Hi Adam

I do not need unicast and non unicast combined. just additional non unicast would be fine.
We just had a misbehaving customer which broadcasted out of his rack to our core. we like to be able to get alerted on that.

I also noticed that the non unicast graphs do not show the tx values, they just appear in the labels section. is that only on my setup or is this like that?
(Screenshot attached)

Regards

  Matthias

On 15/07/16 20:57, Adam Armstrong wrote:
 Hi Matthias,

 Storing this information would have too great an impact on the performance of the ports table, I think.

 It would involve 4 extra fields in the ports table (it would be worthless to store "pps", as it would just be unicast+nonunicast).

 To store this we'd need in/out counters and in/out rate. If we added this, you
would be very unpopular with some of our larger installations... ;)

 Are there really reasons to alert on non-unicast in the wild?

 Thanks,
 Adam.

 Adam.

 Sent from BlueMail <http://www.bluemail.me/r>

 On 15 Jul 2016, at 20:51, Matthias Cramer <matthias.cramer@iway.ch <mailto:matthias.cramer@iway.ch>> wrote:

     Hi There

     Is there a metric for normal packets per second and not only unicast packets (ifInUcastPkts_rate) and also a metric for non unicast packets per second. According to the doc at
     http://www.observium.org/docs/entities/#metrics_1 this is not existent... I would like to make an alert on high pps and if possible on high non unicast pps.

     Any hint is appreciated.

     Regards

       Matthias

 observium mailing list

observium@observium.org
 http://postman.memetic.org/cgi-bin/mailman/listinfo/observium