This is a route you can take.
Take the components one by one. Let $G_i$ denote the event that the $i$-th component is good and let $F_i$ denote the event that is faulty.
Then you are looking for: $$P(G_1G_2G_3\cup G_1G_2F_3\cup G_1F_2G_3\cup F_1G_2G_3)$$
(Here $AB$ stands for $A\cap B$.)
Remark that the $4$ events in this union exclude eachother. So?...
Also e.g. $P(G_1G_2F_3)$ can be calculated as: $$P(G_1)P(G_2\mid G_1)P(F_3\mid G_1G_2)$$
Practicizing this makes it more simple.
Compare $P(G_1G_2F_3),P(G_1F_2G_3),P(F_1G_2G_3)$ with eachother and try to find a pattern. Wonder why. Thinking about it improves your intuition.