Murder question

[Rob Siemborski]

On Fri, 2 Apr 2004, Etienne Goyer wrote:

> Well, what I mean is that there is a time window where mailbox operation
> front the frontend would succeed, but the outcome not received by the
> frontend.  Example :
>
> 1. MUPDATE master down, frontend loose connection
> 2. MUPDATE master up again
> 3. frontend issue CREATE
> 4. backend RESERVE then CREATE succeed
> 5. MUA (or whatever) look for new folder, can't see it.
> 6. frontend reconnected to MUPDATE master
> 7. frontend receive update and now see new folders
>
> I am interested in minimizing the time between 2 and 6, to avoid 5.

I think that you're trying to address something that is, in most practical
applications, a non-issue, given that the mupdate master should never
really be unavailable (and we already know that there are other problems
in the system when it is).

In all likelihood the majority of the time will not be the delay in the
reconnect pause, but the delay as the mailbox list resyncs.

-Rob

-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Rob Siemborski * Andrew Systems Group * Cyert Hall 207 * 412-268-7456
Research Systems Programmer * /usr/contributed Gatekeeper

---
Home Page: http://asg.web.cmu.edu/cyrus
Wiki/FAQ: http://cyruswiki.andrew.cmu.edu
List Archives/Info: http://asg.web.cmu.edu/cyrus/mailing-list.html