To include the whole packet in every sampled header you could try
changing SFL_MAX_DATAGRAM_SIZE in lib/sflow.h to something like 2000
and then recompile (or maybe 10000 if you want to sample whole jumbo
frames too),  but with the settings you show you will get random
1-in-500 packet-samples. Setting the sampling rate to 1:1 will not
perform well at all.  It's not what sFlow is designed for.  If you
really want *all traffic* then you might be better off using a
replication openflow rule.  Maybe something along these lines:
https://blog.sflow.com/2013/11/udp-packet-replication-using-open.html

------
Neil McKee
InMon Corp.

------
Neil McKee
InMon Corp.



On Mon, Dec 9, 2019 at 10:35 AM Oliver Dzombic <i...@layer7.net> wrote:
>
> Hi Folks,
>
> when we have high rates of packets per second sflow will not export this
> traffic to the collector.
>
> Inside of ovs-vswitchd.log we can see tons of entries like:
>
> 2019-12-09T18:15:04.739Z|30436|sflow(handler17)|WARN|sFlow agent error:
> sfl_agent_error: receiver: flow sample too big for datagram
>
>
> Our configuration:
>
>
> ovs-vsctl -- --id=@sflow create sflow agent=127.0.0.1
> target="\"127.0.0.1:6343\"" header=1518 sampling=500 polling=1 -- set
> bridge ovsbridge sflow=@sflow
>
>
> How can we improve this ? We would like to receive all traffic at the
> collector, no matter how many packets or what packetsize will come in.
>
> Any help / hint / idea is greatly appriciated.
>
> Thank you !
>
>
> --
> Mit freundlichen Gruessen / Best regards
>
> Oliver Dzombic
> Layer7 Networks
>
> mailto:i...@layer7.net
>
> Anschrift:
>
> Layer7 Networks GmbH
> Zum Sonnenberg 1-3
> 63571 Gelnhausen
>
> HRB 96293 beim Amtsgericht Hanau
> Geschäftsführung: Oliver Dzombic
> UST ID: DE259845632
> _______________________________________________
> discuss mailing list
> disc...@openvswitch.org
> https://mail.openvswitch.org/mailman/listinfo/ovs-discuss
_______________________________________________
discuss mailing list
disc...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-discuss

Reply via email to