Python module not found on the Server

KNIME Server
1

Hi,

I am using a Python script node after a Conda Propagation Node.
In the node, I use a module called requests.
This module is installed in the conda environment, and is available in the conda propagation node.

Depending on the execution, I obtain sometime a Module Not Found Error. Then, after a second run, this error might be solved.

Any idea to solve this issue ?

Thanks for your help
Regards
Jerome

2

Hi Jerome,

Could you please elaborate a bit?

What version of the KNIME Analytics Platform (KAP) did you use for the workflow processing?
In dependency of that: What Python integration have you used? On newer versions of the KAP you have the new Python extension (since 4.5 in the labs, with 4.7 moved into the regular extensions) as well as the legacy one.

In case that you are using the old Python integration you should have an Advanced section under “Python (legacy)” where you could define the number of prelaunched processes. To exclude caching problems it might help to set both values to 0 and check if the error still occurs afterwards.

Best,
Michael

3

Hi Michael,

Thanks for your answer.

The error rise on the server (4.16.4, large).
I use the node Python Script, not the legacy.
Therefore I haven’t find the Advanced section. Any other place to check?

My concern is that the error rise randomly : sometime, the workflow runs with no error.

Would you have any other advises ?

Best
Jerome

4

Hi @MichaelRespondek ,

Any update on that issue for me ?

Thanks
Best
Jerome

5

Hi Jerome,

Apologies for the delayed reply. This sounds like an issue caused by a latency. Could it be related to the mount/drive where the Python installation is stored?

Best,
Michael

6

Hi Michael,

Oh that is a good point. This might be the issue.
Is there a way to change the connection timeout ?

Best
Jerome

7

Hi @MichaelRespondek ,

Any feedback on that point to change the connection timeout threshold ?

Thanks
Best
Jerome

8

Hi @trj,

this sounds more like a Linux and network mount question; I’m not sure the KNIME experts can supply an answer to your question here. You could try to employ workarounds on a Python level with waits or filesystem checks, but I assume the cleaner solution would be to make sure the drive is properly mounted and quickly accessible on OS level.

Kind regards
Marvin

9

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.