[6lo] draft-ietf-6lo-minimal-fragment-00: When to remove VRB entries?

[Martine Lenders <[e-mail]>]

Hi,

I'm currently in the process of implementing first prototypes of both
minimal fragment forwarding [1] and fragment recovery [2] for RIOT [3].

However, I'm unsure how I can determine when it is safe to remove a VRB
entry at least for the minimal forwarding case (even for a successfully
transmitted datagram). As far as I have seen not even the original VRB
draft [4] mentions a strategy for that.

For the success case I could of course just count the bytes of the
fragments and remove the VRB entry once I reach the datagram size, however
this does not account for possibly received duplicates (unless I keep track
of at least the intervals of all received fragments which of course costs
more memory).

The failure case is easier, since I could just use the timeout used in the
original 6Lo fragmentation, but using this for the success case as an
alternative to my proposal above might lead to a congestion of the VRB.

Are there any other strategies I might have missed?

Kind regards,
Martine Lenders

[1] https://datatracker.ietf.org/doc/draft-ietf-6lo-minimal-fragment/
[2] https://datatracker.ietf.org/doc/draft-ietf-6lo-fragment-recovery/
[3] https://github.com/RIOT-OS/RIOT
[4]
https://datatracker.ietf.org/doc/draft-ietf-lwig-tcp-constrained-node-networks/