Re: [PATCH pve-network] sdn: frr: add bgpd to SDN-managed daemons

[Hannes Laimer <h.laimer@proxmox.com>]

On 2026-05-19 10:54, Gabriel Goller wrote:
> On 19.05.2026 10:17, Hannes Laimer wrote:
>> Every sdn-apply on a node that has a BGP fabric configured trips the
>> set_daemon_status() guard with "bgpd is not SDN managed", aborting FRR
>> config generation. The Rust enabled_daemons() reports bgpd for BGP
>> fabrics, but the Perl allowlist was only updated for OSPF and OpenFabric
>> when those fabric types were originally added.
>>
>> Fixes: 9d8533d ("sdn: fabrics: register bgp as a fabric protocol type")
>> Signed-off-by: Hannes Laimer <h.laimer@proxmox.com>
>> ---
>>  src/PVE/Network/SDN/Frr.pm | 1 +
>>  1 file changed, 1 insertion(+)
>>
>> diff --git a/src/PVE/Network/SDN/Frr.pm b/src/PVE/Network/SDN/Frr.pm
>> index 7c60d28..642610c 100644
>> --- a/src/PVE/Network/SDN/Frr.pm
>> +++ b/src/PVE/Network/SDN/Frr.pm
>> @@ -125,6 +125,7 @@ sub restart {
>>  my $SDN_DAEMONS_DEFAULT = {
>>      ospfd => 0,
>>      fabricd => 0,
>> +    bgpd => 0,
>>  };
>>  
>>  =head3 set_daemon_status(\%daemons, $set_default)
>> -- 
>> 2.47.3
> 
> Hmm maybe I'm misremembering how this works, but shouldn't we set this to 1, so
> that when removing a bgp fabric, the bgpd daemon stays enabled? Because the
> legacy bgp controller and evpn controllers don't do the daemon check afaik and
> just rely on the fact that it's always enabled?

right.. good catch! this should be `1`

will send a v2