2023-01-30 16:27:06 +03:00
package mox
import (
"bytes"
"context"
implement dnssec-awareness throughout code, and dane for incoming/outgoing mail delivery
the vendored dns resolver code is a copy of the go stdlib dns resolver, with
awareness of the "authentic data" (i.e. dnssec secure) added, as well as support
for enhanced dns errors, and looking up tlsa records (for dane). ideally it
would be upstreamed, but the chances seem slim.
dnssec-awareness is added to all packages, e.g. spf, dkim, dmarc, iprev. their
dnssec status is added to the Received message headers for incoming email.
but the main reason to add dnssec was for implementing dane. with dane, the
verification of tls certificates can be done through certificates/public keys
published in dns (in the tlsa records). this only makes sense (is trustworthy)
if those dns records can be verified to be authentic.
mox now applies dane to delivering messages over smtp. mox already implemented
mta-sts for webpki/pkix-verification of certificates against the (large) pool
of CA's, and still enforces those policies when present. but it now also checks
for dane records, and will verify those if present. if dane and mta-sts are
both absent, the regular opportunistic tls with starttls is still done. and the
fallback to plaintext is also still done.
mox also makes it easy to setup dane for incoming deliveries, so other servers
can deliver with dane tls certificate verification. the quickstart now
generates private keys that are used when requesting certificates with acme.
the private keys are pre-generated because they must be static and known during
setup, because their public keys must be published in tlsa records in dns.
autocert would generate private keys on its own, so had to be forked to add the
option to provide the private key when requesting a new certificate. hopefully
upstream will accept the change and we can drop the fork.
with this change, using the quickstart to setup a new mox instance, the checks
at internet.nl result in a 100% score, provided the domain is dnssec-signed and
the network doesn't have any issues.
2023-10-10 13:09:35 +03:00
"crypto"
"crypto/ecdsa"
2023-01-30 16:27:06 +03:00
"crypto/ed25519"
implement dnssec-awareness throughout code, and dane for incoming/outgoing mail delivery
the vendored dns resolver code is a copy of the go stdlib dns resolver, with
awareness of the "authentic data" (i.e. dnssec secure) added, as well as support
for enhanced dns errors, and looking up tlsa records (for dane). ideally it
would be upstreamed, but the chances seem slim.
dnssec-awareness is added to all packages, e.g. spf, dkim, dmarc, iprev. their
dnssec status is added to the Received message headers for incoming email.
but the main reason to add dnssec was for implementing dane. with dane, the
verification of tls certificates can be done through certificates/public keys
published in dns (in the tlsa records). this only makes sense (is trustworthy)
if those dns records can be verified to be authentic.
mox now applies dane to delivering messages over smtp. mox already implemented
mta-sts for webpki/pkix-verification of certificates against the (large) pool
of CA's, and still enforces those policies when present. but it now also checks
for dane records, and will verify those if present. if dane and mta-sts are
both absent, the regular opportunistic tls with starttls is still done. and the
fallback to plaintext is also still done.
mox also makes it easy to setup dane for incoming deliveries, so other servers
can deliver with dane tls certificate verification. the quickstart now
generates private keys that are used when requesting certificates with acme.
the private keys are pre-generated because they must be static and known during
setup, because their public keys must be published in tlsa records in dns.
autocert would generate private keys on its own, so had to be forked to add the
option to provide the private key when requesting a new certificate. hopefully
upstream will accept the change and we can drop the fork.
with this change, using the quickstart to setup a new mox instance, the checks
at internet.nl result in a 100% score, provided the domain is dnssec-signed and
the network doesn't have any issues.
