So I was pondering about geodesic deviations and I'm confused about the following. Let's say I have $3$ geodesics $\gamma_1(t)$ , $\gamma_2(t)$ and $\gamma_3(t)$. I introduce a parameter $s$ such that $\gamma_{ij}(t,s)$ obeys:
$$ \gamma_{12}(t,0) = \gamma_1$$ $$ \gamma_{12}(t,1) = \gamma_2$$ $$ \gamma_{23}(t,0) = \gamma_2$$ $$ \gamma_{23}(t,1) = \gamma_3$$ $$ \gamma_{31}(t,0) = \gamma_3$$ $$ \gamma_{31}(t,1) = \gamma_1$$
Now, I know in the geodesic deviation $\zeta^\mu_{ij}$ between $\gamma_i$ and $\gamma_j$. In a flat spacetime intuitively I imagine the relation locally:
$$ \zeta^\mu_{12} + \zeta^\mu_{23} + \zeta^\mu_{31} = 0 $$
Similarly the relative velocities obey when summed around in a triangle obey:
How does this get modified in curved spacetime?
In other words, For each pair of $\gamma_i,\gamma_j$ I pick a congruence they're part of and get a deviation vector field $\zeta^\mu_{ij}$ on the inside of the triangle. Now I ask asking what happens when you sum these three vector fields in curved space time?