After reboot of QFX5100, some of the interfaces that are directly connected to servers may continuously flap. This article suggests a workaround and also recommends upgrading the NIC firmware on the server side to resolve the problem.
Observe the following output:
root@JTAC> show interfaces terse xe-0/0/17 Dec 06 02:21:03 Interface Admin Link Proto Local Remote xe-0/0/17 up down xe-0/0/17.0 up down inet 10.10.10.2/24 {master:0} root@JTAC> show interfaces terse xe-0/0/17 Dec 06 02:21:04 Interface Admin Link Proto Local Remote xe-0/0/17 up down xe-0/0/17.0 up down inet 10.10.10.2/24 {master:0} root@JTAC> show interfaces terse xe-0/0/17 Dec 06 02:21:06 Interface Admin Link Proto Local Remote xe-0/0/17 up down xe-0/0/17.0 up down inet 10.10.10.2/24 {master:0} root@JTAC> show interfaces terse xe-0/0/17 Dec 06 02:21:07 Interface Admin Link Proto Local Remote xe-0/0/17 up up xe-0/0/17.0 up up inet 10.10.10.2/24 {master:0} root@JTAC> show interfaces terse xe-0/0/17 Dec 06 02:21:08 Interface Admin Link Proto Local Remote xe-0/0/17 up down xe-0/0/17.0 up down inet 10.10.10.2/24 {master:0} root@JTAC> show interfaces terse xe-0/0/17 Dec 06 02:21:09 Interface Admin Link Proto Local Remote xe-0/0/17 up down xe-0/0/17.0 up down inet 10.10.10.2/24
In the logs, we can see the link flapping as shown below:
root@JTAC> show log messages | last 100 Dec 06 02:21:25 Dec 6 02:20:16 JTAC mib2d[4770]: SNMP_TRAP_LINK_UP: ifIndex 672, ifAdminStatus up(1), ifOperStatus up(1), ifName xe-0/0/17 Dec 6 02:20:17 JTAC mib2d[4770]: SNMP_TRAP_LINK_DOWN: ifIndex 672, ifAdminStatus up(1), ifOperStatus down(2), ifName xe-0/0/17 Dec 6 02:20:18 JTAC mib2d[4770]: SNMP_TRAP_LINK_UP: ifIndex 672, ifAdminStatus up(1), ifOperStatus up(1), ifName xe-0/0/17 Dec 6 02:20:19 JTAC mib2d[4770]: SNMP_TRAP_LINK_DOWN: ifIndex 672, ifAdminStatus up(1), ifOperStatus down(2), ifName xe-0/0/17 Dec 6 02:20:21 JTAC mib2d[4770]: SNMP_TRAP_LINK_UP: ifIndex 672, ifAdminStatus up(1), ifOperStatus up(1), ifName xe-0/0/17 Dec 6 02:20:22 JTAC mib2d[4770]: SNMP_TRAP_LINK_DOWN: ifIndex 672, ifAdminStatus up(1), ifOperStatus down(2), ifName xe-0/0/17 Dec 6 02:20:24 JTAC mib2d[4770]: SNMP_TRAP_LINK_UP: ifIndex 672, ifAdminStatus up(1), ifOperStatus up(1), ifName xe-0/0/17 Dec 6 02:20:25 JTAC mib2d[4770]: SNMP_TRAP_LINK_DOWN: ifIndex 672, ifAdminStatus up(1), ifOperStatus down(2), ifName xe-0/0/17 Dec 6 02:20:26 JTAC mib2d[4770]: SNMP_TRAP_LINK_UP: ifIndex 672, ifAdminStatus up(1), ifOperStatus up(1), ifName xe-0/0/17
On the server side, we can see the following interface status:
[root@localhost net]# ethtool enp17s0f0 Settings for enp17s0f1: Supported ports: [ FIBRE ] Supported link modes: 1000baseT/Full 1000baseKX/Full 10000baseT/Full Supported pause frame use: Symmetric Supports auto-negotiation: Yes Supported FEC modes: Not reported Advertised link modes: 1000baseT/Full 1000baseKX/Full 10000baseT/Full Advertised pause frame use: No Advertised auto-negotiation: Yes Advertised FEC modes: Not reported Speed: Unknown! Duplex: Unknown! (255) Port: FIBRE PHYAD: 0 Transceiver: external Auto-negotiation: off Supports Wake-on: d Wake-on: d Current message level: 0x00000007 (7) drv probe link Link detected: no <<<<<<<<Link showing down on the server</span>
This issue is related to the Server NIC firmware (INTEL NIC) auto-negotiation stability issue.
To resolve the issue, upgrade the NIC firmware on the server side.
Meanwhile, a workaround is to restart auto-negotiation on the server side by using the following command:
[root@localhost ~]# ethtool -r <server interface>