Thanks for getting back to me. The reason I thought it shared the load is because of this test I did:
* A simple script making two requests, 1 to google and 1 to youtube * 2 slave servers * 1 master server When I read the results of this (Result tree view(?) output to file) they gave. Threadgroup 1-1: One request from each slave Threadgroup 1-2: One request from each slave. I assume the first number means the first threadgroup and the second number refers to the index of the concurrent thread. I would have expected both threadgroup 1-1 requests to be made by one slave and both threadgroup 1-2 requests to be made by the other. Or am I reading this wrong? Also, if I want multiple servers to carry out the full instruction set once, should I set the number of threads in the test plan to the number of servers, or just 1 thread seeing as each server gets an identical copy and will carry out a single thread? Thanks again, Drew sebb-2-2 wrote: > > On 17/04/2009, drubix <andrew.schr...@gmail.com> wrote: >> >> Hi, >> >> I'm attempting to run a load test against our system but because of the >> way >> out authentication is set up each user needs a different IP address. I >> can >> achieve this using IP aliasing and running multiple instances of JMeter >> with >> different local addresses. >> >> What I'd like to have happen is this: >> * Start each slave JMeter instance on a different RMI port and a >> different >> local IP address > > OK > >> * Start the master JMeter instance which distributes the script to all >> of >> the slaves > > OK, you need to add all the slave addresses to the list of remote hosts. > >> * Have the slaves work through the test script themselves, managing >> their >> own authentication and cookies >> * Repeat the script for a set length of time >> * Have all the slaves report back to the master who collates the data >> >> However, JMeter doesn't seem to work that way by default. After doing >> some >> preliminary tests it seems that final results don't necessarily have a >> correlation between a JMeter instance and the threadgroup it operates >> in. > > Not sure what you mean by that. > >> It seems that JMeter shares the load dynamically rather than just >> dishing >> out an entire thread group to each instance. > > Not so, JMeter sends the same test plan to all the slaves mentioned in > the remote_hosts property. > > What makes you think JMeter shares the load? > >> Unfortunately, as the IPs for >> each thread/instance will no longer match up, this doesn't work for our >> particular situation. >> >> Is there a way to force JMeter to handle each thread with a different >> instance > > No, the entire test plan is sent to all remote hosts. > >> or, even better, a way to handle IPs (attaching a single username, >> password and IP address to a single thread) which would get rid of these >> problems entirely? > > The local IP cannot currently be specified per thread, only per JMeter > instance, and only for the HttpClient version of the HTTP sampler. > > It should be possible to enhance JMeter to do this with the HttpClient > sampler, but it's highly unlikely this can be done with the Java Http > sampler. > >> Thanks for your help, >> >> Drew >> >> -- >> View this message in context: >> http://www.nabble.com/One-thread-per-instance-on-distributed-system-tp23092193p23092193.html >> Sent from the JMeter - User mailing list archive at Nabble.com. >> >> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: jmeter-user-unsubscr...@jakarta.apache.org >> For additional commands, e-mail: jmeter-user-h...@jakarta.apache.org >> >> > > --------------------------------------------------------------------- > To unsubscribe, e-mail: jmeter-user-unsubscr...@jakarta.apache.org > For additional commands, e-mail: jmeter-user-h...@jakarta.apache.org > > > -- View this message in context: http://www.nabble.com/One-thread-per-instance-on-distributed-system-tp23092193p23095314.html Sent from the JMeter - User mailing list archive at Nabble.com. --------------------------------------------------------------------- To unsubscribe, e-mail: jmeter-user-unsubscr...@jakarta.apache.org For additional commands, e-mail: jmeter-user-h...@jakarta.apache.org
