[j-nsp] Finding drops - part 2

Jason Lixfeld jason-jnsp at lixfeld.ca
Wed Jan 30 15:13:18 EST 2019

Hi all,

Another question about finding drops.

This time, identifying the dropped packet counters that would increment as a result of being dropped during a controlled failure scenario.

                         +-----+                        +-----+
Loopback mx1 +------------------------+ mx2 +---+ Tester
                         +--+--+                        +--+--+
                            |                              |
                            |                              |
                            |                              |
                            |           +------+           |
                            +-----------+ qfx1 +-----------+

In the topology above, there’s rLFA protecting both 10.10.4/24 and 10.10.3/24.  Active path is mx1-mx2, backup path via qfx1.  Not sure if the mechanism is important, but figured I’d mention it anyway.

While the tester is generating packets towards the loopback, and the link between mx1 and mx2 is disabled, a certain volume of packets is lost, obviously [1].  
Looking at the all interface counters in the path from the tester to the loopback on mx1, mx2 and qfx1, I don’t see any errors or drops[2] in show interface x extensive.  Is it not reasonable to expect that the drops would show up in any of those counters.

[1] perhaps it’s important to note that the number of packets mx2 receives from the tester is equal to the number of packets the tester is reported as having transmitted towards mx2.

[2] the interface on mx1 and mx2 that connects mx1 and mx2 shows 1 input error (a Framing error) and 1 output carrier transition, obviously the result of the interface getting disabled.


More information about the juniper-nsp mailing list