[bind10-dev] take 2: design and inter-process protocol for shared memory

JINMEI Tatuya / 神明達哉 jinmei at isc.org
Thu Mar 14 18:48:34 UTC 2013


At Thu, 14 Mar 2013 09:20:03 +0100,
Michal 'vorner' Vaner <michal.vaner at nic.cz> wrote:

One more thing:

>  • The ACK. If the Update is a command, it would feel natural to send the ACK as
>    the actual result of the command, not a separate command. The manager could
>    then use some kind of rpcCallAsync (once create that), to simplify its code.
>    The unmap-map should be a fast operation in my understanding.
> 
>    Also, for simplicity, I think we should just be sending the ACK every time.
>    Omitting it on startup won't give us any benefit and it'll be some additional
>    handling.

Regarding the latter part, that's one option.  I thought it's useful
for the manager if it can determine whether an ack is really expected
or a kind of error (sent by someone not requested).

---
JINMEI, Tatuya
Internet Systems Consortium, Inc.


More information about the bind10-dev mailing list