Re: A further question on using the -l option

[Jim M. <jim...@ex...>]

Marcelo Maraboli wrote:

>> iperf -u -c 172.18.1.1 -b 90m -t 300 -l 100B
> 
> 
> you also used "-i 1"... ;)

Nope, got the results from what the server was seeing (it had the -i 1 
option) but I wanted to show the client command that generated the data. 
Sorry, should have made this clear.

> 
>>
>> After a time I see the following from the server output:
>>
>> [  6] 154.0-155.0 sec  7.01 MBytes  58.8 Mbits/sec  0.013 ms 
>> 46088/119609 (39%)
>> [  6] 155.0-156.0 sec  7.00 MBytes  58.7 Mbits/sec  0.018 ms 
>> 43404/116817 (37%)
>> [  6] 156.0-157.0 sec  6.31 MBytes  52.9 Mbits/sec  0.016 ms    
>> 0/66157 (0%)
>> [  6] 157.0-158.0 sec  6.33 MBytes  53.1 Mbits/sec  0.017 ms    
>> 0/66423 (0%)
>>
>> The number of datagrams/s seems to lower and packets stop being 
>> dropped or at least only a few get lost. The time to reach this point 
>> seems to vary but it appears to be consistent behaviour if I repeat 
>> the test. Is there any reason why the Iperf client would start to send 
>> fewer packets, as this is what appears to happen, or am I mistaken 
>> about what is happening?
> 
> 
> The standard test should be 60 seconds according to RFC 2544,
> but this behaviour means that something is "stabilizing" or that
> Iperf is not reporting correct numbers..

Looks like I need to try to find out why this is happening if there is 
no obvious answer. We want to do some tests that last for a long time on 
new fibre connections, hence didn't want things changing, or being 
reported incorrectly, during the test.

Thanks for helping out with both questions,

Jim Mozley