This how to is created by the Dutch Internet Standards Platform (the organization behind [internet.nl](https://internet.nl)) and is meant to provide practical information and guidance on implementing DANE for SMTP.
DANE is short for "**D**NS-based **A**uthentication of **N**amed **E**ntities" and is described in [RFC 6698](https://tools.ietf.org/html/rfc6698). It offers the option to use the DNSSEC infrastructure to store and sign keys and certificates that are used by TLS. So basically DANE can be used to validate provided certificates based on information (a TLSA signature) published in the DNS zone of a specific domain.
DANE can be used in a web context or in an e-mail context. DANE for web is unfortunately not supported by most browsers and therefore has little added value. DANE for SMTP (which is described in [RFC 7672](https://tools.ietf.org/html/rfc7672) on the other hand, is used increasingly and addresses a couple of shortcomings in SMTP using STARTTLS. DANE for SMTP uses the presence of DANE TLSA records to securely signal TLS support and to publish the means by which SMTP clients can successfully authenticate legitimate SMTP servers. This becomes "opportunistic DANE TLS" (which is better than regular "opportunistic TLS") and creates resistance to downgrade and man-in-the-middle (MITM) attacks.
The use of opportunistic TLS (via STARTTLS) is not without risks:
* Because forcing the use of TLS for all mail servers would break backwards compatibility, SMTP uses opportunistic TLS (via STARTTLS) as a mechanism to enable secure transport between mail servers. However, the fact that STARTTLS is opportunistic means that the initial connection from one mail server to another always starts unencrypted making it vulnerable to man in the middle attacks. If a mail server does not offer the 'STARTTLS capability' during the SMTP handshake (because it was stripped by an attacker), transport of mail occurs over an unencrypted connection.
* By default mail servers do not validate the authenticity of another mail server's certificate; any random certificate is accepted (see [RFC 2487](https://tools.ietf.org/html/rfc2487). This was probably done because there is no user who can act on errors in case they occur. Unfortunately this default behavior introduces the risk of a man in the middle attack; an attacker can offer a false certificate enabling the attacker to decrypt encrypted traffic.
* Sending mail servers can deduce a receiving mail server's ability to use TLS, by the presence of a TLSA record. This means that a connection does not have to start unencrypted (awaiting the STARTTLS capability) but can be encrypted from the start.
* TLSA records can be used to validate the certificate provided by the receiving mail server. This implies that the administrator of a domain 'guarantees' that the TLSA record is always correct and can be used to validate the certificate. Because of this guarantee the sending mail server does not have to fallback to unencrypted mail transport when the offered certificate does not match a single TLSA record. If this is the case the sending mail server can abort the transport and not send the e-mail.
# Guaranteeing a valid TLSA record
Because it is important that there is always a valid TLSA record to make sure mail transport can occur, DANE offers a roll-over mechanism. A roll-over is useful when certificates expire and need to be replaced. Since distributing information via DNS can be a bit slow (depending on the TTL settings), it's important to anticipate a certificate change from a DANE perspective. This can be done by applying a roll-over scheme of which there are two:
* Current + next. This roll-over scheme provides two TLSA records per mail server. One with the fingerprint of the current mail server's certificate (usage type 3), and another with the fingerprint of the future mail server's certificate (usage type 3). The latter can, for example, be determined by using a Certificate Signing Request (CSR).
* Current + issuer. This roll-over scheme provides two TLSA records per mail server. One with the fingerprint of the current mail server's certificate (usage type 3), and another with the fingerprint of a certificate within the current mail server's certificate chain of trust; an intermediate or root certificate.
# Tips and tricks for implementation
This section describes several pionts for attention when implementing DANE for SMTP.
* Purchasing of expensive certificates for mail server has no to little added value for the confidentiality since mail server don't validate certificates by default. Depending on the context there can be other advantages which makes organizations decide to use specific certificates.
* It is recommended to use a certificates public key for generating a TLSA signature (selector type "1") instead of the full certificate (selector type "0"), because this enables the reuse of key materials. Notice that the use of Forward Secrecy decreases the need to use a new key-pair on every occasion.
* An issuer certificate (usage type "2") validates only when the full certificate chain is offered by the receiving mail server.
* Mail servers don't validate certificates and therefore don't have their own certificate store. That's why DANE for SMTP only supports usage type "2" (DANE-TA) and usage type "3" (DANE-EE). Usage type "0" (PKIX-TA) and usage type "1" (PKIX-EE) are not supported.
* Make sure the TTL (time-to-live) of your TLSA records is not too high. This makes it possible to apply changes relatively fast in case of problems. A TTL between 30 minutes (1800) and 1 hour (3600) is recommended.
* The refresh value of your full DNS zone should be in accordance with the TTL setting of your TLSA record, to make sure all name servers give the same information when (after expiration of the TLSA TTL) being queried.
* In case of roll-over scheme "current + issuer", the use of the root certificate is preferred because in some contexts ([PKIoverheid](https://en.wikipedia.org/wiki/PKIoverheid)) this makes it easier to switch supplier / certficate without impacting DANE. (Remember [DigiNotar](https://en.wikipedia.org/wiki/DigiNotar)).
* Roll-over scheme "current + next" gives less flexibility but the highest form of certainty, because of "tight pinning".
* Implement monitoring of your DANE records to be able to detect problems as soon as possible.
