[c-nsp] multicast NOT in HW on 7600

[Liviu Pislaru]

hi Phil,

because R2 is not part of the NEW TOPOLOGY, R1 and R2 are not PIM
neighbors anymore.

[...]

R1#sh run | inc ^ip (pim|multi)
ip multicast-routing 
ip multicast-routing vrf XXX
ip pim ssm range multicast-mdt

R1#sh run int Vlan10
interface Vlan10
 description XXX
 mtu 9216
 ip address XXX
 no ip redirects
 no ip unreachables
 no ip proxy-arp
 ip mtu 1524
 ip pim sparse-mode
 load-interval 30
 mpls mtu 1542
 mpls ip
end

------------

R2#sh run | inc ^ip (pim|multi)
ip multicast-routing 
ip pim ssm range multicast-mdt

R2#sh run int Vlan10
interface Vlan10
 description XXX
 mtu 9216
 ip address XXX
 no ip redirects
 no ip unreachables
 no ip proxy-arp
 ip mtu 1524
 ip pim sparse-mode
 load-interval 30
 mpls mtu 1542
 mpls ip
end

[...]

i issue the command "mls ip multicast" AGAIN on R2 and it looks like
this solved my problem.

BEFORE
-------------------------------------------------
R2#sh mls ip multicast summary
0 MMLS entries using 0 bytes of memory
Number of partial hardware-switched flows: 0
Number of complete hardware-switched flows: 0

Directly connected subnet entry install is enabled
Current mode of replication is Egress
Auto-detection of replication mode is enabled
Consistency checker is enabled
Bidir gm-scan-interval: 10



R2#sh mls ip multicast statistics
MLS Multicast configuration and state:
Counters last cleared                         Never
Router Mac                                    001b.0de6.7b80
MLS multicast operating state                 IDLE
Layer 3 Switching H/W Version                PFC III+
Maximum number of allowed outstanding message 20
Maximum size reached from feQ                 0
Maximum size reached from screq               2
Feature Notification sent (simple/rtr-mac)    0/0
Feature Notification Ack received             0
Unsolicited Feature Notification received     0
MSM sent/Received                             0/0
Delete notifications received                 0
sgc oif delete notifications received         0
Flow Statistics messages received             0
Restart Notification messages received        0
Cleanup Send/Resp-rx seq number               0/0



AFTER:
-----------------------------------------------

R2#sh mls ip multicast summary
0 MMLS entries using 0 bytes of memory
Number of partial hardware-switched flows: 0
Number of complete hardware-switched flows: 0

Directly connected subnet entry install is enabled
Hardware shortcuts for mvpn mroutes supported
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Current mode of replication is Egress
Auto-detection of replication mode is enabled
Consistency checker is enabled
Bidir gm-scan-interval: 10



R2#sh mls ip multicast statistics
MLS Multicast configuration and state:
Counters last cleared                         Never
Router Mac                                    001b.0de6.7b80
MLS multicast operating state                 ACTIVE
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

SCB RetryQ size                               1
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Layer 3 Switching H/W Version                PFC III
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Maximum number of allowed outstanding message 20
Maximum size reached from feQ                 0
Maximum size reached from screq               2
Feature Notification sent (simple/rtr-mac)    1/2
Feature Notification Ack received             3
Unsolicited Feature Notification received     0
MSM sent/Received                             0/0
Delete notifications received                 0
sgc oif delete notifications received         0
Flow Statistics messages received             0
Restart Notification messages received        0
Cleanup Send/Resp-rx seq number               0/0



couple of years ago, because of the number of mac-addresses or unicast
routes (depends on SDM profile), at a certain moment, 3750-ME did all
the routing in software (CEF disable) and "ip cef distributed"
reactivate CEF and hardware routing. 

looks the same to me on 7600 with multicast but in this case i don't
know what is the trigger. i'll keep digging.

--
liviu.

On Thu, 2008-08-21 at 09:45 +0100, Phil Mayers wrote:

> Something is configured wrong somewhere. Can you supply the output of:
> 
> sh run | inc ^ip (pim|multi)
> sh run int Vlan10
> sh ip pim int
> sh ip pim nei
> sh ip rpf SOURCEIP
> 
> ...on both R1 and R2?
> 
> It sounds to me as if R2 isn't correctly configured as an RP, or isn't 
> able to send the PIM joins which would push things down into hardware - 
> the initial packets to the RP are always in software, so that's normal, 
> but it's not normal for the traffic to be continually CPU-punted.
>