jbednar
commented
3 years ago I was able to reproduce the issue in a fresh environment created with conda create -c pyviz -n segfault holoviews bokeh matplotlib pandas datashader notebook:
$ python segfault.py
0
1
2
3
4
5
6
7
python(50040,0x112519dc0) malloc: tiny_free_list_remove_ptr: Internal invariant broken (next ptr of prev): ptr=0x7faf53bb10d0, prev_next=0x7fbf53bb10d0
python(50040,0x112519dc0) malloc: *** set a breakpoint in malloc_error_break to debug
8
9
10
11
12
13
14
15
Traceback (most recent call last):
File "/Users/jbednar/miniconda3/envs/segfault/lib/python3.8/site-packages/nbclient/client.py", line 816, in async_execute_cell
exec_reply = await self.task_poll_for_reply
File "/Users/jbednar/miniconda3/envs/segfault/lib/python3.8/asyncio/futures.py", line 260, in __await__
yield self # This tells Task to wait for completion.
File "/Users/jbednar/miniconda3/envs/segfault/lib/python3.8/asyncio/tasks.py", line 349, in __wakeup
future.result()
File "/Users/jbednar/miniconda3/envs/segfault/lib/python3.8/asyncio/futures.py", line 173, in result
raise exceptions.CancelledError
asyncio.exceptions.CancelledError
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "segfault.py", line 10, in <module>
output = ep.preprocess(notebook, {})
File "/Users/jbednar/miniconda3/envs/segfault/lib/python3.8/site-packages/nbconvert/preprocessors/execute.py", line 79, in preprocess
self.execute()
File "/Users/jbednar/miniconda3/envs/segfault/lib/python3.8/site-packages/nbclient/util.py", line 74, in wrapped
return just_run(coro(*args, **kwargs))
File "/Users/jbednar/miniconda3/envs/segfault/lib/python3.8/site-packages/nbclient/util.py", line 53, in just_run
return loop.run_until_complete(coro)
File "/Users/jbednar/miniconda3/envs/segfault/lib/python3.8/site-packages/nest_asyncio.py", line 98, in run_until_complete
return f.result()
File "/Users/jbednar/miniconda3/envs/segfault/lib/python3.8/asyncio/futures.py", line 178, in result
raise self._exception
File "/Users/jbednar/miniconda3/envs/segfault/lib/python3.8/asyncio/tasks.py", line 280, in __step
result = coro.send(None)
File "/Users/jbednar/miniconda3/envs/segfault/lib/python3.8/site-packages/nbclient/client.py", line 540, in async_execute
await self.async_execute_cell(
File "/Users/jbednar/miniconda3/envs/segfault/lib/python3.8/site-packages/nbconvert/preprocessors/execute.py", line 123, in async_execute_cell
cell, resources = self.preprocess_cell(cell, self.resources, cell_index)
File "/Users/jbednar/miniconda3/envs/segfault/lib/python3.8/site-packages/nbconvert/preprocessors/execute.py", line 146, in preprocess_cell
cell = run_sync(NotebookClient.async_execute_cell)(self, cell, index, store_history=self.store_history)
File "/Users/jbednar/miniconda3/envs/segfault/lib/python3.8/site-packages/nbclient/util.py", line 74, in wrapped
return just_run(coro(*args, **kwargs))
File "/Users/jbednar/miniconda3/envs/segfault/lib/python3.8/site-packages/nbclient/util.py", line 53, in just_run
return loop.run_until_complete(coro)
File "/Users/jbednar/miniconda3/envs/segfault/lib/python3.8/site-packages/nest_asyncio.py", line 98, in run_until_complete
return f.result()
File "/Users/jbednar/miniconda3/envs/segfault/lib/python3.8/asyncio/futures.py", line 178, in result
raise self._exception
File "/Users/jbednar/miniconda3/envs/segfault/lib/python3.8/asyncio/tasks.py", line 282, in __step
result = coro.throw(exc)
File "/Users/jbednar/miniconda3/envs/segfault/lib/python3.8/site-packages/nbclient/client.py", line 820, in async_execute_cell
raise DeadKernelError("Kernel died")
nbclient.exceptions.DeadKernelError: Kernel died
stuartarchibald
nvictus
As originally reported in https://github.com/holoviz/datashader/pull/939, the notebook getting_started/3_Interactive.ipynb sometimes fails to execute using either current Datashader master or the 0.11.1 release, reporting that the kernel has died. I have not yet been able to reduce it to a minimal example, but here's at least a cut-down one. First, create
segfault.ipynb, containing:Second, create
segfault.py, containing:The original error was found in CI on Ubuntu, but when I run this on my Mac OS X laptop, I get something like:
I can also reproduce it interactively within Jupyter, by doing
Restart and Run All, but sometimes it takes 10 or 20 tries to see any problem; usually it just nicely creates two Bokeh-based Datashader plots.The messages suggest that memory might be being modified after it was freed, which sounds consistent with this behavior (in that if the memory isn't yet being used for anything there's no harm in doing so). The issues may be Numba related, as so far I have not seen them when I do
env NUMBA_DISABLE_JIT=1 python segfault.py, but that's not conclusive, as I also don't happen to have seen them when I take "matplotlib" out of thehv.extension()call, which seems entirely unrelated to anything Numba is doing.