Hi!

I've been working with dask and dask-distributed for a few weeks and, after going through the documentation, I have a few doubts that I was hoping to clear:

1. Is is possible, in some manner, to do some sort of checkpointing/saving of the state of a very long-running graph of tasks that would allow a later restart?
2. Is it possible to modify in any way a graph while it is being computed?

Thanks!