From Conrad's [notes][1] on rigid geometry:
> More specifically, Gabber has given an example of a sheaf of modules $F$ on the closed unit disk $B^1$ such that $F$ is locally a direct limit of coherent sheaves but with nonzero degree-1 sheaf cohomology, so
$F$ cannot be expressed as a direct limit of coherent sheaves over the entire affinoid space...

What is this example due to Gabber? Is there a reference describing it?
  [1]: http://math.stanford.edu/~conrad/papers/aws.pdf