Re: [OSPF] [spring] One question on E-flag of ABR/ASBR in OSPF SR extension
On 19/05/17 05:40 , Chao Fu wrote: Hi Peter, Yes I think it is better to add the text that would say that ABR MUST NOT set the E-bit. If not, when ABR receives the intra prefix with SID, it would be thought to inherit the SID value including its flags (NP and E) and then flood the SID to other areas. I guess that's why NP is described expressly on ABR not to do so. Then I think E-flag should be clarified also in case the non-directly attached prefix sets it. It could avoid any misunderstanding. Anyway, the E-flag will not be set in this case, right? Thanks. right. Peter Regards, Chao Fu -Original Message- From: Peter Psenak [mailto:ppse...@cisco.com] Sent: Thursday, May 18, 2017 17:47 To: Chao Fu <chao...@ericsson.com>; spr...@ietf.org Subject: Re: [spring] One question on E-flag of ABR/ASBR in OSPF SR extension Hi Chao, On 18/05/17 11:15 , Chao Fu wrote: Hi Peter, It's right if the NP-flag is not set then the received E-flag is ignored. But, if the NP-flag is SET because the prefix is not directly attached to the ABR, E-flag will not be ignored why would an E-bit be set in such case? It's up to the originator of the SID to decide when to set the E-bit. If the ABR does not set the E-bit no PHP would be done and advertised SID will be preserved. Are you trying to add the text that would say that ABR MUST NOT set the E-bit either? I don't think it's necessary. thanks, Peter and the upstream neighbor will replace the Prefix-SID with the Explicit-NULL label 0. I guess actually the packet should be forwarded with the original prefix-SID, and no need to pop the 0 label and look up path again. Regards, Chao Fu -Original Message- From: Peter Psenak [mailto:ppse...@cisco.com] Sent: Thursday, May 18, 2017 16:54 To: Chao Fu <mailto:chao...@ericsson.com>; mailto:spr...@ietf.org Subject: Re: [spring] One question on E-flag of ABR/ASBR in OSPF SR extension Hi Chao, On 18/05/17 09:44 , Chao Fu wrote: Hi, Should we clarify how to set E-flag for ABR/ASBR in OSPF SR extension? In https://www.ietf.org/id/draft-ietf-ospf-segment-routing-extensions-14. txt, the draft describes how to set NP-flag on ABR and ASBR (Section 5 [Page 14]): The NP-Flag (No-PHP) MUST be set for Prefix-SIDs allocated to inter- area prefixes that are originated by the ABR based on intra-area or inter-area reachability between areas. When the inter-area prefix is generated based on a prefix which is directly attached to the ABR, the NP-Flag SHOULD NOT be set. The NP-Flag (No-PHP) MUST be be set for Prefix-SIDs allocated to redistributed prefixes, unless the redistributed prefix is directly attached to the ASBR, in which case the NP-flag SHOULD NOT be set. However, the E-flag (Explicit-Null Flag) is not described. Should we clarify it also? I think E-flag SHOULD NOT be set if the prefix is not directly attached to the ABR or ASBR, and if necessary, it SHOULD be set if the prefix is directly attached to the ABR or ASBR. The existing draft says: "If the NP-flag is not set then the received E-flag is ignored." Given that the draft clearly states when the NP-flag is set on ABR/ASBR above statement should be sufficient. thanks, Peter Regards, Chao Fu ___ spring mailing list mailto:spr...@ietf.org https://www.ietf.org/mailman/listinfo/spring . . ___ OSPF mailing list OSPF@ietf.org https://www.ietf.org/mailman/listinfo/ospf
Re: [OSPF] [spring] One question on E-flag of ABR/ASBR in OSPF SR extension
Hi Peter, Yes I think it is better to add the text that would say that ABR MUST NOT set the E-bit. If not, when ABR receives the intra prefix with SID, it would be thought to inherit the SID value including its flags (NP and E) and then flood the SID to other areas. I guess that's why NP is described expressly on ABR not to do so. Then I think E-flag should be clarified also in case the non-directly attached prefix sets it. It could avoid any misunderstanding. Anyway, the E-flag will not be set in this case, right? Thanks. Regards, Chao Fu -Original Message- From: Peter Psenak [mailto:ppse...@cisco.com] Sent: Thursday, May 18, 2017 17:47 To: Chao Fu <chao...@ericsson.com>; spr...@ietf.org Subject: Re: [spring] One question on E-flag of ABR/ASBR in OSPF SR extension Hi Chao, On 18/05/17 11:15 , Chao Fu wrote: > Hi Peter, > > It's right if the NP-flag is not set then the received E-flag is > ignored. But, if the NP-flag is SET because the prefix is not directly > attached to the ABR, E-flag will not be ignored why would an E-bit be set in such case? It's up to the originator of the SID to decide when to set the E-bit. If the ABR does not set the E-bit no PHP would be done and advertised SID will be preserved. Are you trying to add the text that would say that ABR MUST NOT set the E-bit either? I don't think it's necessary. thanks, Peter > and the upstream neighbor will replace the Prefix-SID with the Explicit-NULL > label 0. I guess actually the packet should be forwarded with the original > prefix-SID, and no need to pop the 0 label and look up path again. > > Regards, > Chao Fu > > > -Original Message- > From: Peter Psenak [mailto:ppse...@cisco.com] > Sent: Thursday, May 18, 2017 16:54 > To: Chao Fu <mailto:chao...@ericsson.com>; mailto:spr...@ietf.org > Subject: Re: [spring] One question on E-flag of ABR/ASBR in OSPF SR > extension > > Hi Chao, > > On 18/05/17 09:44 , Chao Fu wrote: >> Hi, >> >> Should we clarify how to set E-flag for ABR/ASBR in OSPF SR extension? >> >> In >> https://www.ietf.org/id/draft-ietf-ospf-segment-routing-extensions-14. >> txt, the draft describes how to set NP-flag on ABR and ASBR (Section >> 5 [Page >> 14]): >> >> The NP-Flag (No-PHP) MUST be set for Prefix-SIDs allocated to >> inter- >> >> area prefixes that are originated by the ABR based on intra-area >> or >> >> inter-area reachability between areas. When the inter-area >> prefix is >> >> generated based on a prefix which is directly attached to the >> ABR, >> >> the NP-Flag SHOULD NOT be set. >> >> The NP-Flag (No-PHP) MUST be be set for Prefix-SIDs allocated to >> >> redistributed prefixes, unless the redistributed prefix is >> directly >> >> attached to the ASBR, in which case the NP-flag SHOULD NOT be set. >> >> However, the E-flag (Explicit-Null Flag) is not described. Should we >> clarify it also? I think E-flag SHOULD NOT be set if the prefix is >> not directly attached to the ABR or ASBR, and if necessary, it SHOULD >> be set if the prefix is directly attached to the ABR or ASBR. > > The existing draft says: > > "If the NP-flag is not set then the received E-flag is ignored." > > Given that the draft clearly states when the NP-flag is set on ABR/ASBR above > statement should be sufficient. > > thanks, > Peter > > >> >> Regards, >> >> Chao Fu >> >> >> >> ___ >> spring mailing list >> mailto:spr...@ietf.org >> https://www.ietf.org/mailman/listinfo/spring >> > > . > ___ OSPF mailing list OSPF@ietf.org https://www.ietf.org/mailman/listinfo/ospf