search

open-switch / opx-nas-l3

https://openswitch.net
1 stars 9 forks source link

Unable to ping native loopback interface 'lo' #29

Closed geraldmyres closed 5 years ago

geraldmyres commented 6 years ago

Unable to ping Leaf or Spine loopback 'lo' interfaces from hosts attached to Leafs.
Environment: OPX 3.0.0 FRR 5.01 eBGP or OSPF

  1. Configure Leaf1 and Spine1 switches. See attached configs and topology.
    1. Ping Spine loopback 10.0.1.1 from Leaf1 or hosts attached to leaf1. Pings fail.
    2. From OPX, create new loopback 'lo1' on both leaf1 and spine1 ip link add name lo1 type dummy
    3. From FRR, assign IP address to lo1 on both leaf1 and spine1.
    4. Pings now work.
    5. If switch rebooted, lose new loopback 'lo1' .

Os there a way to make loopback lo1 persistent, or to resolve loopback 'lo'?

bgp Leaf1 BGP.txt Spine2 BGP.txt

geraldmyres commented 6 years ago

I think this is a bug

madhu222 commented 5 years ago

Workaround for this issue is to do the networking restart couple of times.

  1. I did some modification on the interface configuration file. Below is the file content I changed in Spine1.

    
auto e101-001-1
iface e101-001-1 inet manual
mtu 9184

auto e101-002-1 iface e101-002-1 inet manual mtu 9184

auto lo iface lo inet manual pre-up sudo /bin/ip link add name lo1 type dummy

auto lo1 iface lo1 inet static address 11.0.1.1 netmask 255.255.255.255


 Networking restart needs to be done one more time to send the netlink message. I changed the /usr/bin/network_restart.sh script .

root@Spine1:~# cat /usr/bin/network_restart.sh
```#!/bin/sh

service networking restart
sleep 5
service networking restart
GarrickHe commented 5 years ago

This should fix this issue: https://github.com/open-switch/opx-nas-interface/pull/67