But this time fibration sequences are not allowed [I didn't impose this condition].
anyways, I was thinking: f o h= g so passing onto cohomology: h* o f*= g* and on S^{2n+1}, g* being nullhomotopic so 0 so h*o f*=0 on S^{2n+1}.
this gives j* o pi* o f* =0 whence j* o (f o pi)*=0
But since f o j= id, we have j* o f*= id
this contradicts the earlier equality.
But there must be something wrong in it. :(