***
He had me work on
a conjectural statement connected to a famous
conjecture of his.
He explained me that the existing ideas are not
sufficient, and it would be nice if I can learn
some tools from a related field. Then I should
try to apply ideas from that area to my problem.
I spent about two years learning the new tools that
he was suggesting and in the meantime I had also
proved some nice (small) result unrelated to his
problem.  
He told me that it would be about 25% of my thesis.
I was perfectly happy with that.

***
Then I started working on the suggested problem
and solved it. Initially he said that would be my
thesis. Then I wrote the authors of the conjecture
and found out that one of them had the problem solved
(about 4 months ahead of me)
and his preprint was already submitted to a journal.

***
As I said before, my adviser does not talk to anybody
and consequently he was not even aware of the status
of my problem. When I told him, he changed his mind
with respect to my graduation and gave me one more
year to solve some other question. He didn't give
me a specific one, realizing that I would have
little time to solve it.

***
Honestly, I was thinking of changing fields
at that point. Finally, the summer was a good time
for me to rethink my options and decided that
I could still do something. And I solved some
other small question and graduated.

***
Unfortunately, mathematics is a field where the
big fish eats the small fish. My advise for you,
if you happen to be in graduate school and read
this: don't choose advisors who are newcomers to
the US (I feel sorry to say that, I'm an international
person myself), are less active or have very few
connections. You may end up in the role of the small
fish.

***