[jira] [Comment Edited] (DRILL-2322) CSV record reader should log which file and which record caused an error in the reader

["Zelaine Fong (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/DRILL-2322?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14965621#comment-14965621 ] 

Zelaine Fong edited comment on DRILL-2322 at 10/20/15 7:55 PM:
---------------------------------------------------------------

My understanding is that the original fix that this Jira was intended to address was the case when there's a non-printable character in the CSV file, and when the query errors out, then the fix will return the line number of the bad record in the CSV file.

Khurram's query includes a cast that results in a conversion error.  That is not addressed by this fix.  The fix for conversion type errors is to skip these records.  That's the part that Sean is working on in DRILL-3764.


was (Author: zfong):
My understanding is that the original fix that this Jira was intended to address was the case when there's a non-printable character in the CSV file, and when the query errors out, then the fix will return the line number of the bad record in the CSV file.

Khurram's query includes a cast that results in a conversion error.  That is not addressed by this fix.  The fix for conversion type errors is to skip these records.  That's the part that Sean is working on in MD-400.

> CSV record reader should log which file and which record caused an error in the reader
> --------------------------------------------------------------------------------------
>
>                 Key: DRILL-2322
>                 URL: https://issues.apache.org/jira/browse/DRILL-2322
>             Project: Apache Drill
>          Issue Type: Improvement
>          Components: Storage - Text & CSV
>    Affects Versions: 0.8.0
>            Reporter: Ramana Inukonda Nagaraj
>            Assignee: Sudheesh Katkam
>             Fix For: 0.9.0
>
>         Attachments: DRILL-2322.1.patch.txt, DRILL-2322.2.patch.txt, DRILL-2322.3.patch.txt
>
>
> I believe the title is self exploratory.
> If the text reader fails for any reason due to an offending record drill should log which file (if there are multiple files) and which line/record the error occurs at. This will improve debugging when dealing with large files/ large number of files.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)