Hi Zafar,
My fault. Let me repeat if I understand correctly. Host1 –– iPE –– P — ePE –– Host2 You mean iPE should handle the error upon receiving traffic from host1. Like TTL/MTU etc. The ICMP originated by P node is not the case you want. Is that correct? BR, Feng 发件人: Zafar Ali (zali) <[email protected]> 发送时间: 2026年4月2日 13:13 收件人: [email protected]; 'Zafar Ali (zali)' <[email protected]> 抄送: 'SPRING WG' <[email protected]>; [email protected]; Zafar Ali (zali) <[email protected]> 主题: [spring] Re: Follow up "SID as source" in CE-CE ICMP case(uniform mode) Hi Feng What I meant is that you need to explain in your draft how the source node (which uses a VPN SID as the source address) handles the ICMP error encountered by the data traffic (e.g., MTU exceeded). Given the SID pseudo codes covers the upper layer processing, I think it should be fine. But please document it in your draft. Thanks Regards … Zafar From: [email protected] <mailto:[email protected]> <[email protected] <mailto:[email protected]> > Date: Wednesday, April 1, 2026 at 11:46 PM To: 'Zafar Ali (zali)' <[email protected] <mailto:[email protected]> > Cc: 'SPRING WG' <[email protected] <mailto:[email protected]> >, [email protected] <mailto:[email protected]> <[email protected] <mailto:[email protected]> >, Zafar Ali (zali) <[email protected] <mailto:[email protected]> > Subject: 回复: [spring] Re: Follow up "SID as source" in CE-CE ICMP case(uniform mode) Hi Zafar, Thanks for your response. In my opinion, the SRv6 compatible with firewall is orthogonal with the ICMP handling. The source SRv6 node will receive ICMP (not encaped by SRv6 header) if the TTL gets expired in the network. There are 2 relevant cases in my head. * Case1: trace route from CE in uniform mode. I did not give proposal. * Case2: Srv6 ping on PE. I gave proposal using VPN SID. If you want have some proposals in the draft. I would say there are 2 existing proposals regarding ICMP handling. Proposal 1 is P node set the VPN SID as ICMP destination address (draft-varhal-6man-icmp-srv6-vpn). Proposal 2 is MPLS style, which forwarding to egress PE(draft-ali-6man-srv6-vpn-icmp-error-handling). Since my draft is to address the firewall issue, I would say handling of ICMP should consider firewall case. To be more clear, the expected ICMP address pair would be a pair of VPN SID. Thus it will not only matter of ingress node, but will also relevant with P and egress PE. BR, Feng 发件人: Zafar Ali (zali) <[email protected] <mailto:[email protected]> > 发送时间: 2026年4月2日 8:51 收件人: [email protected] 抄送: 'SPRING WG' <[email protected] <mailto:[email protected]> >; [email protected] <mailto:[email protected]> ; Zafar Ali (zali) <[email protected] <mailto:[email protected]> > 主题: [spring] Re: Follow up "SID as source" in CE-CE ICMP case(uniform mode) Hi Feng My comments have nothing to do with the CE-CE trace route in uniform mode proposal in 6man. You need to explain how the source node processes the ICMP error if it receives it from the network. You cannot exclude ICMP error handling from your proposal. Hope it helps. Thanks Regards … Zafar From: <mailto:[email protected]> [email protected] < <mailto:[email protected]> [email protected]> Date: Thursday, March 26, 2026 at 2:18 AM To: Zafar Ali (zali) < <mailto:[email protected]> [email protected]> Cc: 'SPRING WG' < <mailto:[email protected]> [email protected]>, <mailto:[email protected]> [email protected] < <mailto:[email protected]> [email protected]> Subject: Follow up "SID as source" in CE-CE ICMP case(uniform mode) Hi Zafar, Thanks for your comments on SPRING meeting last week. You mentioned the CE-CE ICMP issue with uniform mode in provider network, and there are some proposals in 6man. Here comes my proposal: I would exclude that scenario in the scope of “draft-yang-spring-sid-as-source-address”. Will that be an acceptable solution for you? BR, Feng Yang China Mobile Research Institute
_______________________________________________ spring mailing list -- [email protected] To unsubscribe send an email to [email protected]
