[ https://issues.apache.org/jira/browse/CARBONDATA-1168?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16053380#comment-16053380 ]
Kingsley Chen commented on CARBONDATA-1168: ------------------------------------------- I tested it in the version of spark1.6+CarbonData-1.0.0-incubating, it can process 100,000+ in one batch. So is that a bug of version 1.1.0?? Is there anyone can help me??? > Driver Delete data operation is failed due to failure in creating delete > delta file for segment > ----------------------------------------------------------------------------------------------- > > Key: CARBONDATA-1168 > URL: https://issues.apache.org/jira/browse/CARBONDATA-1168 > Project: CarbonData > Issue Type: Bug > Components: sql > Affects Versions: 1.1.0 > Environment: spark1.6+carbon1.1.0 > we have a 20 nodes cluster and 32GB ram each > Reporter: Kingsley Chen > Labels: features > Fix For: NONE > > > We use spark code to delete data from table as below > ------------------spark code---------------------- > val deleteSql = s"DELETE FROM $tableName WHERE $rowkeyName IN > (${rowKeyVals.mkString(",")})" > cc.sql(deleteSql).show() > ------------------spark code---------------------- > when the array size of rowKeyVals is larger than 200, the delete operation > will be failed, and print the log: > Delete data request has been received for default.item > Delete data operation is failed for default.item > Driver Delete data operation is failed due to failure in creating delete > delta file for segment : null block : null > ++ > || > ++ > ++ > That is to say, it only delete success maximun at 200 a batch,and took about > 1min which is > too slow. So my question is how to tuning the performance to make the batch > larger > and delete faster -- This message was sent by Atlassian JIRA (v6.4.14#64029)