Understanding network errors

[Guest lyndsayp]

We're trying to understand the source of the network errors that WAPT includes in its report, as we'd like to run tests that don't generate any network errors.

 

We're using WAPT Pro 2.1 to test an ASP.Net web application, running on IIS 6, on Windows Server 2003. The tests are being conducted using 2 physical, client PC's, connected to the physical web-server machine via a dedicated, isolated network switch. We're currently seeing an average of 45.1% network errors per minute (but no other errors) for a 26 minute run of 50 concurrent users (ramp up from 1, with 2 users being added per minute). This scenario targets a minimalistic web app that gives a linear response time of 0.009-0.09 seconds and an average number of sessions per minute of 48 000.

 

Can anyone shed light on what WAPT counts as being a network error, and whether its likely that we're simply placing an unrealistically high load on the system (thus the network error rate is as expected) or whether we may have a poor hardware/software configuration (thus could reduce the network error rate)?

 

Many thanks

[sergei 0]

Could you tell me the errors that WAPT includes in report? For that, please, enable logs in WAPT and find exact network error codes.

[Guest lyndsayp]

Could you tell me the errors that WAPT includes in report? For that, please, enable logs in WAPT and find exact network error codes.

 

Sure, its:

Response-Failed[Network error: 10048 - "Only one usage of each socket address (protocol/network address/port) is normally permitted."]

[Guest lyndsayp]

Could you tell me the errors that WAPT includes in report? For that, please, enable logs in WAPT and find exact network error codes.

 

The detailed error message led me to this MSDN article. I updated both clients to have MaxUserPort of 65534 and TcpTimedWaitDelay of 30. This reduced the amount of network errors massively, to only 5 every other minute.

 

What's puzzled me now is that this has also caused the total number of sessions to drop:

 

Before: 521226 successful sessions, 544550 failed sessions; total of 1,065,776 sessions

After: 684748 successful sessions, 25 failed sessions; total of 684,773 sessions

 

Any idea why this drop has occurred?

[sergei 0]

The detailed error message led me to this MSDN article. I updated both clients to have MaxUserPort of 65534 and TcpTimedWaitDelay of 30. This reduced the amount of network errors massively, to only 5 every other minute.

 

What's puzzled me now is that this has also caused the total number of sessions to drop:

 

Before: 521226 successful sessions, 544550 failed sessions; total of 1,065,776 sessions

After: 684748 successful sessions, 25 failed sessions; total of 684,773 sessions

 

Any idea why this drop has occurred?

 

Because some time is required to perform a successful session whereas the errors occur very quickly.