VxLAN fix tunnel oper_state issue

[praveenraja1]

What I did

In portsorch layer ensure tunnel port_oper state is updated correctly after tunnel comes up.

Additionally made changes so that

1. non L3 packets which gets encapsulated have outer IP hdr TTL as 64.
2. For L3 packets the inner IpHdr TTL is preserved in outer header.

Why I did it

After tunnel was up and traffic was flowing, tunnel oper_state was displayed as down.

root@sonic:~#  show vxlan remotevtep
+---------+---------+-------------------+--------------+
| SIP     | DIP     | Creation Source   | OperStatus   |
+=========+=========+===================+==============+
| 1.1.1.1 | 2.2.2.2 | EVPN              | oper_down    |
+---------+---------+-------------------+--------------+
Total count : 1

How I verified it

root@sonic:~# show vxlan tunnel
vxlan tunnel name    source ip    destination ip    tunnel map name    tunnel map mapping(vni -> vlan)
-------------------  -----------  ----------------  -----------------  ---------------------------------
vtep                 2.2.2.2                        map_2000_Vlan20    2000 -> Vlan20
                                                    map_3000_Vlan30    3000 -> Vlan30
root@sonic:~#
root@sonic:~#
root@sonic:~#  show vxlan remotevtep
+---------+---------+-------------------+--------------+
| SIP     | DIP     | Creation Source   | OperStatus   |
+=========+=========+===================+==============+
| 2.2.2.2 | 1.1.1.1 | EVPN              | oper_up      |
+---------+---------+-------------------+--------------+
Total count : 1

Details if related

[linux-foundation-easycla]

• ✅ login: praveenraja1 (94c3fa1)
• ❌ The email address for the commit (8f62c2c) is not linked to the GitHub account, preventing the EasyCLA check. Consult this Help Article and GitHub Help to resolve. (To view the commit's email address, add .patch at the end of this PR page's URL.) For further assistance with EasyCLA, please submit a support request ticket.

[VladimirKuk]

I think it would better to use DEFAULT_TUNNEL_ENCAP_TTL as default value for the encap_ttl (instead of 0). It will reuse the existing code and you could also remove the "if (encap_ttl != 0)".

[VladimirKuk]

I think only calling for updateDbPortOperStatus is relevant for tunnel case, everything else (like autoneg, link training, ...) is physical port specific.
Maybe moving updateDbPortOperStatus outside of this "if", should be more correct ?

[bradh352]

Looks like this supersedes #2080 as it includes changes from that, and then also fixes sonic-net/sonic-buildimage#10050

I'd think this really needs to get merged for the next release. VXLAN just doesn't work (at least on my Dell S5248F switches) without the TTL change, and the oper_status fix is very useful for debugging.

[bradh352]

@VladimirKuk what do we need to do to get this merged? I see you had asked if updateDbPortOperStatus(port, status) could be moved outside the if, implying just always call it. I can test to see if that works today if you want.

I can say that this patch is critical to VXLAN support for me, as soon as I apply it, I can ping across the VXLAN tunnels to addresses learned via ARP by the VTEPs. Without it, just nothing...

[bradh352]

@prsunny I see you merge a lot of PRs, I'd really like to see this in the next SONiC stable release branch... it would mean I wouldn't have to continue maintaining my own automated build system and patches. Thanks!

[bradh352]

I'm replacing this PR with #3383

[VladimirKuk]

@VladimirKuk what do we need to do to get this merged? I see you had asked if updateDbPortOperStatus(port, status) could be moved outside the if, implying just always call it. I can test to see if that works today if you want.

I can say that this patch is critical to VXLAN support for me, as soon as I apply it, I can ping across the VXLAN tunnels to addresses learned via ARP by the VTEPs. Without it, just nothing...

Makes sense, since updateDbPortOperStatus wasn't called for tunnel case. Looks good to me.

[bradh352]

@VladimirKuk can you look at my modifications in #3383 to verify it looks right and comment there?

[prsunny]

Closing in favor of #3383