On QFX5100, MC-LAG may fail to work after upgrading from Junos OS 14.1X53 to 17.1R1. When this issue happens, it is likely due to the reason that the "redundancy-group-id-list" option is not configured under the "protocols iccp peer" hierarchy.
After the upgrade, ICCP peers fail to be established with the status below:
{master:0} root@LEAF1> show iccp Redundancy Group Information for peer 192.168.100.1 TCP Connection : In progress <<<<<<<<<<<<<<< Liveliness Detection : Down Backup liveness peer status: Down Client Application: lacpd Client Application: l2ald_iccpd_client
On QFX5100 with Junos OS 14.1X53, redundancy-group-id-list is an optional and hidden command, whereas with Junos OS 17.1R1, it is a mandatory command which needs to be added properly.
redundancy-group-id-list
If similar issue occurs after upgrading to Junos OS 17.1R1, try to add " redundancy-group-id-list " option on both mc-lag peers with the following commands.
{master:0}[edit] user@LEAF0# set protocols iccp peer 192.168.100.2 redundancy-group-id-list 1 {master:0}[edit] user@LEAF2# set protocols iccp peer 192.168.100.1 redundancy-group-id-list 1
Or before the upgrade, add the above commands proactively.
Note: On Junos OS 14.1X53, "redundancy-group-id-list" is a hidden command, so input the entire completely. You may see a warning like below after adding such command on Junos OS 14.1X53 which is expected and does not affect the functionality or upgrade.
root@LEAF2# show protocols iccp local-ip-addr 192.168.100.1; peer 192.168.100.2 { session-establishment-hold-time 100; ## ## Warning: value redundancy-group-id-list ignored: unsupported platform (qfx5100-48s-6q) ## redundancy-group-id-list 1; backup-liveness-detection { backup-peer-ip 172.27.113.75; } liveness-detection { minimum-interval 1000; multiplier 3; } }