Re: [Dovecot] sessionid in logout line

2012-11-02 Thread Pascal Volk
On 11/03/2012 12:51 AM Aleksandar Lazic wrote: > Dear listmembers. > > I have the following > > login_log_format_elements = service=%s user=<%u> session=%{session} > method=%m rip=%r lip=%l mpid=%e %c %k > > Maybe I have overseen the config option to add the %{session} also to > the logout lin

[Dovecot] sessionid in logout line

2012-11-02 Thread Aleksandar Lazic
Dear listmembers. I have the following login_log_format_elements = service=%s user=<%u> session=%{session} method=%m rip=%r lip=%l mpid=%e %c %k Maybe I have overseen the config option to add the %{session} also to the logout line. Here a example from my log: ### Nov 3 00:46:43 dovecot:

Re: [Dovecot] dovecot dimensioning

2012-11-02 Thread mancyb...@gmail.com
Hi thanks to all for your pointers, very interesting. I'll do my homework about local SAS disks and SANs. Thanks and regards :) Mike On Fri, 02 Nov 2012 09:39:56 -0400 Oscar del Rio wrote: > On 11/ 1/12 07:12 PM, Gregory Finch wrote: > > I believe that Javier is talking IOPS, not throughput. Y

[Dovecot] Feature request: add information to error message: client doesn't have lookup permissions for this user: userdb reply doesn't contain uid (change userdb socket permissions)

2012-11-02 Thread Steffen Kaiser
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Please add the information to this error, which socket has the problem and which uid access is and what is expected. For instance, when the quota dict request fails, because of permission problems, you get a very detailed info about the current pr

Re: [Dovecot] "starting" dovecot

2012-11-02 Thread Dave Abrahams
on Thu Nov 01 2012, Dave Abrahams wrote: > My system never issues the "dovecot start" command. I do, however, run > /usr/local/libexec/dovecot/imap on port 9xxx. I talk to the server > through port 9xxx and through the preauth tunnel. Is this arrangement > OK? Are there some things that will

Re: [Dovecot] dovecot dimensioning

2012-11-02 Thread Oscar del Rio
On 11/ 1/12 07:12 PM, Gregory Finch wrote: I believe that Javier is talking IOPS, not throughput. You need a storage system that is able to provide a high random read + write IOPS. That single drive is going to cause you issues when things get busy. And a disaster when the single drive fails!