Re: [Dovecot] Problem with 1.2 and sieve: failed with unsuccessful implicit keep

2009-07-08 Thread Timo Sirainen
On Wed, 2009-07-08 at 22:11 +0200, Julien Danjou wrote: > Like mine. I don't think this is *bad* since it's, AFAIK, Dovecot > only use in homedir to expand %h. Or do I miss another point? Home directory may also be used for other things, although mainly only when Sieve is enabled. > I really sugg

Re: [Dovecot] Problem with 1.2 and sieve: failed with unsuccessful implicit keep

2009-07-08 Thread Julien Danjou
At 1247082407 time_t, Timo Sirainen wrote: > "let the bug"? The idea is anyway that you return > home=/var/mail/vmail/danjou.info/jd and set > > mail_location = maildir:%h > > That makes much more sense to me. I've considered making it an error to > return relative home dirs from userdb, but that

Re: [Dovecot] Problem with 1.2 and sieve: failed with unsuccessful implicit keep

2009-07-08 Thread Timo Sirainen
On Wed, 2009-07-08 at 15:46 -0400, Timo Sirainen wrote: > "let the bug"? The idea is anyway that you return > home=/var/mail/vmail/danjou.info/jd and set > > mail_location = maildir:%h > > That makes much more sense to me. I've considered making it an error to > return relative home dirs from use

Re: [Dovecot] Problem with 1.2 and sieve: failed with unsuccessful implicit keep

2009-07-08 Thread Timo Sirainen
On Wed, 2009-07-08 at 21:36 +0200, Julien Danjou wrote: > At 1247080982 time_t, Timo Sirainen wrote: > > > The thing is that danjou.info/jd/ is the homedir of the user (returned > > > by SQL), > > > > Make your SQL query return absolute path for home dir. > > Sigh. That sounds like you let the bu

Re: [Dovecot] Problem with 1.2 and sieve: failed with unsuccessful implicit keep

2009-07-08 Thread Julien Danjou
At 1247080982 time_t, Timo Sirainen wrote: > > The thing is that danjou.info/jd/ is the homedir of the user (returned > > by SQL), > > Make your SQL query return absolute path for home dir. Sigh. That sounds like you let the bug. Quite sad to read. > > I smell (did not dig into the code) that th

Re: [Dovecot] Problem with 1.2 and sieve: failed with unsuccessful implicit keep

2009-07-08 Thread Timo Sirainen
On Wed, 2009-07-08 at 21:03 +0200, Julien Danjou wrote: > At 1247075539 time_t, Timo Sirainen wrote: > > http://hg.dovecot.org/dovecot-1.2/rev/6d7f6ea02e17 makes it log the > > current directory in the above error message when the path is relative. > > Ok, that helps to see what's wrong, thanks: >

Re: [Dovecot] Problem with 1.2 and sieve: failed with unsuccessful implicit keep

2009-07-08 Thread Julien Danjou
At 1247075539 time_t, Timo Sirainen wrote: > http://hg.dovecot.org/dovecot-1.2/rev/6d7f6ea02e17 makes it log the > current directory in the above error message when the path is relative. Ok, that helps to see what's wrong, thanks: Jul 8 20:50:16 prometheus dovecot: deliver(j...@danjou.info): mkd

Re: [Dovecot] Problem with 1.2 and sieve: failed with unsuccessful implicit keep

2009-07-08 Thread Timo Sirainen
On Wed, 2009-07-08 at 13:46 -0400, Timo Sirainen wrote: > On Wed, 2009-07-08 at 19:37 +0200, Julien Danjou wrote: > > > Jul 8 18:47:51 prometheus dovecot: deliver(j...@danjou.info): > > mkdir_parents_chown(danjou.info/jd/) failed: Permission denied > > (euid=8(mail) egid=8(mail)) > > The problem

Re: [Dovecot] Problem with 1.2 and sieve: failed with unsuccessful implicit keep

2009-07-08 Thread Timo Sirainen
On Wed, 2009-07-08 at 19:37 +0200, Julien Danjou wrote: > Jul 8 18:47:51 prometheus dovecot: deliver(j...@danjou.info): > mkdir_parents_chown(danjou.info/jd/) failed: Permission denied > (euid=8(mail) egid=8(mail)) The problem is this. It tries to create a non-existing directory, but can't becau

[Dovecot] Problem with 1.2 and sieve: failed with unsuccessful implicit keep

2009-07-08 Thread Julien Danjou
Hi, First, thanks for this awesome piece of software. I've just upgraded to latest dovecot, aka 1.2. I've a problem. I've started using Sieve, and everything works fine, I already received a thousand of mails without any problem. However, there's one mail which is stuck in the postfix queue becau