Cyrus 2.3 on shared filesystems

[Attila Nagy]

Ken Murchison wrote:
>> Is replicated murder consists of multiple backend server groups which 
>> has the same mailboxes? So murder will say that user.jsmith is on 
>> server1 AND server2, instead of just saying it's on server1 OR server2?
> Yes.  Any machine in the Murder has local access to any mailbox.
Hmm. If any machine can access any of the mailboxes, why is murder 
necessary?
Above, I described a setup, similar to this (ASCII art follows):

+-----+ +-----+  +-----+ +-----+
|IMAP1| |IMAP2|  |IMAP3| |IMAP4|
+--+--+ +-----+  +--+--+ +-----+
    |   /            |   /
+--+--+          +--+--+
|STOR1|          |STOR2|
+-----+          +-----+

So murder is necessary to make the namespace unified. This way the 
MUPDATE servers say that a mailbox on STOR1 storage is on IMAP1 and 
IMAP2, so IMAP3 can't access mailboxes on STOR1. Does this work?

> Its designed to work on a shared filesystem SAN (GFS, QFS, CXFS) where 
> all of the machines have direct access to each mailbox, but each machine 
> has its own replicated copy of mailboxes.db.
So when I use replicated murder the mailbox database moves to the local 
(backend IMAP) server?

Sorry if the answers are trivial, there are no docs regarding this 
topic, so I cannot RTFM :)

-- 
Attila Nagy                                   e-mail: Attila.Nagy at fsn.hu
Free Services Network (FSN.HU)           phone @work: +361 371 3536
ISOs: http://www.fsn.hu/?f=download            cell.: +3630 306 6758
---
Cyrus Home Page: http://asg.web.cmu.edu/cyrus
Cyrus Wiki/FAQ: http://cyruswiki.andrew.cmu.edu
List Archives/Info: http://asg.web.cmu.edu/cyrus/mailing-list.html