New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
scheduler and worker fail in PyPy3 #2734
Comments
That seems sensible to me. Would you be interested in submiting a PR with
that fix?
(Also, thanks for testing things on PyPy)
…On Thu, May 30, 2019 at 11:20 AM Daniel Cronce ***@***.***> wrote:
Python version: Python 3.5.3 (89428233efed, Apr 12 2018, 16:18:00)
Platform: pypy3-5.10
OS: Fedora 28 5.0.16-100.fc28.x86_64
Calling dask-scheduler or dask-worker fails due to GC tuning from #1653
<#1653>.
***@***.*** ~]$ dask-scheduler
Traceback (most recent call last):
File "/usr/lib64/pypy3-5.10/bin/dask-scheduler", line 10, in <module>
sys.exit(go())
File "/usr/lib64/pypy3-5.10/site-packages/distributed/cli/dask_scheduler.py", line 245, in go
main()
File "/usr/lib64/pypy3-5.10/site-packages/click/core.py", line 764, in __call__
return self.main(*args, **kwargs)
File "/usr/lib64/pypy3-5.10/site-packages/click/core.py", line 717, in main
rv = self.invoke(ctx)
File "/usr/lib64/pypy3-5.10/site-packages/click/core.py", line 956, in invoke
return ctx.invoke(self.callback, **ctx.params)
File "/usr/lib64/pypy3-5.10/site-packages/click/core.py", line 555, in invoke
return callback(*args, **kwargs)
File "/usr/lib64/pypy3-5.10/site-packages/distributed/cli/dask_scheduler.py", line 141, in main
g0, g1, g2 = gc.get_threshold() # #1653
Commenting out these lines solves the problem. Maybe include a check using
platform.python_implementation() before attempting to tune the GC.
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
<#2734?email_source=notifications&email_token=AACKZTEKR3A7HRLPZBAFXX3PX75E7A5CNFSM4HRGRIJKYY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4GWYMU6Q>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AACKZTETI2V2OQ6L4XJMXDLPX75E7ANCNFSM4HRGRIJA>
.
|
Yeah, should be able to. Just gotta play with getting dev environment up. |
I'm having difficulties setting up the dev environment on a clean VM (Fedora Server 29, PyPy 6.0.0 (3.5.3)). The current versions of numpy and PyPy have a bug that prevents installation of the dev-requirements.
After pinning numpy to 1.15.4, numpy and pandas successfully installed. However, running
I haven't been able to find information on how to solve this one. My pandas version and numpy do not seem to conflict and both were installed using
I could submit a PR, but I wouldn't be able to run the test suite with PyPy. I could still verify that the tests succeed with CPython, and that the worker and scheduler run with PyPy. How do we want to proceed? |
NumPy and pandas shouldn't be required. We may attempt to import them to register special serializers, but I believe all the tests requiring them should be skipped. |
Python version: Python 3.5.3 (89428233efed, Apr 12 2018, 16:18:00)
Platform: pypy3-5.10
OS: Fedora 28 5.0.16-100.fc28.x86_64
Calling dask-scheduler or dask-worker fails due to GC tuning from #1653.
Commenting out these lines solves the problem. Maybe include a check using
platform.python_implementation()
before attempting to tune the GC.The text was updated successfully, but these errors were encountered: