distributed.client error with atomgroup or universe type self attribute

[VOD555]

Distributed

An error is raised when one uses atomgroup or universe type self attribute with distributed.client dsak scheduler.

One example is rdf.py (PR #70 ), which had one line

self.ags = ags

And it raised one error;

______________________ test_same_result[distributed-2-1] _______________________

u = <Universe with 43480 atoms>
sels = [[<AtomGroup with 1 atom>, <AtomGroup with 2 atoms>], [<AtomGroup with 2 atoms>, <AtomGroup with 2 atoms>]]
n_blocks = 1, scheduler = None

    @pytest.mark.parametrize("n_blocks", [1, 2, 3, 4])
    def test_same_result(u, sels, n_blocks, scheduler):
        # should see same results from analysis.rdf.InterRDF_s
        # and pmda.rdf.InterRDF_s
        nrdf = rdf.InterRDF_s(u, sels).run()
>       prdf = InterRDF_s(u, sels).run(n_blocks=n_blocks)

/home/shujie/pmda/pmda/test/test_rdf_s.py:109: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
/home/shujie/pmda/pmda/parallel.py:359: in run
    res = blocks.compute(**scheduler_kwargs)
/home/shujie/anaconda3/envs/py36/lib/python3.6/site-packages/dask/base.py:156: in compute
    (result,) = compute(self, traverse=False, **kwargs)
/home/shujie/anaconda3/envs/py36/lib/python3.6/site-packages/dask/base.py:395: in compute
    results = schedule(dsk, keys, **kwargs)
/home/shujie/anaconda3/envs/py36/lib/python3.6/site-packages/distributed/client.py:2230: in get
    direct=direct)
/home/shujie/anaconda3/envs/py36/lib/python3.6/site-packages/distributed/client.py:1593: in gather
    asynchronous=asynchronous)
/home/shujie/anaconda3/envs/py36/lib/python3.6/site-packages/distributed/client.py:647: in sync
    return sync(self.loop, func, *args, **kwargs)
/home/shujie/anaconda3/envs/py36/lib/python3.6/site-packages/distributed/utils.py:277: in sync
    six.reraise(*error[0])
/home/shujie/anaconda3/envs/py36/lib/python3.6/site-packages/six.py:693: in reraise
    raise value
/home/shujie/anaconda3/envs/py36/lib/python3.6/site-packages/distributed/utils.py:262: in f
    result[0] = yield future
/home/shujie/anaconda3/envs/py36/lib/python3.6/site-packages/tornado/gen.py:1099: in run
    value = future.result()
/home/shujie/anaconda3/envs/py36/lib/python3.6/site-packages/tornado/gen.py:1107: in run
    yielded = self.gen.throw(*exc_info)
/home/shujie/anaconda3/envs/py36/lib/python3.6/site-packages/distributed/client.py:1469: in _gather
    traceback)
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

tp = <class 'distributed.scheduler.KilledWorker'>, value = None, tb = None

    def reraise(tp, value, tb=None):
        try:
            if value is None:
                value = tp()
            if value.__traceback__ is not tb:
                raise value.with_traceback(tb)
>           raise value
E           distributed.scheduler.KilledWorker: ('_dask_helper-0b781621-ac0a-4bda-b691-1507e7587020', 'tcp://127.0.0.1:36055')

/home/shujie/anaconda3/envs/py36/lib/python3.6/site-packages/six.py:693: KilledWorker
----------------------------- Captured stderr call -----------------------------
distributed.protocol.core - CRITICAL - Failed to deserialize
Traceback (most recent call last):
  File "/home/shujie/anaconda3/envs/py36/lib/python3.6/site-packages/MDAnalysis-0.19.1.dev0-py3.6-linux-x86_64.egg/MDAnalysis/core/groups.py", line 119, in _unpickle
    u = _ANCHOR_UNIVERSES[uhash]
  File "/home/shujie/anaconda3/envs/py36/lib/python3.6/weakref.py", line 137, in __getitem__
    o = self.data[key]()
KeyError: UUID('38eaec10-83dd-4648-aca7-5e16ce9f7196')

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/shujie/anaconda3/envs/py36/lib/python3.6/site-packages/distributed/protocol/core.py", line 131, in loads
    value = _deserialize(head, fs, deserializers=deserializers)
  File "/home/shujie/anaconda3/envs/py36/lib/python3.6/site-packages/distributed/protocol/serialize.py", line 178, in deserialize
    return loads(header, frames)
  File "/home/shujie/anaconda3/envs/py36/lib/python3.6/site-packages/distributed/protocol/serialize.py", line 57, in pickle_loads
    return pickle.loads(b''.join(frames))
  File "/home/shujie/anaconda3/envs/py36/lib/python3.6/site-packages/distributed/protocol/pickle.py", line 59, in loads
    return pickle.loads(x)
  File "/home/shujie/anaconda3/envs/py36/lib/python3.6/site-packages/MDAnalysis-0.19.1.dev0-py3.6-linux-x86_64.egg/MDAnalysis/core/groups.py", line 127, in _unpickle
    for k in _ANCHOR_UNIVERSES.keys()])))
RuntimeError: Couldn't find a suitable Universe to unpickle AtomGroup onto with Universe hash '38eaec10-83dd-4648-aca7-5e16ce9f7196'.  Available hashes: 
distributed.core - ERROR - Couldn't find a suitable Universe to unpickle AtomGroup onto with Universe hash '38eaec10-83dd-4648-aca7-5e16ce9f7196'.  Available hashes: 
Traceback (most recent call last):
  File "/home/shujie/anaconda3/envs/py36/lib/python3.6/site-packages/MDAnalysis-0.19.1.dev0-py3.6-linux-x86_64.egg/MDAnalysis/core/groups.py", line 119, in _unpickle
    u = _ANCHOR_UNIVERSES[uhash]
  File "/home/shujie/anaconda3/envs/py36/lib/python3.6/weakref.py", line 137, in __getitem__
    o = self.data[key]()
KeyError: UUID('38eaec10-83dd-4648-aca7-5e16ce9f7196')

As mentioned in issue #76, current solution is to avoid setting atomgroup or universe as self attribute.

version of dask

dask 0.19.4

[orbeckst]

See also PR #65