docker-mailserver/docs/content/a-mail-server's-101.md

59 lines
6.5 KiB
Markdown
Raw Normal View History

2020-06-09 21:51:01 +00:00
What is a mail server and how does it perform its duty?
Here's an introduction to the field that covers everything you need to know to get started with docker-mailserver.
## Anatomy of a mail server
A mail server is only a part of a [client-server relationship](https://en.wikipedia.org/wiki/Client%E2%80%93server_model) aimed at exchanging information in the form of emails.
This project provides with the server portion, whereas "the" client can be anything from a console, text-only software (eg. [Mutt](https://en.wikipedia.org/wiki/Mutt_(email_client))) to a fully-fledged desktop application (eg. [Mozilla Thunderbird](https://en.wikipedia.org/wiki/Mozilla_Thunderbird), [Microsoft Outlook](https://en.wikipedia.org/wiki/Microsoft_Outlook)…), to a webmail, etc.
2020-06-09 21:54:21 +00:00
Unlike the client side where usually a single program is used, there are many components making up the server. Specialized piece of software handle atomic tasks, such as receiving emails, dropping emails into mailboxes, sending emails to other mail servers, filtering emails, exposing emails to authorized clients, etc.
2020-06-09 21:54:51 +00:00
The docker-mailserver project has made some informed choices about those components and offers a comprehensive platform to run a feature-full mail server.
2020-06-09 21:51:01 +00:00
## Components
The following components are required to create a [complete delivery chain](https://en.wikipedia.org/wiki/Email_agent_(infrastructure)):
2020-06-09 21:55:45 +00:00
- MUA: a [Mail User Agent](https://en.wikipedia.org/wiki/Email_client) is basically any client/program capable of sending emails to arbitrary mail servers; and most of the times, capable of fetching emails from such mail servers.
2020-06-09 21:51:01 +00:00
- MTA: a [Mail Transfer Agent](https://en.wikipedia.org/wiki/Message_transfer_agent) are the so-called "mail servers", as seen from the MUA's perspective. It's a piece of software dedicated to accepting emails from clients. But it also is responsible for transferring emails to other, specialized programs, which may be either MTA (for further transferring) or "final destination" aka. mailboxes.
- MDA: a [Mail Delivery Agent](https://en.wikipedia.org/wiki/Mail_delivery_agent) is responsible for accepting emails and dropping it into its recipient's mailbox, whichever the form.
There may be other moving parts or sub-divisions, but that's the big picture.
docker-mailserver provides you with the following agents:
- MTA: Postfix
- MDA: Dovecot
One important thing to note is that both programs actually handle multiple tasks. For instance, Postfix is both an SMTP server (accepting email) and an MTA (transfering email); Dovecot is both an MDA (delivering emails in mailboxes) and an IMAP server (allowing MUAs to fetch emails from the so-called mail server).
For both Postfix and Dovecot need to be accessible from the outside to act as servers, they expose themselves through TCP ports, which may be secured using different schemes.
## About security, ports…
### SMTP
A MUA sending an email to a [SMTP](https://en.wikipedia.org/wiki/SMTP) server communicates using data packets exchanged over a network that both the client and the server are part of. In the case of docker-mailserver, the server is Postfix. The MUA may be anything, and its submission/request is (most frequently!) performed as [TCP](https://en.wikipedia.org/wiki/Transmission_Control_Protocol) packets sent over the _public_ internet. This exchange of information may, or may not, be secured in order to counter eavesdropping.
**The best practice as of 2020 would be [SMTPS](https://en.wikipedia.org/wiki/SMTPS) over port 465**. It has the server _enforce_ the client into using an encrypted TCP connection, using [TLS](https://en.wikipedia.org/wiki/Transport_Layer_Security) (see [RFC 8314](https://tools.ietf.org/html/rfc8314)). With this setup, the mail server should deny any client attempting at submitting emails in plain text; it should require a TLS-encrypted exchange to exist from the get go (no connection upgrade using an opt-in STARTTLS mechanism). That SMTPS setup uses port 465 and _Implicit_ (aka. enforced) TLS encryption.
Another well-documented, extensively used mail submission setup is SMTP+STARTTLS. It uses _Explicit_ (aka. opportunistic) TLS over port 587, with an opt-in TLS upgrade of the client-to-server connection using using [STARTTLS](https://en.wikipedia.org/wiki/Opportunistic_TLS). With this setup, the mail server should accept unencrypted requests but should automatically respond to the client with an "offer" to upgrade the connection to a TLS-encrypted one; but it also should allow the client to deny that proposal and eventually still accept unencrypted mail exchange (although some servers may eventually deny unencrypted trafic). Overall, this setup requires more configuration and is less secure by design (hence the name "opportunistic"). As of 2020, it is recommended by RFC 8314 for mail servers to support it, but as a to-be-deprecated protocol and to encourage clients to switch to SMTPS.
A final setup exists and is akin SMTP+STARTTLS, but over port 25. That port has historically been reserved specifically for plain text mail exchange. One may upgrade the connection on port 25 to a TLS-encrypted one, but that should be considered a non-normative usage. It's better reserving port 25 for plain text trafic in order to support older clients, and inter-MTA exchange (although obviously non-secure).
### IMAP
A MUA reading emails from an [IMAP](https://en.wikipedia.org/wiki/IMAP) server communicates using data packets exchanged over a network that both the client and the server are part of. In the case of docker-mailserver, the server is Dovecot. The MUA may be anything, and its retrieval request is (most frequently!) performed as [TCP](https://en.wikipedia.org/wiki/Transmission_Control_Protocol) packets sent over the _public_ internet. This exchange of information may, or may not, be secured in order to counter eavesdropping.
As with SMTP (described above), the IMAP protocol may be secured with either: _Implicit_ (enforced) TLS (aka. [IMAPS](https://en.wikipedia.org/wiki/IMAPS), sometimes written IMAP4S); or _Explicit_ (opportunistic) TLS using STARTTLS.
**The best practice as of 2020 would be IMAPS over port 993**, rather than IMAP+STARTTLS over port 143 (see [RFC 8314](https://tools.ietf.org/html/rfc8314)).
### POP3
Similarly to IMAP, POP3 may be secured with either: _Implicit_ (enforced) TLS (aka. POP3S); or _Explicit_ (opportunistic) TLS using STARTTLS.
**The best practice as of 2020 would be [POP3S](https://en.wikipedia.org/wiki/POP3S) over port 995**, rather than [POP3](https://en.wikipedia.org/wiki/POP3)+STARTTLS over port 110 (see [RFC 8314](https://tools.ietf.org/html/rfc8314)).