using Murder for migration from UW IMAP

[John A. Tamplin]

Quoting Rob Siemborski <rjs3 at andrew.cmu.edu>:

> So, the murder really isn't intended to work like this.

Right, I understand.  I also looked for more generic IMAP/LMTP proxies but most
of the links were dead and the ones that weren't seemed to be lacking something.

> Does UW-IMAPd even support proxy authentication?

No, logins to it will have to be the user who is accessing their mailboxes. 
Once the client logs into the proxy, isn't that same login information passed
onto the backend server?

> If it does, *offhand*, provided you're not using internal mechanisms to
> move the mailboxes (i.e. you are moving them by hand), I don't really see
> a reason this won't work, but I'm not going to recommend it.

Here is pseudo-code of how I envisioned the process:

Populate mupdated with initial database showing all mailboxes on old server
for each user {
     block mail delivery to the user (LMTP returns temporary failure)
     block write access in UW IMAP (minor hack to locking code)
     connect to Cyrus server as user with backdoor password (via saslauthd)
     for each user mailbox {
          create mailbox on Cyrus server


> Do you really have so much data that it would take 2-3 days to move it?

The tests I have done so far (using both imap-utils mbxcvt and custom perl code)
it looks like it will take around 60 hours to convert everything (2300 users,
23000 folders, 70G total), and that is putting /var/imap on a tmpfs filesystem
for the conversion (about a 2x speed improvement there) and striping /cyrus
across 14 drives in an FC-AL array.

-- 
John A. Tamplin
Unix System Administrator