[bug]: Autoconfig dns records not generated

[giomf]

What happened?

Hi,

Im installing my setup right now and I think I found something odd.
Although its mentioned in the docs that the autoconfig dns records will be generated I can not see them when printing them.

Every other dns records seems to be present, however.

How can we reproduce the problem?

I can reproduce the problem by doing the following steps:

1. Create a new Domain
2. Print the dns records

Version

v0.10.x

What database are you using?

RocksDB

What blob storage are you using?

RocksDB

Where is your directory located?

Internal

What operating system are you using?

Docker

Relevant log output

No response

Code of Conduct

• I agree to follow this project's Code of Conduct

[mdecimus]

Hi, perhaps you don't have any HTTPS ports enabled? Autoconfig entries are only generated is there is a way to access them.

[giomf]

Hi,

Maybe I have to add some details. I have a https listener enabled but without tls because it's handled by my reverse proxy. I can trigger the autoconfig endpoints, and they return the correct configuration.

So I have set the dns records myself and everything is working fine. I just expected that the dns records will get printed but they don't.

[docjyJ]

I have the same issue, I use my reverse proxy for encryption, I have to add these fields manually.

[giomf]

I'm starting to get the feeling that Stalwart isn't necessarily suited to being used behind a reverse proxy. In the last few days that I have been setting up stalwart, I have repeatedly encountered unpleasant errors.

The error I described above is an example of this. I have set an HTTP listener with tls implicit to false because TLS is handled via the reverse proxy. As a result, stalwart does not give me the dns entries because it assumes that I am not using TLS. If implicit tls is activated again, the DNS entries appear.

Another example would be SMTP. If I deactivate implicit tls here (again for the reason that stalwart is behind a reverse proxy) the autoconfig entries are no longer set to SSL but STARTTLS, which leads to problems with clients because they want to establish a plain TCP connection and later want to upgrade to TLS. If I enable implicit tls for the listener again, the autoconfig settings are correct but the listener now expects a TLS connection to be made by the proxy which can also lead to problems.

I would love it if, for example, stalwart offered options to set the autoconfig entries manually or if you could completely deactivate tls for the listeners. (server.tls.enable did not work for me)

In the meantime I have decided to continue using the reverse proxy but more as a certificate creator and logging system as all tcp/http connections are simply forwarded to stalwart.

Also Im willing to help, should I at least open a feature request or bug report about this?

[mdecimus]

Stalwart can be used behind a reverse proxy without issues. You just need to make sure you configure the proxy protocol properly and that you provide Stalwart a valid TLS certificate in order to handle STARTTLS requests.

Regarding the automatically generated DNS records, Stalwart has no way to know which protocols you're exposing on your reverse proxy and using which port numbers. That's why you need to manually adapt the sample entries for your environment. If a certain entry is not added to the DNS file, it doesn't mean it's not supported in proxied environments, it means that Stalwart cannot confirm that the entry applies to your environment.