All:
Does anyone know of the default setting for the PIM DM prune limit timer in Cisco's implementation? This is the timer used to prevent a node from sending an upstream prune in response to every incoming packet (see RFC 3973, sec 4.4.1).
In a lab setting I am seeing results that differ from what is described in the RFC as outlined below.
Consider a three node network R1, R2, R3; each router has a single loopback l0 and a single ethernet interface.
The ethernet interfaces are on the same L2 domain with all three interfaces in the same IP subnet, e.g, 10.0.0.1, 10.0.0.2, and 10.0.0.3.
Assume R3 has a receiver on group G configured with:
interface loopback 0
ip igmp join-group G
and R1 sends icmp pings with:
ping G repeat 100 source l0
From my understanding of the RFC the expected behavior is:
- R2 - with no receiver - will send a PRUNE upstream to R1.
- R1 will set its prune pending timer and also send a PRUNE ECHO - prune packet with itself as the upstream neighbor
- R3 will override with PRUNE with a JOIN
- R2 will continue to receive packets for group G on it's RPF interface but will not send another PRUNE until its prune limit timer - per RFC the default value is 210s - has expired.
- After the prune limit timer has expired on R2, the next packet received will trigger a new PRUNE.
In short, the pattern of three packets, PRUNE, PRUNE ECHO, JOIN - should be repeated aproximately every 210s.
What is I have observed in the lab using three CSR1000Vs is:
* an initial PRUNE from R2 followed by a PRUNE override from R3.
* after approximately 125 s R2 sends another PRUNE, followed by R1 sending a PRUNE ECHO and then a JOIN from R3.
* this three packet pattern the repeats approximately every 10s.
What am I missing ?
