[jboss-jira] [JBoss JIRA] Commented: (JBRULES-2477) Time spent in org.mvel2.integration.impl.ClassImportResolverFactory constructor causing transaction timeouts

Edson Tirelli (JIRA) jira-events at lists.jboss.org
Tue Jun 15 21:09:46 EDT 2010


    [ https://jira.jboss.org/browse/JBRULES-2477?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12535861#action_12535861 ] 

Edson Tirelli commented on JBRULES-2477:
----------------------------------------

Sam, can you test your use case using latest Drools [1] and let us know if you still see the bottleneck? Back in April I fixed JBRULES-2067 [2] by creating a class resolution cache to avoid searching the classpath over and over for all necessary classes. There is also a configuration to enable/disable this cache, as per comments in that ticket. 

I will close this issue for now, as I am assuming that will fix your problem. If it is not the case, please re-open it and let me know. In that case we will probably have to submit a ticket to MVEL to fix this in MVEL code, but I will follow up with Mike to get this released soon in MVEL.

[1] http://hudson.jboss.org/hudson/job/drools/lastSuccessfulBuild/artifact/trunk/target/
[2] https://jira.jboss.org/browse/JBRULES-2067

Thanks,
Edson

> Time spent in org.mvel2.integration.impl.ClassImportResolverFactory constructor causing transaction timeouts
> ------------------------------------------------------------------------------------------------------------
>
>                 Key: JBRULES-2477
>                 URL: https://jira.jboss.org/browse/JBRULES-2477
>             Project: Drools
>          Issue Type: Bug
>      Security Level: Public(Everyone can see) 
>          Components: drools-core
>    Affects Versions: 5.0.0.FINAL
>            Reporter: Sam Doyle
>            Assignee: Edson Tirelli
>            Priority: Critical
>
> We have hit a critical bottleneck in our application as a result of attempting to execute rules on more facts then we have in the past.
> I didn't think it was Drools at first but after profiling it is pretty evident that it is.
> The callgraph is showing 560s in org.mvel2.integration.impl.ClassImportResolverFactory constructor and 152530 calls.
> This also translates into org.mvel2.integration.impl.SimpleValueResolver constructor 18,454,778 calls for my present case.
> This is being invoked from the following in CompiledExpression:
>  public Object getValue(Object staticContext, VariableResolverFactory factory) {
>         if (!optimized) setupOptimizers();
>         if (importInjectionRequired) {
>             return execute(false, this, staticContext, new ClassImportResolverFactory(parserContext.getParserConfiguration(), factory));
>         }
> Where importInjectionRequired is almost always going to be true because of:
>  public void setParserContext(ParserContext parserContext) {
>         if ((this.parserContext = parserContext) != null) {
>             this.importInjectionRequired =
>                     parserContext.getImports() != null && parserContext.getImports().size() != 0;
>         }
>     }
> The parsercontext.imports seems to be always greater then 0 and to make matters worse I'm seeing a new Factory not only being created per rule but for the same rule multiple times.

-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: https://jira.jboss.org/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira

        


More information about the jboss-jira mailing list