I’m working on deploying a data-app to the business hub, but struggle with executing the app.
It seems that there is an issue with installing the python environment via conda as the error message reads
ERROR Conda Environment Propagation 3:4164:0:3343: Execute failed: Conda process terminated with error code 137.
ERROR Component Output 3:4164:0:2: Errors in workflow - Contains one node with execution failure (Conda Environment Propagation #3343)
Searching for that error code leads to an oom error. However, the issue seems to appear indifferently of the executor ram (I tried 8-32GB using the pre-built full container).
I’ve generated a support bundle, but there are no explicit errors. Any tips where to dig deeper?
Thank you for reporting this incident. The error code 137 indeed suggests that the executor running the workflow was terminated due to exceeding the allowed memory limit.
For now, the way around this issue is likely to further increase the executor’s memory (if possible).
Alternatively, you could build a custom executor image that already has the desired conda environment installed into the image. (Installing the environment into the image also has the slight advantage that after executor restarts, the environment doesn’t need to be rebuilt.)
The underlying issue is a memory inefficiency in the conda package resolution. We are upgrading the package resolution software (planned for release with KNIME 5.4 still this year), to prevent such incidents from occurring at all.