[
https://issues.apache.org/jira/browse/DRILL-7192?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16827143#comment-16827143
]
Kunal Khatua commented on DRILL-7192:
-------------------------------------
[~vvysotskyi]
The source of this issue is that we areĀ setting the parameter at a {{SESSION}}
level, although the feature works at a {{QUERY}} level (i.e. on the
{{Statement}} object).
Calling {{!set rowlimit 10}} will execute the `Statement.setMaxRows()`
automatically for each new Statement.
However, {{!set rowlimit 0}} will *not* execute the `Statement.setMaxRows()`
automatically for each new Statement. My guess is that since each query's
Statement is a new object with a presumed default 0 (on the client side), it
does not again `ALTER SESSION` behind the scene. I'll verify this with a custom
code, but if that is the case... fixing SQLLine is not the solution. The only
workaround would be to not rely on the {{SESSION}}-level value, but on the
RunQuery.getAutolimitRowcount() by ensuring the value is set in the
{{DrillClient}}
> Drill limits rows when autoLimit is disabled
> --------------------------------------------
>
> Key: DRILL-7192
> URL: https://issues.apache.org/jira/browse/DRILL-7192
> Project: Apache Drill
> Issue Type: Bug
> Affects Versions: 1.16.0
> Reporter: Volodymyr Vysotskyi
> Assignee: Kunal Khatua
> Priority: Major
> Fix For: 1.17.0
>
>
> InĀ DRILL-7048 was implemented autoLimit for JDBC and rest clients.
> *Steps to reproduce the issue:*
> 1. Check that autoLimit was disabled, if not, disable it and restart Drill.
> 2. Submit any query, and verify that rows count is correct, for example,
> {code:sql}
> SELECT * FROM cp.`employee.json`;
> {code}
> returns 1,155 rows
> 3. Enable autoLimit for sqlLine sqlLine client:
> {code:sql}
> !set rowLimit 10
> {code}
> 4. Submit the same query and verify that the result has 10 rows.
> 5. Disable autoLimit:
> {code:sql}
> !set rowLimit 0
> {code}
> 6. Submit the same query, but for this time, *it returns 10 rows instead of
> 1,155*.
> Correct rows count is returned only after creating a new connection.
> The same issue is also observed for SQuirreL SQL client, but for example, for
> Postgres, it works correctly.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)