[
https://issues.apache.org/jira/browse/GERONIMO-4497?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12703193#action_12703193
]
Mike commented on GERONIMO-4497:
--------------------------------
I think this problem may still exist. In the same environment (Geronimo 2.1.4
running an app that loads the BIRT report engine) we are seeing a report that
should finish in 5 to 10 minutes take several hours. While profiling the
Geronimo process we have found that the majority of the time is spent in
MultiParentClassLoader.loadClass(). Taking Geronimo out of the picture and
running this report from the BIRT development environment (Eclipse w/ BIRT
plugin) against the same database the report finishes in under 10 minutes.
Was this fixed confirmed?
> Searching for resources in MultiParentClassLoader is not fully optimized
> ------------------------------------------------------------------------
>
> Key: GERONIMO-4497
> URL: https://issues.apache.org/jira/browse/GERONIMO-4497
> Project: Geronimo
> Issue Type: Bug
> Security Level: public(Regular issues)
> Components: kernel
> Affects Versions: 2.1.4, 2.2
> Reporter: Tim McConnell
> Assignee: Tim McConnell
> Fix For: 2.1.4, 2.2
>
>
> In one failing scenario, the BIRT reporting engine is invoked with an XML
> data file to generate a PDF report file. MultiParentClassLoader.getResource()
> is invoked many times to search for various XML resources (e.g.,
> DocumentBuilderFactory). MultiParentClassLoader searches through the current
> classloader and all the parent classloaders, which it should, but this can
> take a considerable amount of time if the resource is never found. It seems
> that once a resource has been searched for and not found, there is no need to
> search for it again in the same classloader hierarchy.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
