Re: tcp/ip performance problem

[Pranav Desai]

On Thu, Oct 2, 2008 at 12:56 PM, Bernd Eckenfels <ecki@xxxxxxxxxxxx> wrote:
> In article <cc64e8ce0810011222o10f3c0e1tf288fa87b4bc1723@xxxxxxxxxxxxxx> you wrote:
>> hmm, actually I was looking for max queue length reached ... during
>> the run of the listening program. Is there any such value available ?
>
> netstat -s shows "TCP resets sent" counter which is incremented in a overrun
> scenario. Unfortunatelly this is not socket specific, however it is a good
> idea to monitor that counter closely.
>

Isn't that the "times the listen queue of a socket overflowed" value
in netstat -s ?
Regarding the max value, I want to see how close I get to overflowing,
so that I can tune it for a certain load. I will try to add it.

> This is BTW common to java servers (if GC runs).
>
>> patch to netstat ? so the /proc/...rx_queue should be ok to use to get
>> this number ? or were there some additional values exposed by the
>> kernel.
>
> Kernel patched to hand the connection-slots current/max to netstat for
> listening sockets.
>
> I think I will change the header for "netstat -l" to "maxlisten backlog" or
> something?
>
> Gruss
> Bernd
>
> PS: http://net-tools.berlios.de
> --
> To unsubscribe from this list: send the line "unsubscribe linux-net" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
--
To unsubscribe from this list: send the line "unsubscribe linux-net" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html