[SOLVED] SDN/VXLAN state "error" caused by ipv6 RA in vnet

[ferrets]

I've successfully installed SDN on my two Proxmox node and make it working. But I found that the SDN status on the GUI will becomes "avaliable" after I reload the network config for only a very short priorid of time, some time later, it becomes "error".

The vm running on that vxlan is still be able to connect with each other, and I don't know where to find the error log and go further to debug.

 

[spirit]

Hi,
the monitoring is using "ifquery -a -c" command to show status of networking interfaces.

can you send result of this command when you have an error ? (This is strange that it's becomes "error" by itself after some time, without any action)

 

[ferrets]

Hi,
the monitoring is using "ifquery -a -c" command to show status of networking interfaces.

can you send result of this command when you have an error ? (This is strange that it's becomes "error" by itself after some time, without any action)

you nail it!

looks like it's some things wrong with ipv6 address.

but it's still weird that ipv6 is working but report failed.

 

[spirit]

mmm...

what is the output of :

# ifreload -a -d

and

#ip addr

I wonder if it couldn't be related to ipv6 format (maybe short or long format), when ifquery is doing the diff.

 

[spirit]

also, does the ipv6 address displayed in ifquery have the same format than values in /etc/network/interfaces.d/sdn ?

(also Server1 bridge seem to have the same ip display 3x, this is strange)

 

[ferrets]

also, does the ipv6 address displayed in ifquery have the same format than values in /etc/network/interfaces.d/sdn ?

(also Server1 bridge seem to have the same ip display 3x, this is strange)

there is no ipv6 address configured in /etc/network/interfaces.d/sdn.

the ipv6 address is assigned because theres is a vm running RouterOS sending RA, so maybe that's the problem.

(btw, the 3 ipv6 addresses have different prefix that comes from my ISP that changes for a few times, sorry for the confusing)

 

[spirit]

there is no ipv6 address configured in /etc/network/interfaces.d/sdn.

the ipv6 address is assigned because theres is a vm running RouterOS sending RA, so maybe that's the problem.

oh, ok !
seem to be a bug in ifquery, I'll try to see if it's possible to fix this. (I'm not sure that is possible to see if the ip has been configured by RA ou manually)

thanks for the report !

Edit: I have open a bug at ifupdown2 github: https://github.com/CumulusNetworks/ifupdown2/issues/200

 

[spirit]

I'll try improve code to disable RA on vnets, we really don't need it.