On Jun 1, 2011, at 9:23 AM, Randy Bush wrote:
>> Greetings again. Reading through the -12 draft and the mailing list
>> archives, I cannot figure out what a cache is supposed to do when it
>> has new data and thus a new serial number. Does it have to send a
>> Serial Notify PDU? I'm pretty sure that the draft says all of "yes",
>> "no", and "maybe" in different places. There should be one clear
>> answer.
>
> should be a MAY or a SHOULD.
There is a big difference. The WG should pick one (or none, if they like my
idea of getting rid of "sometimes the server speaks without being spoken to").
What do you see as the purpose of "MAY" or "SHOULD", given that the client
router is already polling as often as it likes (with a minimum of once an hour)?
> read up on dns notify.
Did that a very long time ago and a few times since then. Googling for it now
shows how unclear it is to developers and users. :-(
> please point out where it is ambiguous and i will fix.
Section 4:
When the cache updates its database, it sends a Notify message to
every currently connected router.
Section 5.1:
The cache notifies the router that the cache has new data.
Section 6.2:
| -------- Notify ----------> | (optional)
. . .
The cache server SHOULD send a notify PDU with its current serial
number when the cache's serial changes, with the expectation that the
router MAY then issue a serial query earlier than it otherwise might.
This is analogous to DNS NOTIFY in [RFC1996]. The cache SHOULD rate
limit Serial Notifies to no more frequently than one per minute.
The first two indicates "MUST", the third indicates "MAY", the fourth indicates
"SHOULD", but then modifies that with SHOULD-level rate limiting.
--Paul Hoffman
_______________________________________________
sidr mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/sidr