ThiefMaster's questions

In HAProxy (1.5) I have this rule to redirect all requests to a canonical domain:

redirect prefix https://domain.tld code 301 unless { hdr(host) -i domain.tld }

This works fine. However, I also have stats enabled (stats uri /haproxy-stats). For these I would like to disable the redirect as I need to access specific haproxy machines (they are behind a DNS round-robin) through their own host name.

Is there any way of configuring HAProxy in such a way?

When rejecting emails that are classified as spam by whatever means, there's always the chance of false positives. On the other hand, when accepting emails that may be spam and dumping them into a spam folder the owner of the mailbox might not check that folder and miss an important email while the sender assumes it has been delivered.

I think the perfect solution for this would be to reject the email as far as the sender is concerned but still deliver it to a spam folder.

So my questions are:

• Is it possible/allowed for an SMTP server (as far as the SMTP protocol and related RFCs are concerned) to reject an email after receiving everything (i.e. all header and the full message body) without errors?
• Is it possible to achieve this behavior with Postfix, ideally without having to modify its code.

I'm considering to build a server using the Supermicro X10SLL-F board with a Xeon E3-1231 v3 CPU. However, according to my research, this CPU requires a BIOS update (unfortunately Supermicro does not have BIOS changelogs or CPU compatibility lists so I cannot confirm this).

Since I do not own any other Xeon or i3 which would surely work without a BIOS update I wonder if it's possible to use the IPMI interface of the board to perform the BIOS update even without a (compatible) CPU on the board.

I'm using exim and dovecot. All users are virtual, i.e. they come from a database. A lookup for the destination addresses for a given localpart@domain may yield multiple rows containing either the target mailbox or the target email address.

The following router works fine for delivering all emails properly:

pg_aliases:
  driver = redirect
  directory_transport = address_file
  data = ${sg{\
              ${lookup pgsql { SELECT CASE WHEN substr(c_dest, 1, 1) = '/' THEN (c_dest||'@@'||c_domain) ELSE c_dest END FROM aliases WHERE c_local = '${quote_pgsql:$local_part}' AND c_domain = realDomain('${quote_pgsql:$domain}');}}}\
              {/(.+)@@(.+)}\
              {/var/spool/mail/\$2/\$1/Maildir/}}

That "code" is old and initially not written by me, but it works and I see why: If the destination is a mailbox (/mailboxname) it converts it to a path which will cause the directory_transport to be invoked. Otherwise the destination is an address which will cause the redirect driver to deal with it.

However, I'd like to use use dovecot-lda for local delivery now.

This means I'll need to call a transport which will then take the mailboxname and domain from $address_data and deliver to it:

dovecot_lda:
  driver = pipe
  command = /usr/libexec/dovecot/dovecot-lda -d ${extract{dest}{$address_data}}@${extract{domain}{$address_data}} -f $sender_address -a $original_local_part@$original_domain
  ...

However, I don't see a way to make the redirect router call that transport. When using a separate accept router in addition to the redirect router it will only deliver to the first destination mailbox since it doesn't seem to handle multi-row results in the address_data lookup of the router:

pg_dovecot:
  driver = accept
  transport = dovecot_lda
  condition    = ${lookup pgsql { SELECT substr(c_dest, 2) AS dest, c_domain AS domain FROM aliases WHERE c_local = '${quote_pgsql:$local_part}' AND c_domain = realDomain('${quote_pgsql:$domain}') AND substr(c_dest, 1, 1) = '/'; }}
  address_data = ${lookup pgsql { SELECT substr(c_dest, 2) AS dest, c_domain AS domain FROM aliases WHERE c_local = '${quote_pgsql:$local_part}' AND c_domain = realDomain('${quote_pgsql:$domain}') AND substr(c_dest, 1, 1) = '/'; }}

I wonder if there's a way to solve this problem.

I currently have a file server with 3 1.5TB disks in a RAID5 array. Since it's pretty much full, I got three additional disks (also 1.5TB each).

Now I'd like to switch to RAID6 since 6TB space is enough and I'd like to have the increased safety of raid6. While I do have a full backup - i.e. I could simply create a new array and restore the backup - I'd prefer to switch without having to restore a backup. Is this possible and if yes, how?

I'd like to use dovecot with sql-based virtual users (which already works fine in my current exim+courier-imapd setup).

In my setup there are "mailboxes" which have a mailboxname@domain and a password. Besides that, there are "addresses" which map actual email addresses to mailboxes.
When logging in via IMAP or POP3 users use the mailbox name and password. So I simply check the mailbox table in the dovecot passdb query which works fine.

The userdb query is kind of a problem though: When a user logs in, the available information is obviously the mailboxname@domain username which can then be used to create the mailbox path (even possible with a static userdb).
When an email is delivered via LMTP, however, I need to query the addresses table to retrieve the destination mailbox - so I'd need a different query in this case.

One possible solution would probably be using some conditionals on the DB level which execute different queries depending on the %s variable (service name, e.g. lmtp) - but obviously that'd be a huge mess.

Another solution might be simply using LDA instead of LMTP as I can them perform the lookup of the mailbox name in exim and simply pass it to the LDA program - but using LDA not only involves an additional process being launched (not that it matters on my low-volume server, but anyway) and probably worse error handling as LDA can only return exit status codes to exim unlike LMTP which probably returns a verbose error message.

So I'd like to know if there's a way to perform different userdb queries depending on where they are used - in a delivery or when accessing a mailbox.

I currently have a setup using exim as the MTA and courier-imapd for POP3/IMAP access.

Now I plan moving to dovecot instead of courier but for testing purposes I'd like to run it on a different port for some time. Obviously that makes no sense without any incoming emails so I'd like to deliver mails (for a certain domain if possible) to both courier-imapd and dovecot.

So I'm looking for some suggestions on how to configure exim to behave in the specified way. Please note that I'd like to pass the messages to dovecot through its LDA so using a system filter that writes the email to a specific filesystem location won't work.

Since @ChrisS asked for more details: I'd like to test things with dovecot before (deciding to) switch. And the easiest way is doing that with a "copy" of my real mailbox - but since I also want to test dovecot's sieve filters I need to deliver emails to it via its LDA (or LMTP, but I don't see a big advantage in that) - and for courier the emails are written directly to the maildir by exim. Obviously the maildir of dovecot will be in a different location to avoid collisions.