Re: OSPF graceful restart question

[Sanjay BM <sanjaybm at GMAIL.COM>]

Hi Vishwas,

I was wondering how could there be a loop in the case I mentioned?
Assume the following network:

  Router-D ------ Router-A ------ Router-B -------- Router-C

Router-D is a backbone router, Router-A is an ABR, Router-B and
Router-C are in area 2. There is a virtual-link from Router-A to
Router-C.

Suppose Router-C went down for some reason, we would have its area 0
Router-LSA in the database of Router-A and Router-D till it ages out.
Note that at this point Router-A has removed the virtual-link from its
area 0 Router-LSA.

Before Router-C's area 0 Router-LSA ages out, if Router-A enters OSPF
graceful restart, then Router-D as a helper will eventually send
Router-C's area 0 Router-LSA to Router-A.

My questions is: Should Router-A consider this as a LSA inconsistency
and hence exit the restart mode unsuccessfully?

Would the following be a wrong thing to do?
      If Router-A looks at its own pre-start area 0 router-LSA, it
will find that there is no virtual-link in it. Hence Router-A should
not consider Router-C's area 0 Router-LSA as a inconsistency. Thereby,
Router-A should not exit restart mode unsuccessfully.

Kindly let me know.

Thanx,
Sanjay

On Mon, 13 Sep 2004 22:28:26 -0700, Vishwas Manral <vishwas at sinett.com> wrote:
> Hi Sanjay,
>
> Let me get to your question in a broader sense. Graceful Restart is used
> to prevent as far as possible any packet loss when the control plane is
> brought down. On the otherside, we get out of Graceful restart as soon
> as we figure out that a loop could occur. So for any change which could
> result in a loop in the network we should exit hitless restart. (Your
> case included)
>
> The RFC also states of future enhancements for even stricter checks to
> prevent exiting graceful restart, as far as possible (hence utilizing
> Graceful Restart to the fullest).
>
> This is a reply to my mail from John Moy (way back in time)
> http://peach.ease.lsoft.com/scripts/wa.exe?A2=ind0203&L=ospf&T=0&F=&S=&P
> =1888 . I intend to publish this as a draft sometime to meet future
> enhancements for Hitless Restart.
>
> Thanks,
> Vishwas
>
>
>
> -----Original Message-----
> From: Mailing List [mailto:OSPF at PEACH.EASE.LSOFT.COM] On Behalf Of
> Sanjay BM
> Sent: Tuesday, September 14, 2004 12:48 AM
> To: OSPF at PEACH.EASE.LSOFT.COM
> Subject: OSPF graceful restart question
>
> Hi,
>
> RFC3623 Section 2.2, bullet item #2 says:
>
>      2) Router X receives an LSA that is inconsistent with its pre-
>         restart router-LSA.  For example, X receives a router-LSA
>         originated by router Y that does not contain a link to X, even
>         though X's pre-start router-LSA did contain a link to Y.  This
>         indicates that either a) Y does not support graceful restart,
>         b) Y never received the grace-LSA or c) Y has terminated its
>         helper mode for some reason (Section 3.2).  A special case of
>         LSA inconsistency is when Router X establishes an adjacency
>         with router Y and doesn't receive an instance of its own pre-
>         restart router LSA.
>
> Is the opposite of this example also true? i.e, X receives a
> router-LSA originated by router Y that contains a link to X, but X's
> pre-start router-LSA does not contain a link to Y. This indicates that
> router Y was cut-off from the network before router X entered graceful
> restart and router Y's router-LSA has not yet max aged.
>
> Should router X exit restart mode unsuccessfully for this above case?
>
> Thanx,
> Sanjay
>