Opened 6 years ago

Last modified 6 years ago

#410 new editorial

Initial MID on restart after blockwise transfer

Reported by: achim.kraus@… Owned by: draft-ietf-core-block@…
Priority: minor Milestone:
Component: block Version:
Severity: - Keywords:
Cc:

Description

Doing tests on firmware updates using

  • a blockwise transfer (draft-ietf-core-block-20),
  • a random initial value of the MID (as recommended in RFC7252, 4.4. last part of implementation note),
  • I found on restarts short after such a transfer, a rather high probability of MID collision.

(This is obvious caused by having rather more MIDs in use, than in normal communication situations).

Therefore I would propose to add an implementation note in "draft-ietf-core-block" or/and "draft-ietf-lwig-coap" for proper handling of that situation.

On my experience there are two possibilities:
a: on graceful shutdown (after blockwise) store the last MID and load it on restart.
b: if on restart (after blockwise) communication errors occur, wait for at least EXCHANGE_LIFETIME to escape the “probable” MID collision. This may also be the strategy, when a fails to load the MID for any reason.

Change History (1)

comment:1 Changed 6 years ago by achim.kraus@…

I hope "new editorial" is the right one :-).
Otherwise feel free to change that.

Note: See TracTickets for help on using tickets.