我们正在使用drools 5.5最终版本。我们有数千个对象和两个规则,因此我们明智地获取块(100大小)的对象并为每个块创建知识库并触发规则。因为知识库的创建是昂贵的我们我们正在创建性能问题。所以我们创建了一次知识库并在这种情况下使用该知识库,在这个案例中,从第6个块开始执行4到5个块后,虽然匹配在那里但规则没有被解雇。请指出可以是什么完成。
public static KnowledgeBase getPackageKnowledgeBase(PackageDescr pkg){
KnowledgeBuilderConfiguration builderConf = KnowledgeBuilderFactory.newKnowledgeBuilderConfiguration();
KnowledgeBuilder kbuilder = KnowledgeBuilderFactory.newKnowledgeBuilder(builderConf);
kbuilder.add(ResourceFactory.newDescrResource(pkg), ResourceType.DESCR);
Collection<KnowledgePackage> kpkgs = kbuilder.getKnowledgePackages();
if(kbuilder.hasErrors()){
LOGGER.error(kbuilder.getErrors());
}
KnowledgePackage knowledgePackage = kpkgs.iterator().next();
KnowledgeBase kbase= KnowledgeBaseFactory.newKnowledgeBase();
kbase.addKnowledgePackages(Collections.singletonList(knowledgePackage));
return kbase;
}
chunkSize=100;
int start = 0;
Count = -1;
KnowledgeBase kbase=getPackageKnowledgeBase(pkgdscr)//pkgdscr contails all rules got from db
while(Count!=0 && Count <= chunkSize ){
LOGGER.debug("Deduction not getting "+mappedCustomerId);
Objects inputObjects = handler.getPaginatedInputObjects(start);
Count = inputObjects.size();
start=start+chunkSize;
StatefulKnowledgeSession ksession = kbase.newStatefulKnowledgeSession();
for(Object object:inputObjects){
ksession.insert(object);
}
ksession.fireAllRules();
ksession.dispose();
}
答案 0 :(得分:0)
以下是循环的重要部分。在我看来,一旦Count超过chunkSize(100),这个循环就会终止。你确定这不会发生吗?
while(Count!=0 && Count <= chunkSize ){
Objects inputObjects = ...;
Count = inputObjects.size();
...
StatefulKnowledgeSession ksession = ...;
for(Object object:inputObjects){
ksession.insert(object);
}
ksession.fireAllRules();
...
}