Expand all | Collapse all

EX4300: Traffic drop issues

  • 1.  EX4300: Traffic drop issues

    Posted 10-15-2020 01:42

    We have a Layer 2 customer NNI terminating on one of our switch gig port. Our customer is reporting a high packet loss on circuits terminating on this port on NNI. Our setup has customer A end on Ex2300 switch and customer B end a CPE on end user site which we run vlans to join two ends.

    We then run VLANs from switch to our router.

    example config on VLAN on switch is below@:
    show configuration interfaces ge-0/1/1
    mtu 2000;
    encapsulation extended-vlan-bridge;
    unit 0 {
    vlan-id-list 1-4094;
    input-vlan-map {
    vlan-id 1001;
    output-vlan-map pop;

    Router config is as below@:
    show configuration | display set | match "xe-2/0/1 unit 1001 "
    set interfaces xe-2/0/1 unit 1001 description ABC
    set interfaces xe-2/0/1 unit 1001 vlan-tags outer 1001
    set interfaces xe-2/0/1 unit 1001 vlan-tags inner 719
    set interfaces xe-2/0/1 unit 1001 family inet address

    When pinging customer device on we are seeing high packet loss
    run ping
    PING ( 56 data bytes
    --- ping statistics ---
    500 packets transmitted, 435 packets received, 13% packet loss
    round-trip min/avg/max/stddev = 0.447/1.528/20.511/2.904 ms

    Please can you advise as why this is the case.

  • 2.  Re: EX4300: Traffic drop issues

    Posted 10-15-2020 05:55



    A ping test is not very reliable to test the link health, ICMP traffic is given very low priority on Juniper devices. If the RE is busy processing other control traffic ICMP requests could be discarded. The following document explains this in detail: https://kb.juniper.net/InfoCenter/index?page=content&id=KB28157


    To check the link integrity it is suggested to simulate data plane traffic using a traffic generator.



  • 3.  Re: EX4300: Traffic drop issues

    Posted 10-15-2020 06:07

    I suggest to verify on the other, what is happening with the ICMP echo requests. You can run a continuous ping and run a monitor traffic on interface irb to confirm what happens with the ICMP packets: > monitor traffic interface irb no-resolve size 1500 matching "icmp"


    Also you can check for CRC errors or drops on the interfaces on the Juniper side. To discard any software issues also check for core-dumps and make sure that the Junos version is supported and updated,




  • 4.  Re: EX4300: Traffic drop issues

    Posted 10-15-2020 10:21

    All good suggestions.  You could also configure the interface as a 1-port AE (maybe a default best practice, as then to add additional ports for bandwidth, no hit is taken) and then enable LACP.  LACP would also need to be supported on remote end.  If transit traffic through the switch shows drops, then probably points to some poor link condition.  Then look at interface stats/etc on both ends.

  • 5.  Re: EX4300: Traffic drop issues

    Posted 10-16-2020 03:24


    Your examples (icmp loss) looks to be related to "internal policer"


    Check this:



    I agree with all suggestions, the best will be to use "external" ping SRC/DST.


    As temp solution (if your environment permits) - you can adjust internal policer (for sure at both ends) and check if the ping loss is gone.





    Accept as Solution = cool !
    Accept as Solution+Kudo = You are a Star !