Forum Discussion

PhilipBauwens's avatar
PhilipBauwens
New Contributor
14 years ago

Limitation on open sockets ?

Hi,

Does loadui have some kind of limitation on the number of open sockets ?

I have a simple test, which is generating a soap request to a server and waits for the answer.
I send requests at a rate of 10 with a burst of 10. The value for concurrentSamples is set to 10000.

I can see that the number of running request increases to 1000 and then load ui starts queuing. I would expect this would go up to 10000, before queuing is started.
When checking on the server side, I can see only 503 parallel requests, where I am expecting to see 1000 parallel requests, as indicated by loadui.

Verification (netstat -an | grep -c 80) on the server that is running loadui, shows that I only have 524 sockets open.

The limitations is on the loadui system, because the number of parallel requests on the server under test, increases to 937 when sending loadui traffic from 2 machines.

Does loadui have some limit on the number of open sockets ?
Why is loadui indicating 1000 running samples, when there are only 524 connections open ?

2 Replies

  • PM_r's avatar
    PM_r
    Icon for Staff rankStaff
    14 years ago
    Hi,

    There's no limitation enforced by loadUI. You need to change the corresponding setting in the soapUI Runner as well.

    Connections is not the same thing as Running Samples, especially not when your sample is a complete soapUI TestCase.

    Why would you need more than 1000 parallel requests? That's more than Bing handles...

    /Henrik
  • PhilipBauwens's avatar
    PhilipBauwens
    New Contributor
    14 years ago
    The request that we are generating using loadui, result in some database queries that take some time to execute.
    With the expected load and the current processing time in our app/database, this will result in a high number of parallel sessions.

    As we would like to find out the maximum number of parallel sessions that our current implementation can support, we need to have more than 1000 request being handled simultaneously.