Python Nodes: Fail to execute when run concurrently


#1

Sometimes when running Python nodes at the same time it can happen that a node just fails with “stopped executing”. I wasn’t able to pin-point the exact issue but I think it’s when 2 nodes at the same time are loading / outputting data, then one of the fails.

Example:

WARN Python Learner 0:579:0:423 Execution canceled

And log file doesn’t say anything more.

So there for sure something is going wrong (not 100% stable) with the python nodes.


#2

Hi @beginner,

Thanks for letting us know. Just to make sure: Which OS are you using and which KNIME version? We’ll try to track down and fix the problem.

Best,

Christian


#3

Windows Server 2012 R2 (64-bit)

KNIME 3.5.2

With apache arrow for serialization

Reproduction is non-trivial as I’m not sure what causes it.


#4

Error also happens with flat buffers as serialization.


#5

Hi @beginner,

we were able to reproduce the issue (with some pyhton nodes in a parallel chunk loop) and work on a solution.Thanks again!

Best,
Lukas


#6

Similar issue occurs with Python node (Python 3.6) that is inside of a loop. The node spontaneously gets canceled.
Systems: Windows 8.1 Enterprise
KNIME Version: 3.6.0

Console details, at point of error:

ERROR PythonKernel C:\Program Files\Python36\lib\site-packages\sklearn\utils\validation.py:578: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().
DEBUG PythonKernel y = column_or_1d(y, warn=True)
WARN Python Script (1⇒1) 0:3100:3219:3119 Execution canceled

If I re-run the node, it works…