[frontend] Failed to display a pipeline execution graph: "No graph to show"
Environment
-
How did you deploy Kubeflow Pipelines (KFP)? Manifests files from
-
KFP version: 2.0.0-alpha.1 (worked well in 1.7.0, first observed in 1.8.1).
Steps to reproduce
- Pipeline was uploaded to Kubeflow using Kubeflow Python SDK (version 1.8.1)
- Run a pipeline which is displaying as 2168 lines yaml file in Kubeflow UI.
- Once completed, Kubeflow UI shows a message: "No graph to show" instead of pipeline execution graph. Please see a screenshot attached.
Expected result
Kubeflow UI shows a pipeline execution graph.
Materials and Reference
There was a similar error described in https://github.com/kubeflow/pipelines/issues/4289.
Impacted by this bug? Give it a 👍. We prioritise the issues with the most 👍.
Hello @pahask8 , would you mind providing the example you used for this run, so that we can reproduce the result for this issue. In addition, we would like to ask if you were using v1 mode or v2 mode for compiling the pipeline definition.
Hello @jlyaoyuli, thank you for quick update. I used v1 mode for compiling the pipeline. I can try to provide an example of this , can you please suggest, will the complete yaml work for you? We use this pipeline in production and have some private data there. Thanks
Hello @pahask8 , yes the complete yaml works. Can provide a simplified version of pipeline (yaml file)? Thank you :) Noted: We think that the potential reason for this error is because we upgrade node from 12 to 14, which doesn't allow using Buffer API on the Browser side. https://github.com/kubeflow/pipelines/blob/a8a3b93aded13c9c8f5c9207dd1c233488aaf29f/frontend/src/lib/Utils.tsx#L391
Hello @jlyaoyuli, please find the attached pipeline file. Please note I masked all sensitive data there. Hope that helps. Thank you! pipeline.yaml.zip
It's definitely something to do with compressedNodes. The problem starts happening as soon as the pipeline gets big enough to trigger Argo to start using the compressedNodes method of storing the run state.
Any update on when this will be addressed?
Also upping this as something that is impacting our team. We're downgrading KFP because this issue makes the latest versions of KFP unusable for us.