You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We have occasionally seen what might be inter-worker communication issues of unclear origin. Here's a small setup to possibly reproduce those. The sleep is purely there for the tasks to be long enough for auto-scaling to work and can be removed otherwise.
importdaskimportdask.bagimportnumpyasnpimporttimeNUM_TASKS=512x=np.linspace(1, NUM_TASKS, NUM_TASKS)
b=dask.bag.from_sequence(x, npartitions=len(x))
defblackbox(num):
time.sleep(4)
returnnumfutures=b.map(blackbox)
task=futures.fold(lambdax,y: x+y, split_every=8)
res, *_=dask.compute(task, scheduler=client)
res, sum(x) # result should be sum(x)
The text was updated successfully, but these errors were encountered:
We have occasionally seen what might be inter-worker communication issues of unclear origin. Here's a small setup to possibly reproduce those. The
sleep
is purely there for the tasks to be long enough for auto-scaling to work and can be removed otherwise.The text was updated successfully, but these errors were encountered: