When I run the TPCW bookstore application, I found the server behave
Hardware: a client PC, two server PCs. Each PC has two 1.59 GHz AMD
Opteron, 2GB RAM, and a 30 GB disk. One server machine runs theWeb server
and application server software, while the other contains the database. All
the machines run 2.4 Linux Kernel and are connected through a LAN of
Web Server: Tomcat 5.5.12
Database: mysql 5.0.18
Client Workload Generator: the freely available Java TPC-W
implementation developed by the PHARM research group at the University of
Wisconsin-Madison http://www.ece.wisc.edu/~pharm/tpcw.shtml The file descriptor limit has been increased to 8192.
The maxconnection of Tomcat is 1000.
To solve the heap exception of Tomcat, JAVA_OPTS='-Xms256m -Xmx256m'
Q1: When I emulate 400 EBs ( the command line: java rbe.RBE -EB
rbe.EBTPCW1Factory 400 -OUT run1.m -RU 100 -MI 600 -RD 100 -WWW
http://tomcatserver:8080/ -CUST 288000 -DEBUG 2") , the tomcat memory drops
severly from 1G to 5M.
Q2: Sometimes the tests complete without any error. In other runs when
getting Html from Tomcat, the ioexceptions of url.openStream() are caught
and the client is returned with "Unable to open the html ....".
Could you kindly tell me why the tomcat cosumes so much memory? And
what may cause the ioexception of getting html? Does it correlate with the
implementation of TPCW?