While you could already analyze your failed events in BigQuery using SQL, it was not possible to look at some of the most useful fields in that data. Like
error… which contains the error message! (That’s because sometimes
error is a string and sometimes it’s an object with more details. We couldn’t find a fitting BigQuery column type at the time.)
Well, now you can look at all fields! And you don’t need to change anything in your pipeline, just use the updated table definitions and follow the updated instructions in the docs. We load the offending fields (such as
error ) as JSON — a new data type BigQuery added in 2022. We’ve expanded the docs with some examples on how to construct the queries and how to deal with JSON values.
Happy New Year and happy querying!