Moy, John wrote:

> Vishwas-
>
> You have two separate ideas here, I think. First is the
> idea to continue hitless restart regardless of topology
> changes. I think that is a bad idea, and you can read the long
> arguments about this in the archives. However, one of the reasons
> I formatted the body of the grace-LSA as TLV was so that, if
> we wanted to in the future, we could support options like this
> (the "what, me worry?" option) in the future.
>
> As for detecting whether the neighbors will help, and if not, abort
> *any* restart. This we could do. However, I don't think it's
> worth the extra complication.

John,

In most cases, I think your approach will detect that there are
neighbor(s) that will not help. When the restarting router
does an SPF using its pre-restart router LSA, it will abort graceful
restart if the former neighbor's router LSA does not have a back-link (at
least this is how I implemented the inconsistency check in section
2.2 of the draft).

Thanks,
Acee

> Most likely, if the neighbors won't help
> now, they won't help in an hour or two either (I assume that the
> router isn't being restarted/reloaded for fun, but that there is some
> operational reason and hence urgency for doing so). Obviously this
> option wouldn't make any sense for unplanned outages either...
>
> John
>
>
>>-----Original Message-----
>>From: Manral, Vishwas [mailto:VishwasMNETPLANE.COM]
>>Sent: Sunday, January 27, 2002 2:58 AM
>>To: OSPFDISCUSS.MICROSOFT.COM
>>Subject: Re: Change to OSPF Hitless Restart?
>>
>>
>>John,
>>
>>Just another thought.
>>
>>If due to some change, if the helper has a new LSA, it would
>>prevent the
>>restart to be graceful. I thought it may be helpful if we
>>could provide a
>>configurable option, whether to proceed in such a case using hitless
>>restart, or just abort the restart process itself(and work as
>>normal) and
>>try hitless restart at a later stage.
>>
>>-Vishwas
>>
>>
>

• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]