• Community Support
  • USEnglish
  • encoding from different mail clients using mailcow as mail server

O.k. thank’s for sharing your thoughts.

I’ll check/disable in case anything on my asus home router, and will test again. I’ll be back.

I disabled diversion and skynet, the router DNS is unbound.

DKIM passes, encoding/format is faulty:

Date: Fri, 17 Nov 2023 14:32:31 +0100
To: djrEXr4DjZ2CAP@dkimvalidator.com
MIME-Version: 1.0
Message-ID: <41-65576b80-d-677bcc00@203232121>
Subject: =?utf-8?q?T=C3=84SCHT?=
User-Agent: SOGoMail 5.9.0
X-Last-TLS-Session-Version: None

——==-OpenGroupware_org_NGMime-65-1700227951.204159-2——
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable

sogo html text, signatur and footer mit umlaute
=C3=A4=C3=B6=C3=BC=C3=9F

–=C2=A0
signature plain text =C3=B6=C3=A4=C3=BC=C3=9F

plain footer
=C3=B6=C3=A4=C3=BC=C3=9F

——==-OpenGroupware_org_NGMime-65-1700227951.204159-2——
Content-Type: text/html; charset=utf-8
Content-Transfer-Encoding: quoted-printable

<html>sogo html text, signatur and footer mit umlaute<br />=C3=A4=C3=B6=C3=BC=
=C3=9F<br /><br />–&nbsp;<br />signature plain text =C3=B6=C3=A4=C3=BC=C3=9F</=
html>

html footer
=C3=B6=C3=A4=C3=BC=C3=9F

——==-OpenGroupware_org_NGMime-65-1700227951.204159-2——–

What is this ——==-OpenGroupware_org_NGMime-65-1700227951.204159-2——– about?

From the postfix log, sent with TB, no matter same result with SOGo:

B0AD2427BCA: replace: header Received: from cdc97c45df2b (mailcowdockerized-sogo-mailcow-1.mailcowdockerized_mailcow-network [172.22.1.248])??(Authenticated sender: x.y@z.de)??by mail.z.de (Postcow) with ESMTP from mailcowdockerized-sogo-mailcow-1.mailcowdockerized_mailcow-network[172.22.1.248]; from=x.y@z.de to=check-auth@verifier.port25.com proto=ESMTP helo=<cdc97c45df2b>: Received: from [127.0.0.1] (localhost [127.0.0.1]) by localhost (Mailerdaemon) with ESMTPA id B0AD2427BCA??for check-auth@verifier.port25.com; Sat, 18 Nov 2023 01:54:47 +0100 (CET)

The header, the body and the DKIM key is corrupted. No idea what replace: header received means…

DNS-Check in mailcow says everything is fine.

Mailcow is running on top of archlinux. Laptop is WLAN connected. Static IP is 192.168.XX.YYY. Gateway is the router (for test every possible interference disabled), tried DMZ, portforwarding, etc. There’s no reverse proxy running on the laptop. No firewall, no filters on the laptop. Something must be wrong on the way from the container via docker to the router.

As I already mentioned, I’m running a few other mailservers NOT DOCKERIZED, nearly the same configuration (postfix, dovecot, fail2ban, rspamd, sogo (same ver.).

I have to give up. At this very moment I cannot use mailcow in production environment. I’m sorry.

Ok… one more thing came to my mind while reading your last post: from the debug outputs you posted and the .de domains I guess you are German? Did you setup your Archlinux in German as well? If so, maybe it’s some kind of encoding issue with the German locale? I haven’t had a look yet if the locale of the host system is used while setting up the docker containers of mailcow in the first place, but maybe… My host system is running with en_us.UTF-8

You guess right, I’m German.

Here’s the info:

#$ localectl
System Locale: LANG=C.UTF-8
VC Keymap: de
X11 Layout: de

To my mind another possible issue came across. As I have a pool IP from vodafone (cable), this is not a static IP, it might be impossible to setup a mailserver without having a static IP4. They charge you extra if you want a static one.

Is there a workaround possible with i.e. DDNS and cname in DNS? Don’t know, never tried this. There was no reason for this…

    stefan21 it might be impossible to setup a mailserver without having a static IP4.

    That is true. It is working for receiving mails, but no chance to send mails via a dynamic IP.
    Pay for a relay service then (they can read your mails…), or get a business cable product with static IP and the possibility to have a reverse DNS entry, or pay for a decent hoster.

      I´m sending mail from a dynamic IP, no mail´s get rejected so it´s possible.
      Before that i used smtp2go, also work´s great.

        storpotaten no mail´s get rejected so it´s possible.

        With a relay service yes. Otherwise you are very very lucky.

