I have a specific use case where I call another notebook using the
dbutils.notebook.run() function. The other notebook do some processing and return a string in the dbutils.notebook.exit() function to the caller notebook. The returned string has some french special characters in it like à, é, è and because of that, the calling notebook executes for about 5 minutes longer than the called notebook and eventually ends with an exception: com.databricks.WorkflowException: java.nio.charset.MalformedInputException: Input length = 1
If I remove the special characters, everything works fine. Same thing If I use the .encode('ascii', 'remove') function on my string, but I need to have the the correct string returned with my accents. Is there a way to preserve my string intact? I'm on the Databricks Runtime Version 13.3 LTS.
You can reproduce the behavior using these two simple notebooks.
Caller Notebook:
output = dbutils.notebook.run('called_notebook', 600)
Called Notebook:
dbutils.notebook.exit("La mise à jour des tables des données raffinées est terminée")