Issue is resolved. Was not casting to string when fetching from Map.
On Saturday, February 9, 2019, 3:09:48 PM PST, Krishnanand Khambadkone <kkhambadkone@yahoo.com.INVALID>
wrote:
Hi Kunal, I am trying to query spark application event json data. Here is my query.
I am using drill version 1.15.0. I am using a custom UDF in the query.
select s.application_id, get_spark_attrs(s.spark_event,'spark.executor.memory') as spark_attributes
from dfs.`sparkapplicationevent.json` s where (REGEXP_REPLACE(REGEXP_REPLACE(substr(s.spark_event,11),
'[^0-9A-Za-z]"', ''),'(".*)','') = 'SparkListenerEnvironmentUpdate' or REGEXP_REPLACE(REGEXP_REPLACE(substr(s.spark_event,11),
'[^0-9A-Za-z]"', ''),'(".*)','') = 'SparkListenerApplicationStart' or REGEXP_REPLACE(REGEXP_REPLACE(substr(s.spark_event,11),
'[^0-9A-Za-z]"', ''),'(".*)','') = 'SparkListenerApplicationEnd') group by application_id,
spark_attributes order by application_id;
On Saturday, February 9, 2019, 1:42:57 PM PST, Kunal Khatua <kunal@apache.org>
wrote:
This message is eligible for Automatic Cleanup! (kunal@apache.org) Add cleanup rule | More
info
Hi Krishnanand
The mailing lists don't allow attachments. Please provide the Drill version, query, data source
and a link to where the relevant part of the log can be downloaded.
The message by itself is not useful, especially without a stacktrace.
On 2/9/2019 11:39:49 AM, Krishnanand Khambadkone <kkhambadkone@yahoo.com.invalid> wrote:
Any idea why such an error would be thrown from custom UDF. I am using jackson json parser
in this UDF.
Error: SYSTEM ERROR: CompileException: Line 80, Column 24: Assignment conversion not possible
from type "java.lang.Object" to type "java.lang.String"
Fragment 0:0
Please, refer to logs for more information.
|