2023-10-10 13:09:35 +03:00
"crypto/elliptic"
cryptorand "crypto/rand"
2023-01-30 16:27:06 +03:00
"crypto/rsa"
"crypto/tls"
"crypto/x509"
2023-12-22 12:34:55 +03:00
"encoding/base64"
2023-01-30 16:27:06 +03:00
"encoding/pem"
"errors"
"fmt"
2023-05-31 15:09:53 +03:00
"io"
2024-02-08 16:49:01 +03:00
"log/slog"
2023-01-30 16:27:06 +03:00
"net"
2023-03-10 18:25:18 +03:00
"net/http"
2023-03-01 00:12:27 +03:00
"net/url"
2023-01-30 16:27:06 +03:00
"os"
change mox to start as root, bind to network sockets, then drop to regular unprivileged mox user
makes it easier to run on bsd's, where you cannot (easily?) let non-root users
bind to ports <1024. starting as root also paves the way for future improvements
with privilege separation.
unfortunately, this requires changes to how you start mox. though mox will help
by automatically fix up dir/file permissions/ownership.
if you start mox from the systemd unit file, you should update it so it starts
as root and adds a few additional capabilities:
# first update the mox binary, then, as root:
./mox config printservice >mox.service
systemctl daemon-reload
systemctl restart mox
journalctl -f -u mox &
# you should see mox start up, with messages about fixing permissions on dirs/files.
if you used the recommended config/ and data/ directory, in a directory just for
mox, and with the mox user called "mox", this should be enough.
if you don't want mox to modify dir/file permissions, set "NoFixPermissions:
true" in mox.conf.
if you named the mox user something else than mox, e.g. "_mox", add "User: _mox"
to mox.conf.
if you created a shared service user as originally suggested, you may want to
get rid of that as it is no longer useful and may get in the way. e.g. if you
had /home/service/mox with a "service" user, that service user can no longer
access any files: only mox and root can.
this also adds scripts for building mox docker images for alpine-supported
platforms.
the "restart" subcommand has been removed. it wasn't all that useful and got in
the way.
and another change: when adding a domain while mtasts isn't enabled, don't add
the per-domain mtasts config, as it would cause failure to add the domain.
based on report from setting up mox on openbsd from mteege.
and based on issue #3. thanks for the feedback!
2023-02-27 14:19:55 +03:00
"os/user"
2023-01-30 16:27:06 +03:00
"path/filepath"
"regexp"
2024-03-05 18:30:38 +03:00
"slices"
2023-01-30 16:27:06 +03:00
"sort"
change mox to start as root, bind to network sockets, then drop to regular unprivileged mox user
makes it easier to run on bsd's, where you cannot (easily?) let non-root users
bind to ports <1024. starting as root also paves the way for future improvements
with privilege separation.
unfortunately, this requires changes to how you start mox. though mox will help
by automatically fix up dir/file permissions/ownership.
if you start mox from the systemd unit file, you should update it so it starts
as root and adds a few additional capabilities:
# first update the mox binary, then, as root:
./mox config printservice >mox.service
systemctl daemon-reload
systemctl restart mox
journalctl -f -u mox &
# you should see mox start up, with messages about fixing permissions on dirs/files.
if you used the recommended config/ and data/ directory, in a directory just for
mox, and with the mox user called "mox", this should be enough.
if you don't want mox to modify dir/file permissions, set "NoFixPermissions:
true" in mox.conf.
if you named the mox user something else than mox, e.g. "_mox", add "User: _mox"
to mox.conf.
if you created a shared service user as originally suggested, you may want to
get rid of that as it is no longer useful and may get in the way. e.g. if you
had /home/service/mox with a "service" user, that service user can no longer
access any files: only mox and root can.
this also adds scripts for building mox docker images for alpine-supported
platforms.
the "restart" subcommand has been removed. it wasn't all that useful and got in
the way.
and another change: when adding a domain while mtasts isn't enabled, don't add
the per-domain mtasts config, as it would cause failure to add the domain.
based on report from setting up mox on openbsd from mteege.
and based on issue #3. thanks for the feedback!
