xorbitsai / xorbits

Scalable Python DS & ML, in an API compatible & lightning fast way.
https://xorbits.readthedocs.io
Apache License 2.0
1.11k stars 67 forks source link

BUG: AttributeError: Can't get attribute '_make_function' #615

Open smartguo opened 1 year ago

smartguo commented 1 year ago

Describe the bug

I'm using xorbits.init to connect cluster deploying on yarn, then get AttributeError: Can't get attribute '_make_function'

To Reproduce

To help us to reproduce this bug, please provide information below:

  1. Your Python version: 3.9.12
  2. The version of Xorbits you use: 0.4.2
  3. Versions of crucial packages, such as numpy, scipy and pandas: pandas==1.4.2
  4. Full stack of the error.
  5. Minimized code to reproduce the error.

Hadoop version: Hadoop 3.2.2

Code

import xorbits
import xorbits.remote as xr
import xorbits.pandas as pd
xorbits.init(address="http://10.***.***.42:13062")

def f(x):
    print("-1111----")
    return x+1

def multi_tasks():
    tasks = []
    for i in range(5):
        tasks.append(xr.spawn(f, args=(i,)))

    print("start")
    xorbits.run(tasks)
    r = [t.to_object() for t in tasks]
    print(r)
    print("completed")

multi_tasks()

Error message:

start
Metric is not initialized, please call `init_metrics()` before using metrics.
Traceback (most recent call last):
  File "/Users/***/workspace/xorbits/test.py", line 21, in <module>
    multi_tasks()
  File "/Users/***/workspace/xorbits/test.py", line 16, in multi_tasks
    xorbits.run(tasks)
  File "/Users/***/lib/anaconda3/lib/python3.9/site-packages/xorbits/core/execution.py", line 55, in run
    mars_execute(mars_tileables, **kwargs)
  File "/Users/***/lib/anaconda3/lib/python3.9/site-packages/xorbits/_mars/deploy/oscar/session.py", line 1709, in execute
    return session.execute(
  File "/Users/***/lib/anaconda3/lib/python3.9/site-packages/xorbits/_mars/deploy/oscar/session.py", line 1526, in execute
    execution_info: ExecutionInfo = fut.result(
  File "/Users/***/lib/anaconda3/lib/python3.9/concurrent/futures/_base.py", line 445, in result
    return self.__get_result()
  File "/Users/***/lib/anaconda3/lib/python3.9/concurrent/futures/_base.py", line 390, in __get_result
    raise self._exception
  File "/Users/***/lib/anaconda3/lib/python3.9/site-packages/xorbits/_mars/deploy/oscar/session.py", line 1661, in _execute
    execution_info = await session.execute(*tileables, **kwargs)
  File "/Users/***/lib/anaconda3/lib/python3.9/site-packages/xorbits/_mars/deploy/oscar/session.py", line 890, in execute
    task_id = await self._task_api.submit_tileable_graph(
  File "/Users/***/lib/anaconda3/lib/python3.9/site-packages/xorbits/_mars/services/task/api/web.py", line 228, in submit_tileable_graph
    res = await self._request_url(
  File "/Users/***/lib/anaconda3/lib/python3.9/site-packages/xorbits/_mars/services/web/core.py", line 274, in _request_url
    raise exc.with_traceback(tb)
  File "/data3/emr/yarn/local/usercache/***/appcache/application_1683536483423_116585/container_e04_1683536483423_116585_01_000002/mars_env/lib/python3.9/site-packages/xorbits/_mars/services/web/core.py", line 69, in wrapped
  File "/data3/emr/yarn/local/usercache/***/appcache/application_1683536483423_116585/container_e04_1683536483423_116585_01_000002/mars_env/lib/python3.9/site-packages/xorbits/_mars/services/task/api/web.py", line 75, in submit_tileable_graph
  File "/data3/emr/yarn/local/usercache/***/appcache/application_1683536483423_116585/container_e04_1683536483423_116585_01_000002/mars_env/lib/python3.9/site-packages/xorbits/_mars/utils.py", line 468, in deserialize_serializable
  File "xoscar/serialization/core.pyx", line 933, in xoscar.serialization.core.deserialize
  File "xoscar/serialization/core.pyx", line 913, in xoscar.serialization.core.deserialize
  File "xoscar/serialization/core.pyx", line 811, in xoscar.serialization.core._deserial_single
  File "xoscar/serialization/core.pyx", line 268, in xoscar.serialization.core.PickleSerializer.deserial
  File "xoscar/serialization/core.pyx", line 241, in xoscar.serialization.core.unpickle_buffers
AttributeError: Can't get attribute '_make_function' on <module 'cloudpickle.cloudpickle' from '/data3/emr/yarn/local/usercache/***/appcache/application_1683536483423_116585/container_e04_1683536483423_116585_01_000002/mars_env/lib/python3.9/site-packages/cloudpickle/cloudpickle.py'>
smartguo commented 1 year ago

By the way, I run the python script on my mac laptop, and when I run it on Linux server, It turns out ok.

