tomcat-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Davoust, Pascal" <pascal.davo...@eptica.com>
Subject Re: [jdbc-pool] Transaction left open by the connection validation mechanism ?
Date Sat, 05 Jan 2013 00:26:54 GMT
Sorry, I probably did not make myself clear.

I certainly do not suggest to issue the rollback on each connection
checkout, but only as part of the validation process - because this is
where the problem lies and only there - which I would imagine is triggered
only occasionally (when connection is idle for example): I doubt someone
would actually validate connections each time it's checked or out in. But I
might be wrong.

In any case, I suppose that the pool must not hand over a connection at
check-out time with an open transaction triggered by the internal
validation mechanism of the pool - irrespective of what you do with the
connection.

Am I wrong?

Regards
Le 5 janv. 2013 00:20, "Christopher Schultz" <chris@christopherschultz.net>
a écrit :

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA256
>
> Pascal,
>
> On 1/4/13 5:56 PM, Davoust, Pascal wrote:
> > Indeed I was referring to the tomcat jdbc-pool code, not mine :-).
> >
> > Now, I would suggest to consider that this problem is more common
> > than you could imagine at first sight. First because this is
> > randomly occuring, but frequently enough so that it shows up in
> > quite a few areas. A few examples where this problem occurred and
> > was filed (but not analyzed down to the root cause):
> >
> > Spring DataSource Txn Mgr :
> >
> https://jira.springsource.org/browse/SPR-8870?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel
> >
> >  Hyperic : https://jira.hyperic.com/browse/HHQ-5054
> >
> > If I'm correct, then I suggest that the behavior is corrected by
> > default, not requiring additional configuration.
>
> I disagree: issuing a ROLLBACK on every connection-checkout seems
> heavy-handed if it's not necessary. You are talking about a fairly
> rare set of circumstances (e.g. changing tx isolation level for only
> some uses of the connection-pool) and punishing everyone (performance)
> for that circumstance is IMHO unnecessary.
>
> - -chris
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG/MacGPG2 v2.0.17 (Darwin)
> Comment: GPGTools - http://gpgtools.org
> Comment: Using GnuPG with undefined - http://www.enigmail.net/
>
> iEYEAREIAAYFAlDnY5oACgkQ9CaO5/Lv0PARVgCgvPLCsydboLrgcLiQUg9wD/zI
> g/MAn0ZakKw6QgXd7n47UZ3CRtaYrRex
> =j8ND
> -----END PGP SIGNATURE-----
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
> For additional commands, e-mail: users-help@tomcat.apache.org
>
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message