2023-02-27 14:19:55 +03:00
"strconv"
2023-01-30 16:27:06 +03:00
"strings"
"sync"
"time"
"golang.org/x/text/unicode/norm"
implement dnssec-awareness throughout code, and dane for incoming/outgoing mail delivery
the vendored dns resolver code is a copy of the go stdlib dns resolver, with
awareness of the "authentic data" (i.e. dnssec secure) added, as well as support
for enhanced dns errors, and looking up tlsa records (for dane). ideally it
would be upstreamed, but the chances seem slim.
dnssec-awareness is added to all packages, e.g. spf, dkim, dmarc, iprev. their
dnssec status is added to the Received message headers for incoming email.
but the main reason to add dnssec was for implementing dane. with dane, the
verification of tls certificates can be done through certificates/public keys
published in dns (in the tlsa records). this only makes sense (is trustworthy)
if those dns records can be verified to be authentic.
mox now applies dane to delivering messages over smtp. mox already implemented
mta-sts for webpki/pkix-verification of certificates against the (large) pool
of CA's, and still enforces those policies when present. but it now also checks
for dane records, and will verify those if present. if dane and mta-sts are
both absent, the regular opportunistic tls with starttls is still done. and the
fallback to plaintext is also still done.
mox also makes it easy to setup dane for incoming deliveries, so other servers
can deliver with dane tls certificate verification. the quickstart now
generates private keys that are used when requesting certificates with acme.
the private keys are pre-generated because they must be static and known during
setup, because their public keys must be published in tlsa records in dns.
autocert would generate private keys on its own, so had to be forked to add the
option to provide the private key when requesting a new certificate. hopefully
upstream will accept the change and we can drop the fork.
with this change, using the quickstart to setup a new mox instance, the checks
at internet.nl result in a 100% score, provided the domain is dnssec-signed and
the network doesn't have any issues.
2023-10-10 13:09:35 +03:00
"github.com/mjl-/autocert"
2023-01-30 16:27:06 +03:00
"github.com/mjl-/sconf"
"github.com/mjl-/mox/autotls"
"github.com/mjl-/mox/config"
2023-12-05 23:13:57 +03:00
"github.com/mjl-/mox/dkim"
2023-01-30 16:27:06 +03:00
"github.com/mjl-/mox/dns"
2023-12-05 23:13:57 +03:00
"github.com/mjl-/mox/message"
2023-01-30 16:27:06 +03:00
"github.com/mjl-/mox/mlog"
"github.com/mjl-/mox/moxio"
"github.com/mjl-/mox/mtasts"
"github.com/mjl-/mox/smtp"
)
2023-12-05 15:35:58 +03:00
var pkglog = mlog . New ( "mox" , nil )
2023-01-30 16:27:06 +03:00
2023-12-05 23:13:57 +03:00
// Pedantic enables stricter parsing.
var Pedantic bool
2023-01-30 16:27:06 +03:00
// Config paths are set early in program startup. They will point to files in
// the same directory.
var (
ConfigStaticPath string
ConfigDynamicPath string
2023-12-05 15:35:58 +03:00
Conf = Config { Log : map [ string ] slog . Level { "" : slog . LevelError } }
2023-01-30 16:27:06 +03:00
)
add a webapi and webhooks for a simple http/json-based api
for applications to compose/send messages, receive delivery feedback, and
maintain suppression lists.
this is an alternative to applications using a library to compose messages,
submitting those messages using smtp, and monitoring a mailbox with imap for
DSNs, which can be processed into the equivalent of suppression lists. but you
need to know about all these standards/protocols and find libraries. by using
the webapi & webhooks, you just need a http & json library.
unfortunately, there is no standard for these kinds of api, so mox has made up
yet another one...
matching incoming DSNs about deliveries to original outgoing messages requires
keeping history of "retired" messages (delivered from the queue, either
successfully or failed). this can be enabled per account. history is also
useful for debugging deliveries. we now also keep history of each delivery
attempt, accessible while still in the queue, and kept when a message is
retired. the queue webadmin pages now also have pagination, to show potentially
large history.
a queue of webhook calls is now managed too. failures are retried similar to
message deliveries. webhooks can also be saved to the retired list after
completing. also configurable per account.
messages can be sent with a "unique smtp mail from" address. this can only be
used if the domain is configured with a localpart catchall separator such as
"+". when enabled, a queued message gets assigned a random "fromid", which is
added after the separator when sending. when DSNs are returned, they can be
related to previously sent messages based on this fromid. in the future, we can
implement matching on the "envid" used in the smtp dsn extension, or on the
"message-id" of the message. using a fromid can be triggered by authenticating
with a login email address that is configured as enabling fromid.
