Who is Active in MCLAG setup?

[steven.crutchley]

Hi all,

I'm trying to understand which router becomes the Active router in an ICCP MC-LAG setup.
I've read this post and am struggling to get the basics.

If I have this configured:

PE1
=====
redundancy
 iccp
  group 1
   mlacp node 1
   mlacp system mac 0000.1111.2222
   mlacp system priority 1
   member
    neighbor 2.2.2.2

PE2
======
redundancy
 iccp
  group 1
   mlacp node 2
   mlacp system mac 0000.1111.2222
   mlacp system priority 1
   member
    neighbor 1.1.1.1

(full ICCP details not shown)

Who is the active POA for the MC-LAG group? 
Sometimes is it PE1, sometimes it is PE2.

RP/0/RP0/CPU0:PE1#sh bundle be1        
Thu Nov  7 12:53:19.852 UTC

Bundle-Ether1
  Status:                                    Up
 <snip>

  Port                  Device           State        Port ID         B/W, kbps
  --------------------  ---------------  -----------  --------------  ----------
  FH0/0/0/1             Local            Active       0x8000, 0x9001   400000000
      Link is Active
  FH0/0/0/1             2.2.2.2          Standby      0x8000, 0xa001   400000000
      Link is marked as Standby by mLACP peer
RP/0/RP0/CPU0:PE1#

Now the MAC and priority should match (it'll technically work their priorities are difference, since one is arbitrarily selected by that is beside the point here). So the only differentiating factor is node number. PE1 is node 1. PE2 is node 2. But I've had instances where both are Active regardless of this. 

I'm even tweaked the port priorities - thinking that the numerically lower port priorities would want to become active and thus prompt their connected PE to take over as the active. But nothing. 

This becomes crucially important when understanding the various switchover modes (brute-force vs dynamic etc). I've searched the link I posted above, and other documentation, and cant seem to find any correlation. 

Can anyone assist?

[emre]

hi,

We use mlacp in asr99k's, we have the same configuration, there are some special configurations, but there is no problem in its configuration as a standard base, Mac can see it and mlacp sits.

normally mlacp works stable in asr9k series, your problem may be caused by XR-os or bug in os software, if you want to upgrade, read the release notes with the recomented version and

also what is the device on the other side, Huawei, arista etc.
because cisco products do not work stabile with huawei, you need to make static on the e-trunk side

[steven.crutchley]

Hi there,

There is no problem with the MC-LAG setup. 

I am asking how to determine which is the active. 

[emre]

mlacp system priority 1

You can set these priority values by changing
In both nodes

[steven.crutchley]

Yes but as my post explains, these need to match in an MC-LAG scenario and they do no dictate which is Active. 

I am asking what makes the MC-LAG peer Active or Standby.

[emre]

In addition, MC-LAG only looks at the physical redundancy in its structure, that is, it does not check the redundancy of the L2VPN, L3VPN and carrier Ethernet services you run, you need to back them up yourself,

My friend,
MC-LAG is a redundancy structure related to the redundancy of the physical port,
you can specify the priority in this structure by changing the priority values,
''mlacp system priority 1'', that is, the value here

Apart from this, there is no config or command to set the priority.!!

If we come to the following problem, it automatically fails over when there is a physical or hardware failure, and when there is a problem with the fiber cables or cards connected to the ports, it automatically fails over when there is an interruption.

If there is a problem with the port, it does not completely failover the device, it just moves the port to another, but if the device burns out and goes down, it automatically transfers all traffic to the other one.

''I am asking what makes the MC-LAG peer Active or Standby.''

Also, you do not need to configure each device separately. The configuration you make to the primary device will automatically synchronize to the other device.

It is not a very complex configuration, the working logic is very simple, you can also make nV configuration with different switches to increase capacity.

think simple  

[steven.crutchley]

Sorry. Once again. My question is about the election process that determines with of the two POAs becomes the active one. What determines which POA is active?

It cannot be the `mlacp system priority x` command - I have tested this and it does not make a difference. 

This post explicitly states that the priorities should match. The priority is what the MC-LAG POA pair collectively present to the DHD as a LACP System ID (in addition to the MAC address). If the priorities don't match, one of the two is selected by an arbitration process. 

Simply put, I'm asking that if two PE devices, PE1 and PE2 are put into a MC-LAG group using the configuration I showed above, what one will become the Active POA, as show in the following output?

RP/0/RP0/CPU0:PE1#sh bundle be1        
Thu Nov  7 12:53:19.852 UTC

Bundle-Ether1
  Status:                                    Up
 <snip>

  Port                  Device           State        Port ID         B/W, kbps
  --------------------  ---------------  -----------  --------------  ----------
  FH0/0/0/1             Local            Active       0x8000, 0x9001   400000000   <<<<<< why is this Active and not 2.2.2.2?
      Link is Active
  FH0/0/0/1             2.2.2.2          Standby      0x8000, 0xa001   400000000
      Link is marked as Standby by mLACP peer
RP/0/RP0/CPU0:PE1#

[emre]

ok,

What are the following values in your bundle interface configuration,
Can you forward your port configurations in both poa
"mlacp lag-priority" and “bundle port-priority"

can you also forward the following outputs?

"show mlacp"