zebra: Do not flush an existing vni configuration trying to remove wrong vni

[ton31337]

Closes #17952

[piotrsuchy]

Hey @ton31337, thanks for this PR this does indeed fix the issue I described, however there is one edge case:

root@bd6e65241d7a:/# vtysh

Hello, this is FRRouting (version 10.4-dev).
Copyright 1996-2005 Kunihiro Ishiguro, et al.

bd6e65241d7a# configure
bd6e65241d7a(config)# vni 1
bd6e65241d7a(config)# sh run
% Unknown command: sh run
bd6e65241d7a(config)# do sh run
Building configuration...

Current configuration:
!
frr version 10.4-dev
frr defaults traditional
hostname bd6e65241d7a
log syslog informational
vni 1
no ipv6 forwarding
service integrated-vtysh-config
!
end
bd6e65241d7a(config)# no vni 2
bd6e65241d7a(config)# do sh run
Building configuration...

Current configuration:
!
frr version 10.4-dev
frr defaults traditional
hostname bd6e65241d7a
log syslog informational
vni 1
no ipv6 forwarding
service integrated-vtysh-config
!
end
#### EVERYTHING CORRECT, THIS PATCH FIXES THE ABOVE SCENARIO, HOWEVER:

bd6e65241d7a(config)# vrf vrf-test
bd6e65241d7a(config-vrf)# vni 2
bd6e65241d7a(config-vrf)# exit
bd6e65241d7a(config)# do show run
Building configuration...

Current configuration:
!
frr version 10.4-dev
frr defaults traditional
hostname bd6e65241d7a
log syslog informational
vni 1
no ipv6 forwarding
service integrated-vtysh-config
!
vrf vrf-test
 vni 2
exit-vrf
!
end
bd6e65241d7a(config)# no vni 2
bd6e65241d7a(config)# do show run
Building configuration...

Current configuration:
!
frr version 10.4-dev
frr defaults traditional
hostname bd6e65241d7a
log syslog informational
no ipv6 forwarding
service integrated-vtysh-config
!
vrf vrf-test
 vni 2
exit-vrf
!
end

If there is a vrf with that VNI it will still delete the default one, without checking if that's the one specified, so it's like a false positive. We say 'delete 2', it deletes the default one, so '1' in this case, because there is a '2' in the vrf. (the value has to match the value in the VRF). Do you think there is a way to guard against that as well? Because that seems to be our case (in an environment where there are a lot of vrfs).

[ton31337]

@piotrsuchy could you retest with the latest changes?

[piotrsuchy]

It does work exactly how I thought it should now! Thanks!

[donaldsharp]

@mergifiyio backport dev/10.3 stable/10.0 stable/10.1