![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
|
Hi Mathilde: In a general manner, when you lose a next-hop/link you clean all
routes via that next-hop/link. A one hop loss can be detected by NUD. So the retry/remove thing
only applies to a destination, since you cannot NUD it hops away. Parents used to be removed after missing periodic RAs in a row,
like MIP does, but that’s harder with trickle. I think that’s gone
from the spec now. Cheers, Pascal From:
roll-bounces at ietf.org [mailto:roll-bounces at ietf.org] On Behalf Of Mathilde
Durvy (mdurvy) Hi
All, This
is my current understanding: -
Parents are removed based on OF selection (decision to move within the DAG,
change DAG, etc) -
Destination prefixes are removed after the retry / RemoveTimer procedure
described in section 5.10.1.1.1. This is essentially a keep alive mechanism
based on DIO (D bit set) - DAO exchanges. -
Route corresponding to destination prefixes are removed when DAO lifetime
expires? Are they also removed when the corresponding destination prefix is
removed? -
When are routes corresponding to parents removed? Do they also have a lifetime? -
What happens when a neighbor disappears, are the corresponding
parents/destination/routes removed? The
current behavior seems quite asymmetric... There is some kind of keep alive
mechanism for destination prefixes, but not for parents. There is a lifetime
associated with routes corresponding to destination prefixes but not for routes
corresponding to parents. Is that a design choice? I would appreciate some
clarifications. Best, Mathilde
| |||||||||