Ryan Senior
Wed, 04 Aug 2010 08:19:24 -0700
Hello, I'm seeing some odd behavior when doing some performance testing of SwiftOWLIM and don't know how to troubleshoot the issue further. The benchmark I'm using is SP^2 (found here: http://dbis.informatik.uni-freiburg.de/index.php?project=SP2B). Also on that site is a program to generate datasets for use in the test. Using a dataset of size 10,000, 50,000 or 250,000 triples, I do not see this issue. Only on 1 million triples.
When running the second query in the benchmark, it runs successfully 16 times at roughly 32 seconds each invocation. Then the 17th time, the client issues the query and no response is given after 1 hour (and so my code times out the query). From here the triplestore seems to be in some inconsistent state, even though the machine is rebooted. The next query will run fine to completion (3a from the above paper) however every subsequent query will not return any results. The server responds back (there is no timeout) however, always responding with no results, when there should be some. I am getting this behavior consistently (8 out of the 8 times I have ran it). Software Info: SwiftOWLIM 3.3 Sesame 2.3.2 JDK 1.6.0 rev 20 Tomcat 6 The only real configuration change made from the install instructions is increasing the max heap available to Tomcat from 128m to 3g. The machine used for the test is an Amazon EC2 m1.large instance and has the below characteristics: Ubuntu 10.04 - 64-bit 7.5 GB memory 2 CPU ~1.5Ghz each 15GB Disk I looked at catalina.out to see if anything was logged about this and I didn't see anything. Is there somewhere else to look for log information? Is there some way to view other bits of runtime information to see if there is something wrong with the running instance? Thanks, Ryan
_______________________________________________ OWLIM-discussion mailing list OWLIM-discussion@ontotext.com http://ontotext.com/mailman/listinfo/owlim-discussion