Re: iperf (server) termination with "read/write failed"


Are you terminating the clients early or letting them close themselves 
after the specified time (using -t)?? I have been thinking about this 
response to errors of closing the application and I think it stems back to 
the single threaded days. I have no problem with the warning message, but 
to kill the whole server due to a single client misbehaving seems drastic. 
I will let you know how I decide to change that.. Otherwise make sure that 
the client is not canceled before it is able to do the closing handshake 
with the server (which only matters for UDP).

Kevin


On Tue, 21 Jan 2003, Vivek Sawant wrote:

> Hi Iperfers,
> 
> When I run iperf in the (UDP) server mode, it often termianates with an 
> error "read failed" or "write failed". This happens after a client 
> connected to it has terminated. The experiment involves a series of 
> invocations of iperf clients, one at a time, each with different values 
> for certain parameters (mostly, different values for -b and -t 
> parameters). The client always connects to the same server.
> 
> This abnomal termination is a real problem for the automation of this 
> experiment. Not being able to count on the iperf server being around 
> during the course of a run will make my scripts unduely complicated.
> 
> Has this problem seen before? Any idea why this happens and is there a 
> workaround or a fix?
> 
> Thanks.
> 
> \vivek
> 



Other Mailing lists | Author Index | Date Index | Subject Index | Thread Index