You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@ignite.apache.org by "Vladimir Ozerov (JIRA)" <ji...@apache.org> on 2015/11/26 15:06:11 UTC
[jira] [Created] (IGNITE-2009) Cached prepared SQL statement causes
IgniteKernal leak.
Vladimir Ozerov created IGNITE-2009:
---------------------------------------
Summary: Cached prepared SQL statement causes IgniteKernal leak.
Key: IGNITE-2009
URL: https://issues.apache.org/jira/browse/IGNITE-2009
Project: Ignite
Issue Type: Task
Components: cache, interop
Affects Versions: ignite-1.4
Reporter: Vladimir Ozerov
Priority: Blocker
Fix For: 1.5
Problem: when query is executed, it is cache in thread-local in parsed/prepared form. It has reacheable reference to IgniteKernal. As a result IgniteKernal cannot be collected until the thread dies. If this thread is from a pool, it could be a real leak.
Can we employ WeakReferences here?
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
Re: [jira] [Created] (IGNITE-2009) Cached prepared SQL statement
causes IgniteKernal leak.
Posted by Vladimir Ozerov <vo...@gridgain.com>.
As I understand, the main goal of thes thread-local - to allow re-use
query *between
*task execution.
On Thu, Nov 26, 2015 at 5:27 PM, Vladimir Ershov <ve...@gridgain.com>
wrote:
> Hi,
> Why then thread couldn't clear this threadLocal query at the end of the
> task execution?
>
> On Thu, Nov 26, 2015 at 5:06 PM, Vladimir Ozerov (JIRA) <ji...@apache.org>
> wrote:
>
> > Vladimir Ozerov created IGNITE-2009:
> > ---------------------------------------
> >
> > Summary: Cached prepared SQL statement causes IgniteKernal
> > leak.
> > Key: IGNITE-2009
> > URL: https://issues.apache.org/jira/browse/IGNITE-2009
> > Project: Ignite
> > Issue Type: Task
> > Components: cache, interop
> > Affects Versions: ignite-1.4
> > Reporter: Vladimir Ozerov
> > Priority: Blocker
> > Fix For: 1.5
> >
> >
> > Problem: when query is executed, it is cache in thread-local in
> > parsed/prepared form. It has reacheable reference to IgniteKernal. As a
> > result IgniteKernal cannot be collected until the thread dies. If this
> > thread is from a pool, it could be a real leak.
> >
> > Can we employ WeakReferences here?
> >
> >
> >
> > --
> > This message was sent by Atlassian JIRA
> > (v6.3.4#6332)
> >
>
Re: [jira] [Created] (IGNITE-2009) Cached prepared SQL statement
causes IgniteKernal leak.
Posted by Vladimir Ershov <ve...@gridgain.com>.
Hi,
Why then thread couldn't clear this threadLocal query at the end of the
task execution?
On Thu, Nov 26, 2015 at 5:06 PM, Vladimir Ozerov (JIRA) <ji...@apache.org>
wrote:
> Vladimir Ozerov created IGNITE-2009:
> ---------------------------------------
>
> Summary: Cached prepared SQL statement causes IgniteKernal
> leak.
> Key: IGNITE-2009
> URL: https://issues.apache.org/jira/browse/IGNITE-2009
> Project: Ignite
> Issue Type: Task
> Components: cache, interop
> Affects Versions: ignite-1.4
> Reporter: Vladimir Ozerov
> Priority: Blocker
> Fix For: 1.5
>
>
> Problem: when query is executed, it is cache in thread-local in
> parsed/prepared form. It has reacheable reference to IgniteKernal. As a
> result IgniteKernal cannot be collected until the thread dies. If this
> thread is from a pool, it could be a real leak.
>
> Can we employ WeakReferences here?
>
>
>
> --
> This message was sent by Atlassian JIRA
> (v6.3.4#6332)
>