        Maybe, my IP resolves to “94-255-167-***.cust.bredband2.com” the host of my ISP,
        i could be wrong but if everhing else in in place DKIM, SPF and so on the sending of emails seems to work.
        i´m far from an emailserver expert but this is my experience.

          esackbauer That is true. It is working for receiving mails, but no chance to send mails via a dynamic IP.
          Pay for a relay service then (they can read your mails…), or get a business cable product with static IP and the possibility to have a reverse DNS entry, or pay for a decent hoster.

          In business we have only static IP’s. Not for private. Question is, could a dynamic IP have an impact on DKIM? Hmm…?

          Anyone out there running mailcow on top of any linux OS with a dynamic IP?

          storpotaten
          Some servers might still decline your mail because it’s on a DNSBL for dial up/dynamic IPs.
          I know because I am in the same situation 🙂
          This is why I use a relay.

          storpotaten I´m sending mail from a dynamic IP, no mail´s get rejected so it´s possible.

          Thank’s for jumping on.

          AFAIK any email sent from a dynamic IP would be rejected from my customers and suppliers. In business a dynamic IP is no option. It’ll simply won’t work.

          [unknown]

          If you’re using a relay, this is not what I’m looking for. We don’t send email through a relay. We are running our email servers totally on our own. No relay, no smarthost, no connector, … I mean, what for? We are talking about an email server. No need for anything around. That’s it.

            stefan21 If you’re using a relay, this is not what I’m looking for. We don’t send email through a relay. We are running our email servers totally on our own. No relay, no smarthost, no connector, … I mean, what for? We are talking about an email server. No need for anything around. That’s it.

            Sure, it’s “just” an email server, but most of the mails sent by it will be rejected nonetheless due to you dynamic IP, wether it’s private or for business purposes. Looking up the IP blacklist is a default option not only for mailcow.
            And it’s a real pain if you never can be sure if your mail has been delivered or just silently dropped by the SPAM filter…
            Just saying 😃

            Maybe I didn’t make it clear enough: in business we do only use static IP’s and email-servers without relay, smarthost, … All business domains have 10/10 at mail-tester. No issues at all, no matter where or what I test against.

            Not so for private. As I said, the vodafone IP is so-called pool-IP, not static.

            Follow up while digging around:

            1. I had to change my fritzbox 6490 cable in bridge mode. Exposed host for the asus router behind was not enough.
            2. Now the asus shows a WAN IP, which in (real) bridge mode of the fritzbox is a different one as shown in the fritzbox. The WAN IP shown in the asus is to be set in the DNS settings at the hoster.
            3. The laptop is configured as DMZ in the asus.
            4. On the laptop I installed nginx as reversed proxy. I took the example config from mailcow and changed it to my domain.
            5. I started over with a new mailcow-docker installation.
            6. Unbound container was unhealthy. As the asus is running with unbound, I tried to point the forward zone in unbound.conf to the IP of the asus. That didn’t work. Unbound container showed healthy, but error during the acme challenge occurred. I changed the IP to 1.1.1.1. That worked. The challenge completed.
            7. I created a domain and a test user.
            8. I took the dkim key to the DNS (as well as the other settings to make).
            9. I tested from sogo an email to email-tester composed in html. So far o.k., dkim passes, not a 10/10 because auf rDNS setting. This was to expect.
            10. Changed sogo to plain text, email test also o.k.
            11. Tried thunderbird - email tests also o.k.
            12. Did a backup.

            Well - it is was not so easy to track this down. Still some questions open. But maybe it helps someone. I’ll keep this running and testing at home for a while. If the mailcow runs smooth, I’ll migrate my email servers in business. And will donate for this great work.

            regards,
            stefan

              10 months later

              stefan21

              Hey,

              also have the same Fritz!Box (Vodafone) here, but couldn’t get the bridge-mode working.
              Maybe you got a tip?

              Thanks.

              I don’t have a customized FritzBox. It’s an original. Therefore I can’t help. You should ask Vodafone.

                6 days later

                Vodafone is my provider. At home I don’t have a static IP. Therefore I use an unbranded (no customized firmware) fritzbox.

                Read this: ubiquiti - Deutsches Fan Forum Icon Fritz!Box | Cable (6490, 6590, 6591, 6660) Bridge-Mode freischalten - ubiquiti - Deutsches Fan Forum

                And this: mengelke.de Icon Fritz!Box JSTool

                This adds the option “bridged mode” to an unbranded fritzbox.

                In business we use Vodafone only with static IP’s. This allows to run their customized fritzbox firmware in bridged mode.

                If all pre-requisites are fulfilled (read mailcow docs) the mailserver works flawless.