You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ignite.apache.org by "Taras Ledkov (JIRA)" <ji...@apache.org> on 2019/03/18 14:00:00 UTC
[jira] [Comment Edited] (IGNITE-11524) Memory leak caused by
executing a jdbc prepared statement
[ https://issues.apache.org/jira/browse/IGNITE-11524?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16795035#comment-16795035 ]
Taras Ledkov edited comment on IGNITE-11524 at 3/18/19 1:59 PM:
----------------------------------------------------------------
[~vozerov]
1. Fixed,
2. {{JdbcThinConnection#stmts}} is cleared on close. I guess the statements shouldn't close explicitly on connection close:
- The object on server side are cleared by {{onDisconnect}}.
- On client side {{JdbcThinStatement#isClosed}} check connection state.
- It takes additional time on close to send not useful close commands.
I guess clear {{JdbcThinConnection#stmts}} is enough for decrease memory pressure when {{JdbcThinConnection}}s leak at user code.
We have to nullify the content of the {{JdbcThinStatement}} to reduce memory pressure when {{JdbcThinStatement}}s leak at user code.
was (Author: tledkov-gridgain):
[~vozerov]
1. Fixed,
2. {{JdbcThinConnection#stmts}} is cleared on close. I guess the statements shouldn't close explicitly on connection close:
- The object on server side are cleared by {{onDisconnect}}.
- On client side {{JdbcThinStatement#isClosed}} check connection state.
I guess clear {{JdbcThinConnection#stmts}} is enough for decrease memory pressure when {{JdbcThinConnection}}s leak at user code.
We have to nullify the content of the {{JdbcThinStatement}} to reduce memory pressure when {{JdbcThinStatement}}s leak at user code.
> Memory leak caused by executing a jdbc prepared statement
> ---------------------------------------------------------
>
> Key: IGNITE-11524
> URL: https://issues.apache.org/jira/browse/IGNITE-11524
> Project: Ignite
> Issue Type: Bug
> Components: jdbc, sql
> Reporter: Pavel Vinokurov
> Assignee: Taras Ledkov
> Priority: Blocker
> Fix For: 2.8
>
> Attachments: PreparedStatementOOMReproducer.java
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> Executing a prepared statement multiple times lead to OOM.
> VisualVM indicates that heap contains a lot of JdbcThinPreparedStatament objects.
> The reproducer is attached.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)