smartguo commented 1 year ago

This is another example on my mac laptop. Code

import xorbits
import xorbits.remote as xr
import xorbits.pandas as pd
xorbits.init(address="http://10.***.***.42:13062")

option = {
    "key": "***",
    "secret": "***",
    "endpoint_url": "https://cos.ap-beijing.myqcloud.com",
}

df = pd.read_parquet(
    "s3://<bucket>/<path>/pt=20230403/00000.parquet",
    storage_options=option,
)
print(df.head())

Error message:

Metric is not initialized, please call `init_metrics()` before using metrics.
100%|█████████| 100.00/100 [00:01<00:00, 68.90it/s]
Traceback (most recent call last):
  File "/Users/***/workspace/xorbits/test_session.py", line 20, in <module>
    print(df.head())
  File "/Users/***/lib/anaconda3/lib/python3.9/site-packages/xorbits/utils.py", line 37, in inn
    return f(self, *args, **kwargs)
  File "/Users/***/lib/anaconda3/lib/python3.9/site-packages/xorbits/core/data.py", line 293, in __str__
    return self.data.__str__()
  File "/Users/***/lib/anaconda3/lib/python3.9/site-packages/xorbits/core/data.py", line 119, in __str__
    return self._mars_entity.__str__()
  File "/Users/***/lib/anaconda3/lib/python3.9/site-packages/xorbits/_mars/core/entity/core.py", line 102, in __str__
    return self._data.__str__()
  File "/Users/***/lib/anaconda3/lib/python3.9/site-packages/xorbits/_mars/dataframe/core.py", line 2214, in __str__
    return self._to_str(representation=False)
  File "/Users/***/lib/anaconda3/lib/python3.9/site-packages/xorbits/_mars/dataframe/core.py", line 2186, in _to_str
    corner_data = fetch_corner_data(self, session=self._executed_sessions[-1])
  File "/Users/***/lib/anaconda3/lib/python3.9/site-packages/xorbits/_mars/dataframe/utils.py", line 1242, in fetch_corner_data
    return df_or_series._fetch(session=session)
  File "/Users/***/lib/anaconda3/lib/python3.9/site-packages/xorbits/_mars/core/entity/executable.py", line 169, in _fetch
    return fetch(self, session=session, **kw)
  File "/Users/***/lib/anaconda3/lib/python3.9/site-packages/xorbits/_mars/deploy/oscar/session.py", line 1760, in fetch
    return session.fetch(tileable, *tileables, **kwargs)
  File "/Users/***/lib/anaconda3/lib/python3.9/site-packages/xorbits/_mars/deploy/oscar/session.py", line 1562, in fetch
    return asyncio.run_coroutine_threadsafe(coro, self._loop).result()
  File "/Users/***/lib/anaconda3/lib/python3.9/concurrent/futures/_base.py", line 445, in result
    return self.__get_result()
  File "/Users/***/lib/anaconda3/lib/python3.9/concurrent/futures/_base.py", line 390, in __get_result
    raise self._exception
  File "/Users/***/lib/anaconda3/lib/python3.9/site-packages/xorbits/_mars/deploy/oscar/session.py", line 1728, in _fetch
    data = await session.fetch(tileable, *tileables, **kwargs)
  File "/Users/***/lib/anaconda3/lib/python3.9/site-packages/xorbits/_mars/deploy/oscar/session.py", line 1019, in fetch
    fetched_data = await fetcher.get()
  File "/Users/***/lib/anaconda3/lib/python3.9/site-packages/xorbits/_mars/services/task/execution/mars/fetcher.py", line 60, in get
    fetched_data = await storage_api.get.batch(
  File "/Users/***/lib/anaconda3/lib/python3.9/site-packages/xoscar/batch.py", line 147, in _async_batch
    return [await self._async_call(*args_list[0], **kwargs_list[0])]
  File "/Users/***/lib/anaconda3/lib/python3.9/site-packages/xoscar/batch.py", line 96, in _async_call
    return await self.func(*args, **kwargs)
  File "/Users/***/lib/anaconda3/lib/python3.9/site-packages/xorbits/_mars/services/storage/api/web.py", line 138, in get
    return deserialize_serializable(res.body)
  File "/Users/***/lib/anaconda3/lib/python3.9/site-packages/xorbits/_mars/utils.py", line 469, in deserialize_serializable
    return deserialize(header2, buffers2)
  File "xoscar/serialization/core.pyx", line 862, in xoscar.serialization.core.deserialize
  File "xoscar/serialization/core.pyx", line 811, in xoscar.serialization.core._deserial_single
  File "xoscar/serialization/core.pyx", line 268, in xoscar.serialization.core.PickleSerializer.deserial
  File "xoscar/serialization/core.pyx", line 241, in xoscar.serialization.core.unpickle_buffers
AttributeError: Can't get attribute '_unpickle_block' on <module 'pandas._libs.internals' from '/Users/***/lib/anaconda3/lib/python3.9/site-packages/pandas/_libs/internals.cpython-39-darwin.so'>
ChengjieLi28 commented 1 year ago

After confirming with @smartguo, the client using the mac arm chip submits the spawn task to the linux x86 server. And the python env in linux is the whole env packed on mac. Cloudpickle may have some issues in this scenario.

smartguo commented 1 year ago

The python env in linux is the whole env packed on linux itself, it looks like that Cloudpickle don't support seder on different system or different arch. However, this scenario doesn't matter, using server's nookbook to develop is fine.