Magnetic monopoles are solitons, i.e. field configurations on space (which is 3 dimensional). In pure $SU(N)$ gauge theory, magnetic monopoles can be constructed via 't Hooft's abelian projection (https://arxiv.org/abs/hep-th/0010225).
However, there is also a result that states that principle $SU(N)$ bundles over 3-manifolds are trivial. Doesn't this contradict the existence of monopoles?
Edit: Maybe it has to do with the fact that the gauge group should be $SU(N)/\mathbb Z_N$ in which case there are non-trivial bundles?