upload is unhappy even though the shares are already distributed

[zooko]

Here's a test case that I added to src/allmydata/test/test_upload.py​:

    def test_problem_layout_0123_03_2_1(self):
        # Zooko stumbled on this case when investigating #1118.
        self.basedir = self.mktemp()
        d = self._setup_and_upload(k=2, n=4)

        # server 0: shares 0, 1, 2, 3
        # server 1: shares 0, 3
        # server 2: share 1
        # server 3: share 2
        # With this layout, an upload should just be satisfied that the current distribution is good enough, right?
        def _setup(ign):
            self._add_server_with_share(server_number=0, share_number=None)
            self._add_server_with_share(server_number=1, share_number=0)
            self._add_server_with_share(server_number=2, share_number=1)
            self._add_server_with_share(server_number=3, share_number=2)
            # Copy shares
            self._copy_share_to_server(3, 1)
            client = self.g.clients[0]
            client.DEFAULT_ENCODING_PARAMETERS['happy'] = 4
            return client

        d.addCallback(_setup)
        d.addCallback(lambda client:
            client.upload(upload.Data("data" * 10000, convergence="")))
        return d

When I run it, it fails like this:

[ERROR]: allmydata.test.test_upload.EncodingParameters.test_problem_layout_ticket_1118

Traceback (most recent call last):
  File "/Users/zooko/playground/tahoe-lafs/trunk/src/allmydata/immutable/upload.py", line 510, in _got_response
    return self._loop()
  File "/Users/zooko/playground/tahoe-lafs/trunk/src/allmydata/immutable/upload.py", line 363, in _loop
    self._get_progress_message()))
allmydata.interfaces.UploadUnhappinessError: shares could be placed on only 3 server(s) such that any 2 of them have enough shares to recover the file, but we were asked to place shares on at least 4 such servers. (placed all 4 shares, want to place shares on at least 4 servers such that any 2 of them have enough shares to recover the file, sent 4 queries to 4 peers, 4 queries placed some shares, 0 placed none (of which 0 placed none due to the server being full and 0 placed none due to an error))

Why does the upload not succeed? I added debugprints and here are some of them:

2010-07-18 09:54:15-0600 [-] response to allocate_buckets() from peer xgru5adv: alreadygot=(0, 1, 2, 3), allocated=()
2010-07-18 09:54:15-0600 [-] response to allocate_buckets() from peer ysbz4st7: alreadygot=(0, 3), allocated=()
2010-07-18 09:54:15-0600 [-] response to allocate_buckets() from peer b3llgpww: alreadygot=(2,), allocated=(1,)
2010-07-18 09:54:15-0600 [-] response to allocate_buckets() from peer rvsry4kn: alreadygot=(1,), allocated=(0,)
2010-07-18 09:54:15-0600 [-] server selection unsuccessful for <Tahoe2PeerSelector for upload t2uvf>: shares could be placed on only 3 server(s) such that any 2 of them have enough shares to recover the file, but we were asked to place shares on at least 4 such servers.: placed all 4 shares, want to place shares on at least 4 servers such that any 2 of them have enough shares to recover the file, sent 4 queries to 4 peers, 4 queries placed some shares, 0 placed none (of which 0 placed none due to the server being full and 0 placed none due to an error): sh0: rvsry4kn, sh1: b3llgpww+rvsry4kn, sh2: b3llgpww, sh3: ysbz4st7+xgru5adv