[
https://issues.apache.org/jira/browse/GERONIMO-4906?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12770416#action_12770416
]
Jean-Jacques Parent commented on GERONIMO-4906:
-----------------------------------------------
I am still investigating the origin of my problem, may be some additional
informations:
I used a special oracle driver with trace functionalities. It was not possible
to set a high level of logs because of bad performances but we have the same
information: connection interrupted (stopped).
I succeeded to reproduce the problem in my development environment by
unplugging the network cable of my server. Indeed, the following actions in the
application after re-plugging the cable are giving: connection interrupted.
Some questions according to this kind of situation:
1 is this means that after a problem, maybe a network one, all the connection
pool managed by Geronimo falls into an invalidate state and this pool will
always give a connection even if this one is broken. (Unless this connection is
removed by the idle timeout or by restarting the server)
2 Why the Geronimo pool does not killed those corrupted connexions?
3 is it a good idea to set the connection idle timeout to 0, in order to force
the pool to kill the maximum of connections especially when those become unsafe?
We are still investigating the reasons why the connection pool become unsafe,
but I don't understand why broken connections are still available in the pool...
> DB connection problem
> ---------------------
>
> Key: GERONIMO-4906
> URL: https://issues.apache.org/jira/browse/GERONIMO-4906
> Project: Geronimo
> Issue Type: Bug
> Security Level: public(Regular issues)
> Components: databases
> Affects Versions: 2.0.1
> Environment: Geronimo running on Windows server 2003 (Virtual Machine)
> Oracle 10g
> Driver jdbc 10.2.0.1
> Spring-Version: 2.0.2
> Hibernate-Version: 3.2.0.cr4
> Reporter: Jean-Jacques Parent
> Attachments: config.xml, Errors_samples.txt
>
>
> Your expertise on my problem will be helpful.
> Once a week, I get a DB connectivity problem with my application.
> No new DB transaction can be open. After a few minutes, the situation can
> come back as normal, if not the Geronimo server has to be restarted.
> I ask our technicians about connectivity and performance, but they found
> nothing wrong.
> I give you the stack trace in attachment with the different kind of error
> message.
> You will see that it looks like a network problem, but maybe this could also
> be due to a jdbc, timeout...?
> My pool configuration is as follow
> Pool Min Size:25
> The minimum number of connections in the pool. The default is 0.
> Pool Max Size:140
> The maximum number of connections in the pool. The default is 10.
> Blocking Timeout: 5000 (in milliseconds)
> The length of time a caller will wait for a connection. The default is
> 5000.
> Idle Timeout: 5 (in minutes)
> How long a connection can be idle before being closed. The default is
> 15.
> Any help will be appreciate.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
