function readOnly(count){ }
Starting November 20, the site will be set to read-only. On December 4, 2023,
forum discussions will move to the Trailblazer Community.
+ Start a Discussion
Bob BI 2Bob BI 2 

Einstein - snowflake external dataset

Hi everyone,

I want to bring a dataset from Snowflake to Einstein.
I am admin in both systems.
Permissions seem valid on both ends, and the connection is successful.

However, when I try to bring tables from snowflake, I don't get the expected results.

Even more, I get different responses for trying to bring two different tables from the same db/schema in Snowflake.

I have two cases:


1. Case 1 -

I have a table in Snowflake with the following data types (number, varchar, timestamp).


When I want to create the dataset, the schema is visible, but only specific types of fields - Number(38,0) and timestamp_NTZ(9). Einstein doesn't pull the varchar type into the schema, even if I short it to 255 chars.

In this case, only a partial view of the table schema is visible. However, the data can be viewed in preview, and Einstein manages to pull the table (though partial).

2. Case 2

I try another table in the same schema and with identical permissions set.
I can see the entire Snowflake schema (including varchar) , but when I try to view the data in preview I get an error:

 

"
Application error

Can’t retrieve field values, but you can still edit the object settings.: Unable to get result data: Adapter read method has returned failure.The agent failed to read data from the source because of the following error: [SQL compilation error:

Object 'TEST__' does not exist or not authorized.]

Error in Job ID connector.4Vw_OF1WQqTP7ykCag6X_-:
"

The case 2 error suggests permission issues, but in case 1 I manage to pull the data without a permission error message.

 

I'm not even sure the two issues are connected.

Can someone help / Have ideas? 

Thanks

Best Answer chosen by Bob BI 2
SwethaSwetha (Salesforce Developers) 
HI Bob,
I see that you have logged a case with Salesforce support for this issue and the team is actively working on the fix. I will update this thread with the fix provided by the team.

Can you please mark this answer as best to close the thread. Thank you

All Answers

SwethaSwetha (Salesforce Developers) 
HI Bob,
I see that you have logged a case with Salesforce support for this issue and the team is actively working on the fix. I will update this thread with the fix provided by the team.

Can you please mark this answer as best to close the thread. Thank you
This was selected as the best answer
SwethaSwetha (Salesforce Developers) 
Update on root cause:

Case 1:
That is expected behavior and the limit we support with the Column size is 32k. 

https://support.snowflake.net/s/question/0D50Z00009SDHEoSAP/is-there-case-insensitivity-for-table-name-or-column-names

Snowflake stores tables and columns as UPPERCASE, which is also how they are displayed. However, you can reference them both in any case. Values inside of columns, however, are case-sensitive.

We also observed that when you create Table from UI on snowflake, the table name is stored as in Upper case completely. 

The solution here is to use the Uppercase names for the Tables. 

Case 2:
When using the snowflake connector with Einstein Analytics TableName not found error was occurring.

Exact error: "Application error
Can’t retrieve field values, but you can still edit the object settings.: Unable to get result data: Adapter read method has returned failure.The agent failed to read data from the source because of the following error: [SQL compilation error: Object 'TEST__' does not exist or not authorized.] Error in Job ID connector.4Vw_OF1WQqTP7ykCag6X_-: "

This was due to a bug in the Snowflake connector. The Salesforce team has logged a bug with the connector teams and following up with the concerned.

Salesforce case reference: #26470873