1

We are uploading 45000 rules to Kie Workbench. These rules are declared in a single excel sheet. We are planning to upload 5 files to a single KIe_project totalling upto 200,000 rules.

Problem Statement

  1. Currently for 20,000 rules , the validation and build takes lot of time We have to raise the VM options to be a follows, otherwise we are getting very high response times and some times java.io.buffersize exceed exception Xms=512m Xmx=7168 MaxPermGen=4096 If the rule file contains validation errors then validation alone will take 15 mins

  2. Converting a xls file to GDST format and then Build&Deploy project is taking more than 1 hour. and its not satisfactory that we are getting to know after 1 hour that there are validation errors or build failure or success in deployment

  3. Other connected users are unable to perform any operation on Kie-Workbench during upload/validation/deployment of such large files

  4. After successful convertion of an xls file to a gdst format, Guided Decision table edito is unable to load for even 10k records.

  5. During each validaiton or upload, we are getting the error that unable to deploy artifact to http://repo1.maven.org/maven2 . We are only uploading to Kie-WB, why is it going to the mentioned repo for deployment. Secondly we have our own nexus repository deployed in organization where assets needs to be deployed, not in the http://repo1.maven.org/maven2. For infor we are using maven's in kie-project's pom to deploy to our repository.

learner
  • 906
  • 2
  • 10
  • 39
  • Odds are high that you have made a poor decision for implementing your rules. Decision tables should not be used to generate thousands of rules with minor variations in the literals. - If you want to receive help, describe in some detail what you are trying to achieve. – laune Apr 30 '15 at 11:17
  • Hi Laura, I didn't get you . When should we use decision tables. You are right that we have minor literal changes in each rule. We are evaluating all facts on fixed variables and the outcome is also a fixed set of fact variables . I will upload my test files and model on to stackoverflow – learner Apr 30 '15 at 11:27
  • Hello Lorna, decision tables are best when you have various combinations of constraints (with minor variance due to different constraining literal values), but *not* when you need to evaluate the *same set of constraints* with a very large number of different constraining literal values. – laune Apr 30 '15 at 12:36

0 Answers0