suppression lists are automatically managed per account. if a delivery attempt
results in certain smtp errors, the destination address is added to the
suppression list. future messages queued for that recipient will immediately
fail without a delivery attempt. suppression lists protect your mail server
reputation.
submitted messages can carry "extra" data through the queue and webhooks for
outgoing deliveries. through webapi as a json object, through smtp submission
as message headers of the form "x-mox-extra-<key>: value".
to make it easy to test webapi/webhooks locally, the "localserve" mode actually
puts messages in the queue. when it's time to deliver, it still won't do a full
delivery attempt, but just delivers to the sender account. unless the recipient
address has a special form, simulating a failure to deliver.
admins now have more control over the queue. "hold rules" can be added to mark
newly queued messages as "on hold", pausing delivery. rules can be about
certain sender or recipient domains/addresses, or apply to all messages pausing
the entire queue. also useful for (local) testing.
new config options have been introduced. they are editable through the admin
and/or account web interfaces.
the webapi http endpoints are enabled for newly generated configs with the
quickstart, and in localserve. existing configurations must explicitly enable
the webapi in mox.conf.
gopherwatch.org was created to dogfood this code. it initially used just the
compose/smtpclient/imapclient mox packages to send messages and process
delivery feedback. it will get a config option to use the mox webapi/webhooks
instead. the gopherwatch code to use webapi/webhook is smaller and simpler, and
developing that shaped development of the mox webapi/webhooks.
for issue #31 by cuu508
2024-04-15 22:49:02 +03:00
var ErrConfig = errors . New ( "config error" )
2023-01-30 16:27:06 +03:00
// Config as used in the code, a processed version of what is in the config file.
//
// Use methods to lookup a domain/account/address in the dynamic configuration.
type Config struct {
Static config . Static // Does not change during the lifetime of a running instance.
logMutex sync . Mutex // For accessing the log levels.
2023-12-05 15:35:58 +03:00
Log map [ string ] slog . Level
2023-01-30 16:27:06 +03:00
dynamicMutex sync . Mutex
Dynamic config . Dynamic // Can only be accessed directly by tests. Use methods on Config for locked access.
dynamicMtime time . Time
DynamicLastCheck time . Time // For use by quickstart only to skip checks.
2023-03-29 11:55:05 +03:00
// From canonical full address (localpart@domain, lower-cased when
// case-insensitive, stripped of catchall separator) to account and address.
// Domains are IDNA names in utf8.
2023-01-30 16:27:06 +03:00
accountDestinations map [ string ] AccountDestination
}
type AccountDestination struct {
2023-03-29 22:11:43 +03:00
Catchall bool // If catchall destination for its domain.
Localpart smtp . Localpart // In original casing as written in config file.
2023-01-30 16:27:06 +03:00
Account string
Destination config . Destination
}
2023-02-06 17:17:46 +03:00
// LogLevelSet sets a new log level for pkg. An empty pkg sets the default log
2023-01-30 16:27:06 +03:00
// value that is used if no explicit log level is configured for a package.
// This change is ephemeral, no config file is changed.
2023-12-05 15:35:58 +03:00
func ( c * Config ) LogLevelSet ( log mlog . Log , pkg string , level slog . Level ) {
2023-01-30 16:27:06 +03:00
c . logMutex . Lock ( )
defer c . logMutex . Unlock ( )
l := c . copyLogLevels ( )
l [ pkg ] = level
c . Log = l
2023-12-05 15:35:58 +03:00
log . Print ( "log level changed" , slog . String ( "pkg" , pkg ) , slog . Any ( "level" , mlog . LevelStrings [ level ] ) )
2023-01-30 16:27:06 +03:00
mlog . SetConfig ( c . Log )
}
2023-02-06 17:17:46 +03:00
// LogLevelRemove removes a configured log level for a package.
