Junos
Highlighted
Junos

MC-LAG Active/Standby causing up to 5 seconds of packet loss

‎09-29-2017 07:41 AM

I have two MX nodes in an MC-LAG active/standby.

 

the downstream client is an ex3400:

 

we have the following config on Both MXs ( the same apart from the status-control line )

vlan-tagging;
encapsulation flexible-ethernet-services;
mac 00:00:5e:00:53:55;
aggregated-ether-options {
    minimum-links 1;
    link-speed 1g;
    lacp {
        active;
        periodic fast;
        system-id 00:00:00:00:00:01;
        admin-key 1;
    }
    mc-ae {
        mc-ae-id 10;
        redundancy-group 1;
        chassis-id 0;
        mode active-standby;
        status-control standby;
    }
}
unit 200 {
    vlan-id 200;
    family inet {
        address 10.160.1.38/29;
    }
}

on the downstream client we have:

aggregated-ether-options {
    minimum-links 1;
    link-speed 1g;
    lacp {
        active;
        periodic fast;
    }
}
unit 0 {
    family ethernet-switching {
        interface-mode trunk;
        vlan members 200;
  }
}

interface irb unit 200 family inet address 10.160.1.34/29;

vlan 200 vlan-id 200 l3-interface irb.200;

when we disconnect the active link between the MX and the EX we are getting a number of seconds of packet loss.

I would have though this would have been substantially faster.

 

any pointers?

 

thanks

 

 

5 REPLIES 5
Junos

Re: MC-LAG Active/Standby causing up to 5 seconds of packet loss

‎09-29-2017 08:50 AM
Junos

Re: MC-LAG Active/Standby causing up to 5 seconds of packet loss

‎09-30-2017 04:48 AM

The enhanced-converge option is ONLY for active/active setups.

 

We went for the active/standby option as we want to have multiple downstream clients with the same configuration.

 

they all also use the same vlan ID but with a seperate IP subnet per client device.

 

if we used active/active, we needed to use:

 

  • vlan rewrites on the client facing interfaces
  • vlans on the MX so that we could use IRB interfaces
  • stitch the bridge-domain to the client interface and also the ICL interface.

this is substantially more configuration and complexity than the active/standby option.

 

everything is working on the active/standby setup, but we want to improve the convergence if possible.

Junos

Re: MC-LAG Active/Standby causing up to 5 seconds of packet loss

‎09-30-2017 05:59 AM

If you are running at least 15.1 on the MX and 13.2 on the EX you can add backup-liveness-detection to shorten failover timing.

 

Steve Puluka BSEET - Juniper Ambassador
IP Architect - DQE Communications Pittsburgh, PA (Metro Ethernet & ISP)
http://puluka.com/home
Junos

Re: MC-LAG Active/Standby causing up to 5 seconds of packet loss

‎09-30-2017 04:30 PM

Hi Steve

 

are you talking here about running BFD between the downstream client and the MC-LAG Peers.

 

we already have BFD setup for our ICCP link.

What I am failing over is the client facing MC-AE interface, the ICCP link is rock solid.

 

 

also our final end device is a MultiService Access node to provide residential broadband services.  we are currently in a testing phase but dont seem to be getting the results I had expected.

 

If this doesnt work out as JTAC seem to be a waste of space, I am going to try ACTIVE/ACTIVE with EVPN.

Junos

Re: MC-LAG Active/Standby causing up to 5 seconds of packet loss

‎10-01-2017 02:06 PM

Sorry, I thought I included the link but guess I forgot to paste this in.  This is a BFD specifically for link failure detection in ICCP to enhance failover timings.

 

https://www.juniper.net/documentation/en_US/junos/topics/reference/configuration-statement/backup-li...

Steve Puluka BSEET - Juniper Ambassador
IP Architect - DQE Communications Pittsburgh, PA (Metro Ethernet & ISP)
http://puluka.com/home