The headline says it all. You really need to be frustration tolerant in this business, even if there weren't any politics going on at the moment, to get all worked up about.
So I've submitted this paper where I've blocked the function of parts of the fly brain ( Drosophila) with a genetic tool that requires two components, a driver and an effector. I get the effect with two driver lines which drive the effector in the same brain area, the mushroom-bodies. So one of the referees asks me to use a driver line for outside of the mushroom-bodies as a control for the specificity of the genetic manipulation. I think that's a really good control, actually, one I should've done myself before submission! So I go and have friends send me two different lines for another part of the brain, the fan-shaped body. I had no problems whatsoever with the mushroom-body lines and the fan-shaped body lines have been tested before with a slightly different system, but very similar. So I'm quite surprised to find that one line doesn't fly and another flies really, really poorly. Which is a Very Bad Thing(TM) because my setup involves flying (flight simulator)! All this sending of lines, crossing and everything takes weeks before I can start the experiments. So I think I bite the bullet and use the less bad of the two lines. 2 months later, it turns out that the effect is not there, as expected. That's good. But while the data are significant, the values are so low, that I don't find them all that convincing. And if I don't find them convincing, the referee won't either. So now I've lost several months chasing a decent control line which in the best case doesn't yield any new insights and in the worst case completely destroys the main thrust of the paper.
And this is exactly how science almost always is! And the few times when it's not like that is when you get addicted