I have been playing quite a bit with frr and sdn over the last few days, I was having really weird unreliability with my cpeh thunderbolt mesh network.
I eventually traced it to the the fact that:
I eventually traced it to the the fact that:
- i set the loopback interface IP addresses with frr (instead of a /etc/network/interfaces file)
- when
ifupdown2
,ifreload
, orpvesh set nodes/pve1/network
runs it wipes out the addresses frr set - frr doesn't seem to know the interface changes and won't reapply the interface settings automatically or when a service reload happens (but an frr restart works)
- set
post-up sh -c 'sleep 5 && /usr/bin/systemctl restart frr.service'
in the lo stanza in/etc/network/interfaces
- i originally had a a script in
/etc/network/if-up.d
to bounce the service, but that didn't seem to process for interface lo (it works for interface en05 and en06 i have) - is the expected behaviour / a special conditions for the loopback (my google searches turned up no clear info on this) - i don't really like modifying the
/etc/network/interfaces
file for the loopback so could move this to (i assume) and/etc/networks/interfaces.d/loopback
file - would that have any downsides to proxmox - is there a better / more elegant / more correct way to do this?
Last edited: