This happens when I try to select from a Snowflake table and aggregate using sum on a numeric (type/Float) field. Fields that have a type of type/Number in the underlying Snowflake table do not have this issue.
By query definition do you mean the raw SQL that is generated via the visual editor? If so, here it is:
SELECT
DATE_TRUNC(
'week',
"MY_NAMESPACE".""MY_TABLE""."DATE"
) AS "DATE",
SUM(
"MY_NAMESPACE".""MY_TABLE""."FLOAT_COLUMN"
) AS "sum"
FROM
"ANALYTICS"."MY_NAMESPACE".""MY_TABLE""
GROUP BY
DATE_TRUNC(
'week',
"MY_NAMESPACE".""MY_TABLE""."DATE"
)
ORDER BY
DATE_TRUNC(
'week',
"MY_NAMESPACE".""MY_TABLE""."DATE"
) ASC
the FLOAT_COLUMN was the issue that was failing after upgrading. Updating that column type on the Snowflake table to type/Number made it work again.
Yes this happened across the existing question on 49 and new question in 50. We did change from float to number, did a schema sync and that resolved the issue.
Interestingly enough the issue was not across all float columns, just one float column on one specific table.
If you have access to the Metabase application database, could you run a query on the metabase_field table to see if there are any differences between this one float and the others?
SELECT * FROM metabase_field
or to narrow it down to just the data type classifications