Cisco Cisco Dynamic Fabric Automation for OpenStack Troubleshooting Guide
Cisco Dynamic Fabric Automation Production Troubleshooting Guide
68
join, and (3) repeat the same process as described above for (*,G) state. The primary difference will be that LHR
acorn3 will select an RPF neighbor of the FHR leaf switch (if the source is inside the fabric), or again for one of the
border leaf switches (if the source is outside of the fabric).
border leaf switches (if the source is outside of the fabric).
In situations where a receiver is located inside the fabric and is interested in a group sourced outside of the fabric,
the IGMP join will be translated to a BGP Join at the LHR leaf switch, and that BGP Join will be translated into a
PIM Join at the border leaf switch and forwarded to the RPF PIM neighbor outside of the fabric. If there are redun-
dant border leaf switches and the source is outside the fabric, only the border leaf switch which is determined as
the fabric forwarder for the respective group will forward the multicast traffic into the fabric to prevent duplicates.
the IGMP join will be translated to a BGP Join at the LHR leaf switch, and that BGP Join will be translated into a
PIM Join at the border leaf switch and forwarded to the RPF PIM neighbor outside of the fabric. If there are redun-
dant border leaf switches and the source is outside the fabric, only the border leaf switch which is determined as
the fabric forwarder for the respective group will forward the multicast traffic into the fabric to prevent duplicates.
You have now successfully verified BGP join propagation through the fabric from server leaf switch to the border leaf
switches, and have the commands needed to be able to isolate which device is having a problem if troubleshooting is
needed.
switches, and have the commands needed to be able to isolate which device is having a problem if troubleshooting is
needed.
5.1.3 Fabric Forwarder Election for Redundant Border Leaf Switches
When there are redundant border leaf switches and a source outside of the fabric, the border leaf switch that will be
responsible for forwarding the multicast traffic into the fabric is known as the fabric forwarder (do not confuse this with
vPC+ forwarder which is discussed later). The fabric forwarder is determined on a per (S,G) basis. This means that
both border leaf switches could be fabric forwarders for different (S,G)s. The fabric forwarder for a specific (S,G) is
determined by a hash algorithm on the LHR leaf switch that is used to specify which border leaf switch the BGP (S,G)
Join should be destined to. Both border leaf switches will actually receive this (S,G) Join, however only the border leaf
switch in which the join is destined to (based on the hash algorithm on the LHR) will be elected the Fabric Forwarder
and forward the (S,G) traffic into the fabric.
responsible for forwarding the multicast traffic into the fabric is known as the fabric forwarder (do not confuse this with
vPC+ forwarder which is discussed later). The fabric forwarder is determined on a per (S,G) basis. This means that
both border leaf switches could be fabric forwarders for different (S,G)s. The fabric forwarder for a specific (S,G) is
determined by a hash algorithm on the LHR leaf switch that is used to specify which border leaf switch the BGP (S,G)
Join should be destined to. Both border leaf switches will actually receive this (S,G) Join, however only the border leaf
switch in which the join is destined to (based on the hash algorithm on the LHR) will be elected the Fabric Forwarder
and forward the (S,G) traffic into the fabric.
Note: The hash algorithm procedures are similar to the upstream-multicast hop (UMH) selection (per RFC 6513).
In this example we will verify whether acorn6 or acorn5 is elected as fabric forwarder for (S,G) (172.16.1.101,
225.1.110.112), which is sourced behind the Layer-3 edge router outside of the fabric.
225.1.110.112), which is sourced behind the Layer-3 edge router outside of the fabric.
High-level Overview for Fabric Forwarder Election