[c-nsp] Bug with IOS-XR and SPAN ports?

[Hank Nussbacher]

On 24/11/2016 16:33, James Bensley wrote:
> On 9 November 2016 at 21:24, Hank Nussbacher <hank at efes.iucc.ac.il> wrote:
>> We upgraded our AS9010 this morning from 5.1.3 to 5.3.3 and encountered
>> an issue with SPAN ports.
>>
>> Our SPAN port config looks like this:
>> monitor-session No1 ethernet
>>  destination interface TenGigE0/1/1/4
>> !
>> monitor-session No2 ethernet
>>  destination interface TenGigE0/1/1/5
>> !
>> interface GigabitEthernet0/1/0/10
>>  monitor-session No1 ethernet
>> !
>> interface TenGigE0/0/1/2
>>  monitor-session No2 ethernet
>> !
>> interface TenGigE0/1/1/7
>>  monitor-session No1 ethernet
>> !
>> interface TenGigE0/1/1/7.100
>>  monitor-session No1 ethernet
>> !
>> interface TenGigE0/1/1/7.333
>>  monitor-session No1 ethernet
>> !
>> interface TenGigE0/1/1/7.667
>>  monitor-session No1 ethernet
>>
>> On IOS-XR 5.1.3 - no problem with the above config.  On IOS-XR 5.3.3 we
>> found that when the destination SPAN port gets overloaded, in addition
>> to dropping the pkt destined to the destination port (no big deal),
>> IOS-XR also silently drops the pkt from the port that was being SPANed
>> (yes, a big deal).  Has anyone encountered this, or has a workaround or
>> a Cisco bugid?
>>
>> Thanks,
>> Hank
> Hi Hank,
>
> Did you get to the bottom of this? This could be quite a problem. Did
> you find out if it was a bug per chance?
We have an open TAC case on this and had to have it moved from the Far
East to Europe TAC due to timezone differences. 
They are busy duplicating the situation and hope to respond shortly. 
I'll update here when I have more info.

In the interim we have stabilized by removing SPAN on the primary source
port and only SPAN on the sub-interface, i.e.

interface TenGigE0/1/1/7
 *no monitor-session No1 ethernet*
!
interface TenGigE0/1/1/7.100
 monitor-session No1 ethernet

Regards,
-Hank


>
> Cheers,
> James.
>