Mailing-List: contact user-help@cayenne.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@cayenne.apache.org
Received-SPF: pass (herse.apache.org: local policy)
Mime-Version: 1.0 (Apple Message framework v752.2)
In-Reply-To: 
 <49D23F7404DC5B4086D7352312597E0FD269BC@FNHH-SVMEXDB002.Freenet-AG.de>
References: 
 <49D23F7404DC5B4086D7352312597E0FD269BC@FNHH-SVMEXDB002.Freenet-AG.de>
Content-Type: text/plain; charset=ISO-8859-1; delsp=yes; format=flowed
Message-Id: <A39D3684-6A8E-4CA0-898B-2857C37F7471@pvv.ntnu.no>
Content-Transfer-Encoding: quoted-printable
From: Tore Halset <halset@pvv.ntnu.no>
Subject: Re: AW: postgres, idle in transaction
Date: Wed, 4 Apr 2007 10:30:19 +0200
To: user@cayenne.apache.org

Hello.

I am using cayenne with PostgreSQL, but have not seen this problem. =20
Are you using a new PostgreSQL release and and a jdbc-driver matching =20=

that version? Search http://jdbc.postgresql.org/changes.html has one =20
bugfix related to this from back in 2004.

In a (old) non-cayenne environment I have had some problems with =20
connections not properly closed after use. I tracked down that =20
problem by saving the stacktrace (new Exception()) and a timestamp =20
every time a Connection is required. That way I had a register over =20
all Connections and was able to find the ones that was not closed =20
after use.

Could you try to create a small example that reproduce this problem?

  - Tore.

On Apr 4, 2007, at 8:31, Peter Schr=F6der wrote:

> thank you for your reply, but we are experiencing "idle in =20
> transaction" connections that are hung and have to be killed =20
> manually. there are a lot of other connections in state "idle" =20
> which we recognized as the connection-pool. so we are already =20
> worrying ;-).
> we have a lot of background/scheduled tasks in our app, so we dont =20
> really know where the problem occurs. currently we are searching =20
> our logs to find something usefull.