Re: Negative Latency
On 20/10/2009, Carl Shaulis cshau...@homeaway.com wrote: Riddle me this batman! If I execute a test of 100 threads using my machine as the master and a linux machine as the slave with each thread executing a single request I do NOT get any negative response times. If I schedule a test of 100 threads to run 5 minutes looping indefinitely the single request, I am getting negative response times. This does not make sense to me. More thoughts? The elapsed times are calculated by the sampler, so clock skew won't affect them. The times Which version of JMeter are you using? JVM? The elapsed time calculation depends on both of these. Do the timestamps look reasonable? Carl On 10/20/09 1:14 PM, Peter Lin wool...@gmail.com wrote: both systems must be insync. That's fundamental to all distributed applications, including distributed testing. peter On Tue, Oct 20, 2009 at 2:13 PM, Carl Shaulis cshau...@homeaway.com wrote: The difference appears to be about 10 seconds between the clock on my machine and the slave server. I added a constant timer and that made no difference. Do the two machines really have to be set down to the exact second? I would think we are measuring the delta between start and stop on the same machine, so the clocks should not matter. Thanks, Carl On 10/20/09 1:06 PM, Deepak Shetty shet...@gmail.com wrote: are the time clocks on both machines in sync? On Tue, Oct 20, 2009 at 11:02 AM, Carl Shaulis cshau...@homeaway.comwrote: Hello, We have recently set up a distributed JMeter environment. I am using my MacBook Pro as the Master and a Linux machine as the slave. I executed a very simple test for 5 minutes, where 500 concurrent users access a static html page. The results showed an average response time of 0 ms. Looking more closely at the data there are numerous transactions that look like this. Thread Name: SorryPageTest 1-97 Sample Start: 2009-10-20 12:42:29 CDT Load time: -897 Latency: -897 Size in bytes: 1723 Sample Count: 1 Error Count: 0 Response code: 200 Response message: OK How can you get a negative load time and negative latency with a 200 response code? Help! Carl - 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 - 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
Re: Negative Latency
Thanks for the reply! We are updating the JVM on our linux machine, since it was 1.4.2 and my machine was 1.5.0. Carl On 10/23/09 10:05 AM, sebb seb...@gmail.com wrote: On 20/10/2009, Carl Shaulis cshau...@homeaway.com wrote: Riddle me this batman! If I execute a test of 100 threads using my machine as the master and a linux machine as the slave with each thread executing a single request I do NOT get any negative response times. If I schedule a test of 100 threads to run 5 minutes looping indefinitely the single request, I am getting negative response times. This does not make sense to me. More thoughts? The elapsed times are calculated by the sampler, so clock skew won't affect them. The times Which version of JMeter are you using? JVM? The elapsed time calculation depends on both of these. Do the timestamps look reasonable? Carl On 10/20/09 1:14 PM, Peter Lin wool...@gmail.com wrote: both systems must be insync. That's fundamental to all distributed applications, including distributed testing. peter On Tue, Oct 20, 2009 at 2:13 PM, Carl Shaulis cshau...@homeaway.com wrote: The difference appears to be about 10 seconds between the clock on my machine and the slave server. I added a constant timer and that made no difference. Do the two machines really have to be set down to the exact second? I would think we are measuring the delta between start and stop on the same machine, so the clocks should not matter. Thanks, Carl On 10/20/09 1:06 PM, Deepak Shetty shet...@gmail.com wrote: are the time clocks on both machines in sync? On Tue, Oct 20, 2009 at 11:02 AM, Carl Shaulis cshau...@homeaway.comwrote: Hello, We have recently set up a distributed JMeter environment. I am using my MacBook Pro as the Master and a Linux machine as the slave. I executed a very simple test for 5 minutes, where 500 concurrent users access a static html page. The results showed an average response time of 0 ms. Looking more closely at the data there are numerous transactions that look like this. Thread Name: SorryPageTest 1-97 Sample Start: 2009-10-20 12:42:29 CDT Load time: -897 Latency: -897 Size in bytes: 1723 Sample Count: 1 Error Count: 0 Response code: 200 Response message: OK How can you get a negative load time and negative latency with a 200 response code? Help! Carl - 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 - 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 - To unsubscribe, e-mail: jmeter-user-unsubscr...@jakarta.apache.org For additional commands, e-mail: jmeter-user-h...@jakarta.apache.org
