Hi Andy, Sorry that the picture didn't show (I don't know why). The CPU consumption of the Fuseki process is over 600% (the server has 8 cores).
For now, I think the problem is probably caused by too much log IOs in our own code. We customized LARQ and included it in Fuseki. There were too many "System.out.println()" in our code. I've removed all the console output code and it seemed ok up to now. Thanks Tao -----Original Message----- From: Andy Seaborne [mailto:[email protected]] Sent: Tuesday, April 16, 2013 6:55 PM To: [email protected] Subject: Re: Abnormal CPU consumption by Fuseki On 16/04/13 08:24, Tao (陶信东) wrote: > Hi, > > We’ve being using Fuseki/TDB for 2 years. Recently we noticed that > sometimes it was using huge amount of our CPU resources, as can be > seen from the cited email. I'm afraid that list of threads doesn't tell me anything; maybe someone else can seee something in the list. No Jena code is mentioned except for the start up thread (which is hanging around (= "WAITING") doing nothing). Several threads seem to be reading HTTP requests (in Jetty) (are the clients being slow?) What is the setup? (machine size? cores? version of software?) What has changed recently? Does it have enough heap (maybe the GC is taking a lot of the time)? What causes you to think it is consuming a lot of CPU? Andy > > Did anyone encounter the same situations? What is the most probable problem? > > Thanks > > Tao > > *From:*陈仕员[mailto:[email protected]] > *Sent:* Tuesday, April 16, 2013 3:10 PM > *To:* 陶信东 > *Cc:* 夏炜 > *Subject:* qna00002Fuseki进程cpu很高 > > jstack 日志： > > 2013-04-16 15:05:32 >
