On 2006-04-10 16:27:01, ulisses wrote:
> Hello
Hello again,
I'm answering myself here:
I had to increase the values of
/proc/sys/net/core/rmem_max
/proc/sys/net/core/wmem_max
/proc/sys/net/core/rmem_default
/proc/sys/net/core/wmem_default
in my case the following value was enough: 1351680
Ulisses
>
> I'm receiving ~20Mbps of netflow v5 packets from ~30 Cisco and Juniper
> routers, some of them are almost directly connected with the collector.
>
> I'm capturing traffic with flow-fanout with a powerful machine, without
> any link problems or interrupt or cpu load problems.
>
> Looking at the logs of the flow-fanout I see an high rate of drops
> and resets, for instance:
>
> flow-fanout[20649]: STAT: now=1144678980 startup=1144653562
> src_ip=130.206.a.b dst_ip=130.206.c.d
> d_ver=5 pkts=16426 flows=471176 lost=136003 reset=2 filter_drops=0
> send_nobufs=0
>
> The flow drop rate is ~22%:
>
> 136003/(471176+136003) = .22
>
> and that is not including the sequence resets.
>
> In general, the network is under-loaded so, there is no reason on why this
> drops occur in
> the network.
>
> Any comments about it?
>
> Any suggestion on how to debug this?
>
> Ulisses
>
_______________________________________________
Flow-tools mailing list
[EMAIL PROTECTED]
http://mailman.splintered.net/mailman/listinfo/flow-tools
