distributed.worker - WARNING - Could not find data

[TonyHuBD]

Code Sample, a copy-pastable example to reproduce your bug.

feature_matrix, feature_defs = ft.dfs(
    entityset=es, 
    target_dataframe_name="acc",
    agg_primitives=["count", "sum"],
    trans_primitives=[ "MultiplyNumericBoolean"],
    cutoff_time=cutoff_times,
    cutoff_time_in_index=True,
    training_window="24 hour",
    max_depth=2,
    verbose=True,
    n_jobs = 36
    
)

Warning message

2024-05-18 18:25:09,703 - distributed.worker - WARNING - Could not find da
ta: {'bytes-e7e617a37c90e401634a701acdeac78d': ['tcp://127.0.0.1:41237', '
tcp://127.0.0.1:43061', 'tcp://127.0.0.1:46849', 'tcp://127.0.0.1:40099',
'tcp://127.0.0.1:46353']} on workers: [] (who_has: {'EntitySet-878e4a0191d
9aef3e784de699c988216': ['tcp://127.0.0.1:41237', 'tcp://127.0.0.1:43061',
'tcp://127.0.0.1:46493', 'tcp://127.0.0.1:40099', 'tcp://127.0.0.1:46353'
], 'bytes-e7e617a37c90e401634a701acdeac78d': ['tcp://127.0.0.1:41237', 'tc
p://127.0.0.1:43061', 'tcp://127.0.0.1:46849', 'tcp://127.0.0.1:40099', 't
cp://127.0.0.1:46353']})
2024-05-18 18:25:09,704 - distributed.scheduler - WARNING - Worker tcp://1
27.0.0.1:45941 failed to acquire keys: {'bytes-e7e617a37c90e401634a701acde
ac78d': ('tcp://127.0.0.1:41237', 'tcp://127.0.0.1:43061', 'tcp://127.0.0.
1:46849', 'tcp://127.0.0.1:40099', 'tcp://127.0.0.1:46353')}
2024-05-18 18:28:00,306 - distributed.worker - WARNING - Could not find da
ta: {'bytes-e7e617a37c90e401634a701acdeac78d': ['tcp://127.0.0.1:45941', '
tcp://127.0.0.1:41237', 'tcp://127.0.0.1:46849', 'tcp://127.0.0.1:46353']}
on workers: [] (who_has: {'bytes-e7e617a37c90e401634a701acdeac78d': ['tcp
://127.0.0.1:45941', 'tcp://127.0.0.1:41237', 'tcp://127.0.0.1:46849', 'tc
p://127.0.0.1:46353']})

I don't know why. I can't get the result when I use n_jobs.

[thehomebrewnerd]

@TonyHuBD n_jobs seems to be working as expected for me. Unfortunately, the information above isn't detailed enough to know what might be going wrong. If you are not using the most recent versions of Featuretools or Dask, my first suggestion would be to try to upgrade to the latests released versions and try again.

[BohdanBilonoh]

I can confirm that latest featuretools is not working with distributed>=2024.8.2 because of this line in featuretools and this func in the latest distributed. it could be fixed by passing client to Future like so

...
num_scattered_workers = len(
    client.who_has([Future(es_token, client=client)]).get(es_token, []),
)
...

@thehomebrewnerd please let me know what you think about the fix so I could make a PR