[
https://issues.apache.org/jira/browse/IMPALA-8335?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16801143#comment-16801143
]
Herman Chen commented on IMPALA-8335:
-------------------------------------
Actually I discovered that the avro schema must be specified as a part of the
CREATE TABLE statement, then Impala is able to respect column names containing
uppercase. The challenge there is that specifying the schema inlined, i.e.
TBLPROPERTIES "avro.schema.literal", very easily hits the limit. So it most
likely requires using "avro.schema.url" to refer to the schema file.
> Avro support requires field names to be all lowercase
> -----------------------------------------------------
>
> Key: IMPALA-8335
> URL: https://issues.apache.org/jira/browse/IMPALA-8335
> Project: IMPALA
> Issue Type: Question
> Components: Catalog, Docs
> Reporter: Herman Chen
> Priority: Major
>
> I am using external tables where the data resides in S3 in avro format. My
> avdl contains fields such as:
> string id;
> string sql;
> string templateId;
> union \{ null, string } pool;
> union \{ null, string } user;
> string startTime;
> string endTime;
> However, Impala is able to read the all-lowerfields like "id" and "sql", but
> for any field containing uppercase like "templateId" or "startTime", Impala
> shows them all as NULL. I believe this is due to how Impala treats all
> identifier as lowercase, including column name. Is it so? Is there any way to
> work around it? If not it poses a major roadblock for working with
> pre-existing avro data.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
