How to revert manually normalized discovery models?

[ClayB]

Hello,

We have a few discovery models that have been manually normalized and would like them reverted back so that they can be picked up by the daily normalization job. It's my understanding that you can't use the "Revert" button method on discovery models that were normalized by pattern rules or manually. I have seen mention of this being done with a script though. I'm looking for additional information on this method or an alternative method for getting these reverted. Thank you for any information you can provide.

[dreinhardt]

Hi @ClayB,

they reason why correct manually normalized models are never being normalized by the content service is the following, Support Case Answer which also includes the next steps to "revert" a discovery model to be picked up by the content service normalization. ServiceNow is aware of this limitation and will provide a solution (later ...)

[-] Normalization Suggestion are not generated if suggested values are same as normalized values. This is expected behavior.
[-] If you further want these "manually normalized" Discovery Models can be into the final state "Normalized", since suggestions are not generated if suggested values are same as normalized values, they you have to manually normalise these Discovery Models with right 'norm_publisher' field and rest of the fields (norm_product and norm_version) as EMPTY and wait for suggestions to be generated so that you can Normalize Discovery Models based on Suggestions from Content. This will generate normalization suggestion and later can be accepted.

Best, Dennis

[ClayB]

Hi Dennis,

Thanks as always for the helpful information. Just so I'm understanding, the discovery model fields for Publisher, Product, and Version should be blank (see picture) in order for the normalization suggestions to generate?

Kind Regards,

Clay B

[dreinhardt]

Hi @ClayB,

they you have to manually normalize these Discovery Models with right 'norm_publisher' field and rest of the fields (norm_product and norm_version) as EMPTY and wait for suggestions to be generated.

In my opinion, however, it is a matter of luck whether your discovery model can be normalized exactly. You should be aware that with the settings mentioned, you will have to accept a greater loss of data quality and that installations cannot be assigned correctly until normalization is completed.

Optionally, you can also report all non-normalized discovery models from a manufacturer/product as XML to the content team and they will process the data step by step – we like to use this method for publisher onboarding that will take place in the next 3-6 months so that we can start with a good baseline of models.

I hope that ServiceNow will have a solution for the above-mentioned restriction in the next 1-2 releases, and in the meantime we are working with validated models that are manually normalized. In the end, it's only cosmetic or KPI.

Best, Dennis