[Dovecot] Converting Courier mailboxes to Dovecot
Hi,
I have around 100 mailboxes (maildirs) to convert from Courier IMAP (4.1.3) to Dovecot.
(From a recent backup) I've tested a few mailboxes on Dovecot 1.0.0, and while the client doesn't notice any difference (at least in Thunderbird) I'm hoping to get rid of the old Courier folders and convert any useful information they may have into Dovecot.
I've tried Timo's migration script (http://www.dovecot.org/tools/courier-dovecot-migrate.pl) but it fails at line 230 with a "no file/directory" error. At that line, it appears to be finding Courier's keywords folder (courierimapkeywords), and looking for a file called ":list".
However in my mailboxes, there's no file by that name, only what looks like a set of standard IMAP filenames, most of which are empty files.
If I can't salvage anything useful, is there any adverse effects to me deleting all the "courier*" files/folders from each mailbox and subfolders so when the client connects back when Dovecot's running, it creates it's own UIDs and keywords etc?
Thanks
Andy.
On 5/4/07, Andy Shellam andy.shellam-lists@mailnetwork.co.uk wrote:
If I can't salvage anything useful, is there any adverse effects to me deleting all the "courier*" files/folders from each mailbox and subfolders so when the client connects back when Dovecot's running, it creates it's own UIDs and keywords etc?
I rsync'ed from one mailserver to another and excluded the courier related files. Then I crawled each user's imap folders on both systems with a perlscript and wrote out the results to two files.
As you suggest various logs/indexes were created by dovecot and all my accounts were the same. Baring a few deliveries. You will loose folder subscription information remember and I know nothing about shared folder stuff.
Btw, the difference between courier and dovecot is insane the dovecot machine I am using is not a 3rd of the courier machine and the speed is quite impressive. I hope to see an additional advantage to using dovcot's lda deliver to prep these indexes as well.
-- Gabriel Millerd
participants (2)
-
Andy Shellam
-
Gabriel Millerd