lubomir.brindza's questions

I'm trying to execute a scheduled task on a large-ish pool of virtual servers, but I want to minimize the resulting performance impact on the hypervisor, ideally executing it on one server at a time.

Think something like this:

    Hr07.Min01.host_1::
        "scheduled_service_restart"    usebundle   =>  service_restart;
    Hr07.Min02.host_2::
        "scheduled_service_restart"    usebundle   =>  service_restart;
    Hr07.Min03.host_3::
        "scheduled_service_restart"    usebundle   =>  service_restart;
    ...

Now, this works; but the list will grow significantly, and isn't exactly pretty to look at.

Can I match the hostname sequence number (or sequence number of any general class) against the integer in the MinXY hardclass somehow?

I'm trying to set up my postfix server to alter the sender address of outgoing e-mails for one specific recipient domain (example.com) -- otherwise I'd just configure smtp_generic_maps in main.cf and be done with it.

From what I understand, for this sort of configuration, there is a need to set up an extra transport that handles the traffic I want to get mangled.

So far I've set up an extra smtp transport:

transport:

example.com    drewrite:

master.cf:

smtp         unix  -       -       -       -       -       smtp
drewrite     unix  -       -       -       -       -       smtp 
    -o smtp_generic_maps=hash:/etc/postfix/rewrite_domain

rewrite_domain:

[email protected]     [email protected]

The problem is that amavis is also configured to run on that server, and it seems to hinder the address rewriting in some way I honestly can't wrap my head around.

The curious thing is that when I bypass the transport and set the smtp_generic_maps in main.cf to overwrite the sender address for all recipients; the address gets rewritten OK.

This led me to believe that perhaps the transport is not set up correctly; but postfix logs show that it is in fact being used (and when I did manage to bork it earlier I got deferred messages in the queue).

The only way I've been able to get this working was to disable amavis globally.

I've updated my mail server last night, from Debian etch to lenny. So far I've encountered a problem with my postfix installation, mainly that I managed to broke the IMAP access somehow.

When trying to connect to the IMAP server with Thunderbird, all I get in mail.log is:

Feb 12 11:57:16 mail imapd-ssl: Connection, ip=[::ffff:10.100.200.65] 
Feb 12 11:57:16 mail imapd-ssl: LOGIN: ip=[::ffff:10.100.200.65], command=AUTHENTICATE 
Feb 12 11:57:16 mail authdaemond: received auth request, service=imap, authtype=login 
Feb 12 11:57:16 mail authdaemond: authmysql: trying this module 
Feb 12 11:57:16 mail authdaemond: SQL query: SELECT username, password, "", '105', '105', '/var/virtual', maildir, "", name, "" FROM mailbox WHERE username = '[email protected]'  AND (active=1) 
Feb 12 11:57:16 mail authdaemond: password matches successfully 
Feb 12 11:57:16 mail authdaemond: authmysql: sysusername=<null>, sysuserid=105, sysgroupid=105, homedir=/var/virtual, [email protected], fullname=<null>, maildir=xoxo.sk/[email protected]/, quota=<null>, options=<null> 
Feb 12 11:57:16 mail authdaemond: Authenticated: sysusername=<null>, sysuserid=105, sysgroupid=105, homedir=/var/virtual, [email protected], fullname=<null>, maildir=xoxo.sk/[email protected]/, quota=<null>, options=<null>

...and then Thunderbird proceeds to complain that it cant' login / lost connection. Thunderbird is definitely not configured to connect through SSL/TLS.

POP3 (also provided by Courier) is working fine.

I've been mainly looking for a way to make the courier-imap logging more verbose, like can be seen for example here.

Edit: Sorry about the mess, I've found that I've been funneling the log through grep imap, which naturally didn't display entries for authdaemond.

The verbose logging configuration entry is found in /etc/courier/imapd under DEBUG_LOGIN=1 (set to 1 to enable verbose logging, set to 2 to enable dumping plaintext passwords to logfile. Careful.)