[ 
https://issues.apache.org/jira/browse/ASTERIXDB-1056?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15265950#comment-15265950
 ] 

Abdullah Alamoudi commented on ASTERIXDB-1056:
----------------------------------------------

We actually made it parallel then we had to revert to making it a single thread 
in order for record identification to work.
Record Identification assumes the same records will be produced from the same 
partition in the exact order.

There is a bug in HDFS but no test case yet to uncover it. I have discussed 
with Till and Yingyi a way to handle this by producing RIDs which will enable 
us to make this parallel again.

> Loading (parsing) ADMs from LocalFS is single-threaded per NC
> -------------------------------------------------------------
>
>                 Key: ASTERIXDB-1056
>                 URL: https://issues.apache.org/jira/browse/ASTERIXDB-1056
>             Project: Apache AsterixDB
>          Issue Type: Improvement
>          Components: Adapters, AsterixDB, Operators
>            Reporter: asterixdb-importer
>            Assignee: Yingyi Bu
>            Priority: Minor
>
> Loading (parsing) ADMs from LocalFS is single-threaded per NC



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

Reply via email to