Closed james-tn closed 1 year ago
Is the following error caused by the cli version issue? The "Run Azure ML Pipeline" step was never completed successfully.
2022-05-16T18:54:53.5569884Z ERROR: Met error <class 'marshmallow.exceptions.ValidationError'>:Validation for PipelineJobSchema failed: 2022-05-16T18:54:53.5571134Z 2022-05-16T18:54:53.5571645Z { 2022-05-16T18:54:53.5572090Z "jobs": { 2022-05-16T18:54:53.5572782Z "prep-job": { 2022-05-16T18:54:53.5573302Z "key": [ 2022-05-16T18:54:53.5574419Z "Pipeline node name should be a valid python identifier(upper or lower letter, number, underscore and start with a letter or underscore). Currently got prep-job." 2022-05-16T18:54:53.5575270Z ] 2022-05-16T18:54:53.5575711Z }, 2022-05-16T18:54:53.5576318Z "train-job": { 2022-05-16T18:54:53.5576837Z "key": [ 2022-05-16T18:54:53.5577924Z "Pipeline node name should be a valid python identifier(upper or lower letter, number, underscore and start with a letter or underscore). Currently got train-job." 2022-05-16T18:54:53.5578763Z ] 2022-05-16T18:54:53.5579186Z }, 2022-05-16T18:54:53.5580129Z "evaluate-job": { 2022-05-16T18:54:53.5580713Z "key": [ 2022-05-16T18:54:53.5581832Z "Pipeline node name should be a valid python identifier(upper or lower letter, number, underscore and start with a letter or underscore). Currently got evaluate-job." 2022-05-16T18:54:53.5582670Z ] 2022-05-16T18:54:53.5583100Z }, 2022-05-16T18:54:53.5583731Z "register-job": { 2022-05-16T18:54:53.5584238Z "key": [ 2022-05-16T18:54:53.5585335Z "Pipeline node name should be a valid python identifier(upper or lower letter, number, underscore and start with a letter or underscore). Currently got register-job." 2022-05-16T18:54:53.5586160Z ] 2022-05-16T18:54:53.5586597Z } 2022-05-16T18:54:53.5587015Z } 2022-05-16T18:54:53.5587430Z }
That's exactly it. The dev team needs to follow this basic practice. If you downgrade to cli version 2.2.1 this will also fix
@cynthiajiangatl It's a simple fix, btw -- just edit I believe pipeline.yml and replace the "-" in the job steps names with "_" (remember also the references to steps), and it'll work. I also had this problem last week.
This has been fixed in the current version. Closing the issue for now.
Why?
If you don't fix the version, if there's an upgrade from the library the pipeline may break. Currently the install CLI action does not fix the version which means that Devops Agent will pick up the latest Azure ML CLI version while the job/pipeline definition still follow old version. This may break the pipeline.
How?
Anything else?