[
https://issues.apache.org/jira/browse/CALCITE-1121?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15252511#comment-15252511
]
Julian Hyde commented on CALCITE-1121:
--------------------------------------
I think you are running into a different issue. Since there is no {{GROUP BY}}
we would run a "select"-type Druid query, and we with it with {{threshold:
MAX_INTEGER}}, which basically means bring all rows back in one batch. This
would blow memory of the processing node, and the node would die, so no
subsequent queries would work. As of
https://github.com/julianhyde/calcite/commit/fc438538b610553515fe2c85c4ed4313c3b89695,
we now fetch 16K rows at a time.
When we implement {{SELECT .. [ ORDER BY .. ] LIMIT}} we should push the limit
down.
> Druid adapter
> -------------
>
> Key: CALCITE-1121
> URL: https://issues.apache.org/jira/browse/CALCITE-1121
> Project: Calcite
> Issue Type: Bug
> Reporter: Julian Hyde
> Assignee: Julian Hyde
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