This part of the how to describes the steps that should be taken with regard to your outbound e-mail traffic. This enables other parties to use DANE for validating the certificates offered by your e-mail servers.
Generate the DANE SHA-256 hash with `openssl x509 -in /path/to/primary-mailserver.crt -noout -pubkey | openssl pkey -pubin -outform DER | openssl sha256`. This command results in the following output.
For the secondary mail server we generate the DANE SHA-256 hash using
`openssl x509 -in /path/to/secondary-mailserver.crt -noout -pubkey | openssl pkey -pubin -outform DER | openssl sha256`. This command results in the following output.
We use the provided bundle file for generating the DANE hashes belonging to the root certificate. In order to do that, we first split the bundle file into multiple certificates using `cat ca-bundle-file.crt | awk 'BEGIN {c=0;} /BEGIN CERT/{c++} { print > "bundlecert." c ".crt"}'`. In this specific case this results in two files: _bundlecert.1.crt_ and _bundlecert.2.crt_.
For each file we view the **subject** and the **issuer**. We start with the first file using `openssl x509 -in bundlecert.1.crt -noout -subject -issuer`. This results in the following output.
Based on the information of these two certificates, we can conclude that the second certificate (bundlecert.2.crt) is the root certificate; since root certificates are self-signed the **subject** and the **issuer** are the same. The other certificate (bundlecert.1.crt) is an intermediate certificate which is (in this case) signed by the root certificate.
Because we prefer the root certificate to be our roll-over anchor, we generate the DANE SHA-256 hash using `openssl x509 -in bundlecert.2.crt -noout -pubkey | openssl pkey -pubin -outform DER | openssl sha256`. This results in the following output.
Since both certificates for the primary and secondary come from the same Certificate Authority, they both have the same root certificate. So we don't have to repeat this with a different bundle file.
Now that we have the SHA-256 hash, we can construct the DANE roll-over DNS records. We make the following configuration choices:
* Usage field is "**2**"; we generated a DANE hash of the root certificate which is in the chain the chain of trust of the actual leaf certificate (DANE-TA: Trust Anchor Assertion).
* Selector field is "**1**"; because we use the root certificate's public key to generate a DANE hash.
* Matching-type field is "**1**"; because we use SHA-256.
This setting tells Postfix to perform DNS lookups using DNSSEC. This is an important prerequisite for DANE to be effective, since regular DNS lookups can be manipulated. Without DNSSEC support, Postfix cannot use DANE.
By default Postfix uses opportunistic TLS (smtp_tls_security_level = may) which is susceptible to man in the middle attacks. You could tell Postfix to use mandatory TLS (smtp_tls_security_level = encrypt) but this breaks backwards compatibility with mail servers that don't support TLS (and only work with plaintext delivery). However, when Postfix is configured to use the "dane" security level (smtp_tls_security_level = dane) it becomes resistant to man in the middle attacks, since Postfix will connect to other mail servers using "mandatory TLS" when TLSA records are found. If TLSA records are found but are unusable, Postfix won't fallback to plaintext or unauthenticated delivery.
This tells Postfix to perform lookups using DNS. Although this is default behavior it is important to make sure this is configured, since DANE won't be enabled if lookups are performed using a different mechanism.
When applying a DANE roll-over scheme using an "issuer certificate" (an intermediate or root certificate), Postfix must be able to provide the certificates of the used issuer in the chain of trust. Hence this setting.
# Additional information for implementing DANE for SMTP on Exim
**Specifics for this setup**
* Linux Debian Buster (testing)
* Exim 4.91
**Assumptions**
* DNSSEC is used
* Mail server is operational
## Inbound and outbound e-mail traffic
This part of the how to describes some generic steps that should be taken with regard to both inbound and outbound e-mail traffic.
### Install or generate key pair
You can use a commercial or Let's Encrypt certificate, but you can also generate your own key pair by using the provided Exim tools. Use `sudo bash /usr/share/doc/exim4-base/examples/exim-gencert` to generate a key pair.
### Configure TLS
In Exim you should configure TLS by adding the following to **main/03_exim4-config_tlsoptions**
MAIN_TLS_ENABLE = yes
tls_advertise_hosts = *
tls_certificate = /path/to/certificate.crt
tls_privatekey = /path/to/private.key
## Outbound e-mail traffic
This part of the how to describes the steps that should be taken with regard to your outbound e-mail traffic. This enables other parties to use DANE for validating the certificates offered by your e-mail servers.
### DNSSEC validating resolvers
Make sure to configure DNSSEC validating resolvers on the mail server. When using the locale systemd resolver, make sure to add `DNSSEC = yes` to **/etc/systemd/resolved.conf**.
### Configure DNSSEC validation in Exim
In Exim you explicitly need to configure DNSSEC validation by adding `dns_dnssec_ok = 1` to **main/02_exim4-config_options** since some resolvers only validate DNSSEC on request.
### Configure DANE
In order to use DANE, you should tell Exim to check for DANE records when sending e-mail. You can configure DANE validation to be mandatory by adding `hosts_require_dane = *` to **transport/30_exim4-config_remote_smtp**. This means that TLS connections are not accepted when the domain you are trying to send mail to does not have a valid TLSA record. Since this is rather strict and not recommended to be the default, you are probably better of by configuring DANE validation to be additional. This can be done by adding `hosts_try_dane = *` to **transport/30_exim4-config_remote_smtp**.
