Error when publishing to SAS Micro Analytic Service (MAS) destination

[alisio_meneses]

Hi there!

I am using SAS Viya 3.5 and when I try to publish a binary classification model created using SAS Model Studio (8.5) to a SAS Micro Analytic Service (MAS) destination I get the following error:

"An error occurred. Please refer to the log for details or contact your system administrator. The log can be found at:"

https://mydomain/modelPublish/models/78efa1df-1c29-4e37-b199-e5f5927686e4/log

The given error link contains the following json:

{"creationTimeStamp":"2022-05-25T00:06:05.547Z","modifiedTimeStamp":"2022-05-25T00:06:10.659Z","createdBy":"userhere","modifiedBy":"userhere","links":[{"method":"GET","rel":"up","href":"/modelPublish/models/09ba42be-dafa-47ad-8a3f-ea2af3129bc6","uri":"/modelPublish/models/09ba42be-dafa-47ad-8a3f-ea2af3129bc6","type":"application/vnd.sas.models.publishing.publish"},{"method":"GET","rel":"self","href":"/modelPublish/models/09ba42be-dafa-47ad-8a3f-ea2af3129bc6/log","uri":"/modelPublish/models/09ba42be-dafa-47ad-8a3f-ea2af3129bc6/log","type":"application/json"}],"version":1,"id":"af9ddfa8-64c5-4853-abf6-10f63f58ba06","publishName":"Published model Gradient Boosting (Pipeline 1)","destinationName":"maslocal","modelId":"fa5f2a1d-59fe-4904-a24e-aba49aa75ec7","modelName":"Gradient_Boosting__Pipeline_1_","log":"ERROR==={\"errorCode\":0,\"message\":\"In declaration of method predictedscore: parameter \\\"EM_CLASSIFICATION\\\" is 'in_out'; therefore, the type size (32) will be ignored.\",\"links\":[],\"version\":2,\"httpStatusCode\":0}"}

I guess the most important part of the error message is this at the end:

In declaration of method predictedscore: parameter \\\"EM_CLASSIFICATION\\\" is 'in_out'; therefore, the type size (32) will be ignore

Although I can read the problem description, I really don't know what to make of it. 

More details:

• It is possible publish the same model successfully to a SAS Cloud Analytic Services (CAS);
• The desired MAS destination has working previously published models;
• The model was created using the automated pipeline creation feature, but the same problem happened in a model from a different, manually created pipeline;
• Same thing happens, an error, when publishing the same model from SAS Model Manager.

Can anyone point me in the right direction on how to solve this problem?

Thank you!

[alexal · Accepted solution]

It seems like your pipeline contains an analytic store (astore) model. Users who need to work with analytic store models must have Read and Write permissions to both the /opt/sas/viya/config/data/modelsvr/astore and /models/astores/viya directories. By default, the sas group has full permission to these directories. Please see more details here: https://go.documentation.sas.com/doc/en/calcdc/3.5/calmodels/n10916nn7yro46n119nev9sb912c.htm#n0z2r4... 

[alisio_meneses]

Hi Alexal, thanks for your reply. I'll have to check filesystem permissions with the admins, in the mean time, how do I know if a model contains an analytic store?