Channing wrote:

Checkout http://wiki.dovecot.org/Plugins. You can search the mailing list archives for details on how to implement it. Timo has documented it a few times on the list.

I'll assume you're talking about the Trash plugin, yes? I've seen this and understand it, but it is quota based -- we don't use quotas, and it doesn't do what I want anyways. Something similar but was based on date (days) rather than Quota, and didn't empty that mailbox just deleted the

XXX days messages.

thx, -te

PS: is there a searchable web archives link that I missed? The standard Mailman archives have no searching capability.

-- Troy Engel | Systems Engineer Fluid, Inc | http://www.fluid.com

Scott Alter wrote:

Actually, the plugin you are looking for is named Expire, but it does not work with the current Dovecot version. You need the CVS head version, which I do not believe is recommended for use in production. The Trash plugin is completely separate.

Aha! Thanks for those links and the info; even if I can't use it right now it's good to know it's in the works with HEAD. I'll probably use Mike T's trick for now and just keep an eye out on the newer releases...

Since I'm new to dovecot I'm not familiar with the release roadmap -- is this something slated for inclusion to a future RC for 1.0, or are we talking it's pushed out to 1.1 (sic) or whatnot? I'm guessing the latter...

thanks! -te

-- Troy Engel | Systems Engineer Fluid, Inc | http://www.fluid.com

Mick T wrote:

I personally use a shell script to achieve this:

find /var/virtualmail/ -regex '.*/\.\(Trash\|Junk\)\(/.*\)?\/\(cur\|new\)/.*' -type f -mtime +30
-exec rm '{}' \;

Nice. :) What's your load like doing this compared to number of users? I'm using a Maildir-on-NFS setup here, so there's always that to account for as well... I'm guessing you run this at 5am or so daily?

What about the Dovecot indexes -- will they adjust/fix/repair automagically when the user logs in? Same with dovecot-uidlist, it's self-healing?

-te

-- Troy Engel | Systems Engineer Fluid, Inc | http://www.fluid.com

Mick T wrote:

I have around 5000 mailboxes on local SCSI drives, but yeah I nice -20
the script, and run early in the morning. It doesn't take too long to run

Cool, thanks. (sorry about the Mike T instead of Mick T in the other response, oops) I found that I get a nice speed boost by using a bit more bash-fu with find's target, since my mailboxes are NFS I want tailor to what I know we have. Since all mailboxes are /home/X/XXXX/ (e.g. /home/t/tengel/) I am going with:

find /home/*/*/ -regex '.*/Maildir/\.Junk/\(cur\|new\)/.*' -type f -ctime +14 -exec rm -f '{}' \;

I ran a quick test using some ls -l action instead of rm to get a speed test idea:

# time find /home/*/*/ -regex '.*/Maildir/\.Junk/\(cur\|new\)/.*' -type f -ctime +7 -exec ls -l '{}' \; > results.log

real 23m15.837s user 0m54.882s sys 1m22.567s

That'll do! Not the fastest, but it works. :) -te

-- Troy Engel | Systems Engineer Fluid, Inc | http://www.fluid.com

On Thu, Feb 01, 2007 at 10:24:55PM -0800, Sean Kamath wrote:

You'll probably have better luck with:

find /home/?/????/.*/Maildir/.Junk -regex '\(cur\|new\)/.*' -type f - ctime +14 -print | xargs rm -f

This will not work well with spaces in the paths. Switch back to - exec if so. You might also run into limitations with find with 5000

If using GNU find / xargs you can use the '-print0' find
option and the '-null' xargs option which should handle white-
space'd filename properly.





							John

-- "I'm sorry but our engineers do not have phones." As stated by a Network Solutions Customer Service representative when asked to be put through to an engineer.

"My other computer is your windows box." Ralf Hildebrandt