Problem with wan load-balancing and dhcp with /32 address

Hello,

I am trying to use wan-failover over two connections: FTTH and LTE, both via DHCP.

The problem I have is with the FTTH connection, where I get a /32 address (e.g., address X.a.b.c/32 with gateway Y.e.f.g.0) and wan_lb complains with:

wan_lb[680038]: failure to insert default route on active path with this command: ip route replace table 201 default dev eth0 via Y.e.f.g.0

If I execute the command manually, the error message is:

Error: Nexthop has invalid gateway

The command succeeds if I add the onlink option.

Here is my configuration:

show load-balancing wan

 flush-connections
 interface-health eth0 {
     failure-count 3
     nexthop dhcp
     success-count 5
     test 10 {
         resp-time 5
         target 1.1.1.1
         ttl-limit 1
         type ping
     }
 }
 interface-health eth1.11 {
     failure-count 3
     nexthop dhcp
     success-count 5
     test 10 {
         resp-time 5
         target 1.0.0.1
         ttl-limit 1
         type ping
     }
 }
 rule 10 {
     failover
     inbound-interface eth1.10
     interface eth0 {
         weight 100
     }
     interface eth1.11 {
         weight 1
     }
     protocol all
 }

eth0 is the concerned FTTH connection, eth1.11 is the LTE connection and eth1.10 is my LAN.

I have tried to manually set the gateway address in nexthop, but the result is the same.

At the moment, the only solution I have found is to replace nexthop dhcp with nexthop <eth0 address>.

I would like to ask:

  • Is that the result of a bug in wan_lb not properly handling the /32 address and the corresponding gateway?
  • Is there any drawback in using the eth0 IP address for nexthop?
  • What should nexthop be exactly?
  • Am I doing something wrong?

I am using version 1.4-rolling-202304290647.

Thanks in advance and, please, let me know if I can provide any additional information.

this version is not up to date, many fixed was done during this time . try to update a newer rolling-release :

https://vyos.net/get/nightly-builds/

Thanks for your reply.

I have tried to update to the last rolling release (vyos-1.5-rolling-202311030022-amd64.iso) but the update process fails and the only error I can see is about a failure in migrating my configuration. Then I am not even able to login to vyos.

Is there a way to verify my configuration and spot any possible migration problem before upgrading?

Additionally, I think the code “replacing” the route is this one:

and has been updated for the last time 8 months ago.

Grepping for ip route replace table, it looks like no other options are passed with respect to what I observe in my case. Hence it may be probable for the problem I have to persist in the last rolling releases too.

Of course, I am not a vyos developer and I can only guess.

it looks as a bug in the migration-script , Could you share full configuration ?or create a case with this incident in our https://vyos.dev/

Here is my full config.

I cannot create a case because my account is still pending approval.

vyos.config (29.6 KB)

My account has been approved. Do you want me to open a case or having access to my configuration file is enough?

if you can create the case , it would be better . Please , try to explain who to reproduce the bus and versions running.

I have opened a case for the error in migrating the configuration: ⚓ T5721 Error in migrating configuration from VyOS 1.4.