Okay,  I'm not familiar with the testing tool you are using but I'll take a 
guess here.

Go to MSDN and look of the "linger" structure for sockets.  Read up on it.  
Hopefully it's here:

http://msdn.microsoft.com/en-us/library/ms739165(VS.85).aspx

also have a read of this:

http://msdn.microsoft.com/en-us/library/ms738547(VS.85).aspx

>Changed from 10 to 10000 at SetLinger

May not be what you want if high load servicing is what you are after.  You 
probably don't want ungracefull closes (linger < 1000) but you probably can't 
afford to give the network 10s to get things closed gracefully).

Read up on it and figure out what you really want.

That said it would be nice to tinker with your test to get a handle on what is 
causing the trouble (I'm betting it's Windows, but that's my bias).

>The result:
>C:\programacao\proxy>openload 192.168.1.20
>URL: http://192.168.1.20:80/
>Clients: 5
>MaTps 254.96, Tps 254.96, Resp Time  0.019, Err   0%, Count   257
>MaTps 255.86, Tps 264.00, Resp Time  0.019, Err   0%, Count   521
...
>MaTps 258.92, Tps 245.28, Resp Time  0.020, Err   0%, Count  3928
<Dead lock here>

Are you saying you formed 3928 connections in about 4 seconds?  Running 5 
clients in parallel that connect, get some data, disconnect?  Wow.

Try this:  

Set Linger down to 1000ms (1 Sec).  Should work great on your LAN.

Put a delay in your test.  For all five clients: connect, get some data, 
disconnect, delay 2 seconds.

Rerun the test and report what you get.

Need more load?  Increase the number of clients, don't decrease the delay.

Good luck.


-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php
_______________________________________________
synalist-public mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/synalist-public

Reply via email to