CEF Load-Sharing and EIGRP UCLB

[jiri1]

Hi! I have a question about CEF load-sharing buckets with EIGRP UCLB.

How does the load-sharing work with a very small ratio?

It all works as it is supposed to but I do not understand how the buckets are distributed since there is only 16 of them... ?

Or how is it working in general since there is no bucket for G0/0 or am I reading the output wrongly?

EXAMPLE:

R1#sh ip cef 1.1.1.1 in
1.1.1.1/32, epoch 0, RIB[I], refcnt 5, per-packet sharing
  sources: RIB 
  feature space:
    IPRM: 0x00028000
  ifnums:
    GigabitEthernet0/0(2): 10.0.0.2
    GigabitEthernet0/1(3): 10.0.1.3
    GigabitEthernet0/2(4): 10.0.2.4
  path list 0FA2CC94, 3 locks, per-packet, flags 0x49 [shble, rif, hwcn]
    path 1139E584, share 4/4, type attached nexthop, for IPv4
      nexthop 10.0.0.2 GigabitEthernet0/0, IP adj out of GigabitEthernet0/0, addr 10.0.0.2 1029CDD8
    path 1139E5EC, share 67/67, type attached nexthop, for IPv4
      nexthop 10.0.1.3 GigabitEthernet0/1, IP adj out of GigabitEthernet0/1, addr 10.0.1.3 1029CCA8
    path 1139E654, share 16/80, type attached nexthop, for IPv4
      nexthop 10.0.2.4 GigabitEthernet0/2, IP adj out of GigabitEthernet0/2, addr 10.0.2.4 116C7518
  output chain:
    loadinfo 1139F008, per-packet, 2 choices, flags 0002, 5 locks
      flags [for-rx-IPv4]
      16 hash buckets
        < 0*> IP adj out of GigabitEthernet0/1, addr 10.0.1.3 1029CCA8
        < 1 > IP adj out of GigabitEthernet0/2, addr 10.0.2.4 116C7518
        < 2 > IP adj out of GigabitEthernet0/1, addr 10.0.1.3 1029CCA8
        < 3 > IP adj out of GigabitEthernet0/2, addr 10.0.2.4 116C7518
        < 4 > IP adj out of GigabitEthernet0/1, addr 10.0.1.3 1029CCA8
        < 5 > IP adj out of GigabitEthernet0/2, addr 10.0.2.4 116C7518
        < 6 > IP adj out of GigabitEthernet0/1, addr 10.0.1.3 1029CCA8
        < 7 > IP adj out of GigabitEthernet0/2, addr 10.0.2.4 116C7518
        < 8 > IP adj out of GigabitEthernet0/1, addr 10.0.1.3 1029CCA8
        < 9 > IP adj out of GigabitEthernet0/2, addr 10.0.2.4 116C7518
        <10 > IP adj out of GigabitEthernet0/1, addr 10.0.1.3 1029CCA8
        <11 > IP adj out of GigabitEthernet0/2, addr 10.0.2.4 116C7518
        <12 > IP adj out of GigabitEthernet0/1, addr 10.0.1.3 1029CCA8
        <13 > IP adj out of GigabitEthernet0/2, addr 10.0.2.4 116C7518
        <14 > IP adj out of GigabitEthernet0/1, addr 10.0.1.3 1029CCA8
        <15 > IP adj out of GigabitEthernet0/2, addr 10.0.2.4 116C7518
      Subblocks:
        None
R1#

Thanks in prior for any help.

[MHM Cisco World]

are you set the max path to 3 or to 2 ?

[jiri1]

4

R1#sh ip pro
*Jul 17 19:38:27.581: %SYS-5-CONFIG_I: Configured from console by consoletocols
*** IP Routing is NSF aware ***

Routing Protocol is "application"
  Sending updates every 0 seconds
  Invalid after 0 seconds, hold down 0, flushed after 0
  Outgoing update filter list for all interfaces is not set
  Incoming update filter list for all interfaces is not set
  Maximum path: 32
  Routing for Networks:
  Routing Information Sources:
    Gateway         Distance      Last Update
  Distance: (default is 4)

Routing Protocol is "eigrp 1"
  Outgoing update filter list for all interfaces is not set
  Incoming update filter list for all interfaces is not set
  Default networks flagged in outgoing updates
  Default networks accepted from incoming updates
  EIGRP-IPv4 Protocol for AS(1)
    Metric weight K1=1, K2=0, K3=1, K4=0, K5=0
    Soft SIA disabled
    NSF-aware route hold timer is 240
    Router-ID: 10.0.2.1
    Topology : 0 (base) 
      Active Timer: 3 min
      Distance: internal 90 external 170
      Maximum path: 4
      Maximum hopcount 100
      Maximum metric variance 21
          
  Automatic Summarization: disabled
  Maximum path: 4
  Routing for Networks:
    0.0.0.0
  Routing Information Sources:
    Gateway         Distance      Last Update
    10.0.1.3              90      00:00:11
    10.0.0.2              90      00:00:11
    10.0.2.4              90      00:00:13
  Distance: internal 90 external 170
          

[MHM Cisco World]

the three path have same metric or different ? if different what is value for each one?

[jiri1]

P 1.1.1.1/32, 3 successors, FD is 131072
        via 10.0.0.2 (2690816/130816), GigabitEthernet0/0
        via 10.0.1.3 (156416/130816), GigabitEthernet0/1
        via 10.0.2.4 (131072/130816), GigabitEthernet0/2

Which corresponds to already shown ratio (4:67:80).

[MHM Cisco World]

and the variance is "21"?

[jiri1]

So I can use all the paths, is there any problem with it?

[Giuseppe Larosa]

Hello @jiri1 ,

you can use all of them because the Reported Distance is lower then the Feasible Distance for all three paths. All of them are feasible successors.

The variance value of 21 allows to include the first path via Gi0/0 but the high metric makes this path much less used then the others.

21* FD  > 2690816  

21 * 131072 > 2690816

but the CEF buckets do not report any bucket for this next-hop so UC load balancing in EIGRP uses something else.

Hope to help

Giuseppe

[jiri1]

Thanks, that is what i thought, what else does it use?

Is there any other function of CEF that I am not familiar with or is the process switching in use?

[Giuseppe Larosa · Accepted solution]

Hello @jiri1 ,

for sure it is not process switching or UC load balancing would be more a problem then a feature (it would cause very high CPU load)

What I mean is that the UC EIGRP LB must use some hidden data structure that can be seen as similar to the CEF table but with more buckets.

This is my guess.

Hope to help

Giuseppe

[jiri1]

Ok, so, since EIGRP UCLB is a proprietary function there is no way of knowing for sure.

Thanks for help, I think this will be enough for me I appreciate your help.

[MHM Cisco World]

...

[MHM Cisco World]

I am so interested in this case, so can you confirm that all three interface is run CEF, 

i think like this because CEF select only two, so can you check if all three interface has CEF?

[jiri1]

Indeed, they are all running CEF.