It is an interesting case. How is your convergence doing? What are the values of your residuals to reach convergence? Have you tested mesh independence of your results? (Check link below)
Do you have a specific mesh method in place? Do you have any inflation applied to your mesh? Some help about meshing at second link
very low Re range is a tricky issue. From your results it seems that you have some wall effect, please try to apply a slip wall condition on your mesh boundary and to consider a large geometry around the cylinder. A space of 5d or 10d should be enough (d is the diameter of the cylinder). Another secret is to increase the number of cells around the object.
Asymmetric around the cylinder is common for real fluid. Especially for unsteady case it is very important to take into consideration the vortex shedding and the Strouhal number. Please, consider a time step 100 time lower than the characteristic vortex shedding time. In attach you can find a ppt of mine about these issues (sorry for the italian!)
At low Re, you are unlikely to get vortex shedding, but two counter-rotating stationary vortices often called Fӧppl vortices. If you still get oscillations and instability at low Re numbers, you might consider the issues raised in the above answers.