Thomas Jarosch wrote:
>
>
> > > I just wanted to know if the user rename implementation
> > > is considerd as stable?
> >
> > What implementation are you talking about? If you are talking about the
> > patch that posted to the list a while back, functionally it works.
>
> Exactly :-)
>
> > > Are there any known bugs?
> >
> > I doesn't lock any of the meta-data files (seen state, quota,
> > subscriptions) before it copies and deletes them. I'm waiting for
> > changes to be made to the cyrus_db interface which allow for file
> > locking before I commit the changes.
>
> Do you know when these changes will be available/done?
> Is there a kind of roadmap?
Not really. Its functionality that Larry wants to have in the
distribution (due to popular) demand, its just a matter of finding the
time to look at the filelocking issue.
> > If you are going to use the patch, then I'd first disable the user's old
> > login and make sure that they don't have any open connections to the
> > server. This way you won't have to worry about the lack of
> > file-locking.
>
> Disabling the login should be no problem.
> Is there a safe way to close an active connection only for a certain user?
I don't know about "safe", but you can grep for the login in the
imapd.log to find those process ids corresponding to your user and then
just kill them.
FWIW, I tested a scenario when my test user was accessing the INBOX when
I did the rename, and nothing bad happened (in fact the server handled
it gracefully), but this was using flat files for the meta-data.
Ken
--
Kenneth Murchison Oceana Matrix Ltd.
Software Engineer 21 Princeton Place
716-662-8973 x26 Orchard Park, NY 14127
--PGP Public Key-- http://www.oceana.com/~ken/ksm.pgp
