permutation." Actually you were right from the beginning. The isomorphisms between $D_6$ and $S_3$ are given naturally by the way the geometric transformations permute the numbered corners. ("Isomorphisms" in plural because we have freedom in the way we number the corners.)
Of course, to formally prove they are isomorphic, you need to get a bit more technical, like in the other answers. A way to formalize this conceptual argument is to note/show that $D_6$ acts faithfully on the three corners. This implies that the action gives rise to an injective homomorphism $D_6\to S_3$, and since the groups are of equal finite order, it's an isomorphism.
Edit: Woops, now I’m the one who messed up! I fixed the cycle notation, should be fine now.