Re: [Lsr] Question about OSPF (transit area routing loop)

[Sergey SHpenkov <[e-mail]>]

Hi, Aijun

чт, 12 мар. 2020 г. в 04:21, Aijun Wang <[e-mail]>:

> Hi, Sergey:
>
>
>
> ABR-1 advertised the less optimal route information to other routers(in
> virtual link environment, for avoiding other possible loop scenario,
> mentioned in G.2 of RFC 2328 as pointed out by Acee), then the other
> end(ABR-3) of the virtual link should also prefer this less optimal route
> information.
>
> This is similar to your proposal at
> https://mailarchive.ietf.org/arch/browse/lsr/?gbt=1&index=rPZK_euJTkK0TcRhnI-l1sehxRA
>
>
>
> On the other hand, is it necessary to build another virtual link between
>  ABR-2 and ABR-3 in your topology, or else the traffic between ASBR and
> ABR-3 will be unidirectional? (ASBR can reach the ABR-3, but ABR-3 can’t
> reach the ASBR?)
>
>
>
If create another virtual link between ABR_2 and ABR_3 the path from ABR_3
to ASBR in this topology will not change.

>
>
> Best Regards.
>
>
>
> Aijun Wang
>
> China Telecom
>
>
>
> *发件人:* [e-mail] [mailto:[e-mail]] *代表 *Sergey
> SHpenkov
> *发送时间:* 2020年3月12日 4:31
> *收件人:* Aijun Wang
> *抄送:* [e-mail]
> *主题:* Re: [Lsr] Question about OSPF (transit area routing loop)
>
>
>
> Hi Aijun,
>
>
>
>
>
> ср, 11 мар. 2020 г. в 13:46, Aijun Wang <[e-mail]>:
>
> Hi, Sergey:
>
>
>
> If so, ABR-3 should also receive this SumLSA-4 for the ASBR(with cost
> 300), and then prefer the path via ABR-2 to reach ASBR(with cost 20).
>
> Then there will be no loop then?
>
>
>
> no, because sumLSA-4 in step 16.3 will be less preferred for ABR_3 than
> previously calculated using the Dijkstra algorithm (with cost 3)
>
>
>
> Or, how many SumLAS-4 will be advertised by ABR-1? If it selects and
> advertises only one (3 or 300), then the loop will not be emerged.
>
> Currently, it seems it advertises this SumLAS-4 with the cost 300 to RT_1
> and with the cost 3 to ABR-3?
>
>
>
> only one with a cost of 300
>
>
>
> Best Regards.
>
>
>
> Aijun Wang
>
> China Telecom
>
>
>
> *发件人:* [e-mail] [mailto:[e-mail]] *代表 *Sergey
> SHpenkov
> *发送时间:* 2020年2月26日 15:20
> *收件人:* [e-mail]
> *主题:* Re: [Lsr] Question about OSPF (transit area routing loop)
>
>
>
> Acee,
>
>
>
> Because ABR_1 creates SumLSA-4 for the ASBR not from the backbone
> area. The cost of SumLSA-4 for ASBR is 300.
>
>
>
> Thanks,
>
> Sergey
>
>
>
> вт, 25 февр. 2020 г. в 22:44, Acee Lindem (acee) <[e-mail]>:
>
> Hi Sergey,
>
> I don’t see why RT_1 wouldn’t go through ABR_1 to get to the ASBR.
>
> Thanks,
>
> Acee
>
>
>
> *From: *Lsr <[e-mail]> on behalf of Sergey SHpenkov <
> [e-mail]>
> *Date: *Tuesday, February 25, 2020 at 2:38 PM
> *To: *"lsr@ietf..org <[e-mail]>" <lsr@ietf..org <[e-mail]>>
> *Subject: *[Lsr] Question about OSPF (transit area routing loop)
>
>
>
> Hi,
>
> In section 16.3 of the OSPF RFC 2328 standard, it is stated that all ABR
> routers
>
> connected to a transit area are required to check the sumLSA contained
> within
>
> this area in order to possibly improve the intra-area and inter-area
> backbone routes
>
> for themselves.
>
>
> See the picture:
>
> The RT_1 and ABR_3 routers will use different paths to the ASBR router:
>
> ABR_3 -> RT_1 -> ABR_1 -> ASBR = cost 3
> RT_1 -> ABR_3 -> ABR_2 -> ASBR = cost 21
>
> route loop between RT_1 and ABR_3
>
> Please explain this situation
>
> Thanks,
> Sergey
>
>
>
>