Negative Latency
Hello, We have recently set up a distributed JMeter environment. I am using my MacBook Pro as the Master and a Linux machine as the slave. I executed a very simple test for 5 minutes, where 500 concurrent users access a static html page. The results showed an average response time of 0 ms. Looking more closely at the data there are numerous transactions that look like this. Thread Name: SorryPageTest 1-97 Sample Start: 2009-10-20 12:42:29 CDT Load time: -897 Latency: -897 Size in bytes: 1723 Sample Count: 1 Error Count: 0 Response code: 200 Response message: OK How can you get a negative load time and negative latency with a 200 response code? Help! Carl
Re: Negative Latency
are the time clocks on both machines in sync? On Tue, Oct 20, 2009 at 11:02 AM, Carl Shaulis cshau...@homeaway.comwrote: Hello, We have recently set up a distributed JMeter environment. I am using my MacBook Pro as the Master and a Linux machine as the slave. I executed a very simple test for 5 minutes, where 500 concurrent users access a static html page. The results showed an average response time of 0 ms. Looking more closely at the data there are numerous transactions that look like this. Thread Name: SorryPageTest 1-97 Sample Start: 2009-10-20 12:42:29 CDT Load time: -897 Latency: -897 Size in bytes: 1723 Sample Count: 1 Error Count: 0 Response code: 200 Response message: OK How can you get a negative load time and negative latency with a 200 response code? Help! Carl
Re: Negative Latency
The difference appears to be about 10 seconds between the clock on my machine and the slave server. I added a constant timer and that made no difference. Do the two machines really have to be set down to the exact second? I would think we are measuring the delta between start and stop on the same machine, so the clocks should not matter. Thanks, Carl On 10/20/09 1:06 PM, Deepak Shetty shet...@gmail.com wrote: are the time clocks on both machines in sync? On Tue, Oct 20, 2009 at 11:02 AM, Carl Shaulis cshau...@homeaway.comwrote: Hello, We have recently set up a distributed JMeter environment. I am using my MacBook Pro as the Master and a Linux machine as the slave. I executed a very simple test for 5 minutes, where 500 concurrent users access a static html page. The results showed an average response time of 0 ms. Looking more closely at the data there are numerous transactions that look like this. Thread Name: SorryPageTest 1-97 Sample Start: 2009-10-20 12:42:29 CDT Load time: -897 Latency: -897 Size in bytes: 1723 Sample Count: 1 Error Count: 0 Response code: 200 Response message: OK How can you get a negative load time and negative latency with a 200 response code? Help! Carl - To unsubscribe, e-mail: jmeter-user-unsubscr...@jakarta.apache.org For additional commands, e-mail: jmeter-user-h...@jakarta.apache.org
Re: Negative Latency
both systems must be insync. That's fundamental to all distributed applications, including distributed testing. peter On Tue, Oct 20, 2009 at 2:13 PM, Carl Shaulis cshau...@homeaway.com wrote: The difference appears to be about 10 seconds between the clock on my machine and the slave server. I added a constant timer and that made no difference. Do the two machines really have to be set down to the exact second? I would think we are measuring the delta between start and stop on the same machine, so the clocks should not matter. Thanks, Carl On 10/20/09 1:06 PM, Deepak Shetty shet...@gmail.com wrote: are the time clocks on both machines in sync? On Tue, Oct 20, 2009 at 11:02 AM, Carl Shaulis cshau...@homeaway.comwrote: Hello, We have recently set up a distributed JMeter environment. I am using my MacBook Pro as the Master and a Linux machine as the slave. I executed a very simple test for 5 minutes, where 500 concurrent users access a static html page. The results showed an average response time of 0 ms. Looking more closely at the data there are numerous transactions that look like this. Thread Name: SorryPageTest 1-97 Sample Start: 2009-10-20 12:42:29 CDT Load time: -897 Latency: -897 Size in bytes: 1723 Sample Count: 1 Error Count: 0 Response code: 200 Response message: OK How can you get a negative load time and negative latency with a 200 response code? Help! Carl - 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