2023-12-05 15:35:58 +03:00
func ( c * Config ) LogLevelRemove ( log mlog . Log , pkg string ) {
2023-02-06 17:17:46 +03:00
c . logMutex . Lock ( )
defer c . logMutex . Unlock ( )
l := c . copyLogLevels ( )
delete ( l , pkg )
c . Log = l
2023-12-05 15:35:58 +03:00
log . Print ( "log level cleared" , slog . String ( "pkg" , pkg ) )
2023-02-06 17:17:46 +03:00
mlog . SetConfig ( c . Log )
}
2023-01-30 16:27:06 +03:00
// copyLogLevels returns a copy of c.Log, for modifications.
// must be called with log lock held.
2023-12-05 15:35:58 +03:00
func ( c * Config ) copyLogLevels ( ) map [ string ] slog . Level {
m := map [ string ] slog . Level { }
2023-01-30 16:27:06 +03:00
for pkg , level := range c . Log {
m [ pkg ] = level
}
return m
}
// LogLevels returns a copy of the current log levels.
2023-12-05 15:35:58 +03:00
func ( c * Config ) LogLevels ( ) map [ string ] slog . Level {
2023-01-30 16:27:06 +03:00
c . logMutex . Lock ( )
defer c . logMutex . Unlock ( )
return c . copyLogLevels ( )
}
func ( c * Config ) withDynamicLock ( fn func ( ) ) {
c . dynamicMutex . Lock ( )
defer c . dynamicMutex . Unlock ( )
now := time . Now ( )
if now . Sub ( c . DynamicLastCheck ) > time . Second {
c . DynamicLastCheck = now
if fi , err := os . Stat ( ConfigDynamicPath ) ; err != nil {
2023-12-05 15:35:58 +03:00
pkglog . Errorx ( "stat domains config" , err )
2023-01-30 16:27:06 +03:00
} else if ! fi . ModTime ( ) . Equal ( c . dynamicMtime ) {
if errs := c . loadDynamic ( ) ; len ( errs ) > 0 {
2023-12-05 15:35:58 +03:00
pkglog . Errorx ( "loading domains config" , errs [ 0 ] , slog . Any ( "errors" , errs ) )
2023-01-30 16:27:06 +03:00
} else {
2023-12-05 15:35:58 +03:00
pkglog . Info ( "domains config reloaded" )
2023-01-30 16:27:06 +03:00
c . dynamicMtime = fi . ModTime ( )
}
}
}
fn ( )
}
// must be called with dynamic lock held.
func ( c * Config ) loadDynamic ( ) [ ] error {
2023-12-05 15:35:58 +03:00
d , mtime , accDests , err := ParseDynamicConfig ( context . Background ( ) , pkglog , ConfigDynamicPath , c . Static )
2023-01-30 16:27:06 +03:00
if err != nil {
return err
}
c . Dynamic = d
c . dynamicMtime = mtime
c . accountDestinations = accDests
2023-12-05 15:35:58 +03:00
c . allowACMEHosts ( pkglog , true )
2023-01-30 16:27:06 +03:00
return nil
}
2024-04-18 12:14:24 +03:00
// DynamicConfig returns a shallow copy of the dynamic config. Must not be modified.
func ( c * Config ) DynamicConfig ( ) ( config config . Dynamic ) {
c . withDynamicLock ( func ( ) {
config = c . Dynamic // Shallow copy.
} )
return
}
2023-01-30 16:27:06 +03:00
func ( c * Config ) Domains ( ) ( l [ ] string ) {
c . withDynamicLock ( func ( ) {
for name := range c . Dynamic . Domains {
l = append ( l , name )
}
} )
sort . Slice ( l , func ( i , j int ) bool {
return l [ i ] < l [ j ]
} )
return l
}
func ( c * Config ) Accounts ( ) ( l [ ] string ) {
c . withDynamicLock ( func ( ) {
for name := range c . Dynamic . Accounts {
l = append ( l , name )
}
} )
return
}
2023-03-29 22:11:43 +03:00
// DomainLocalparts returns a mapping of encoded localparts to account names for a
// domain. An empty localpart is a catchall destination for a domain.
func ( c * Config ) DomainLocalparts ( d dns . Domain ) map [ string ] string {
2023-01-30 16:27:06 +03:00
suffix := "@" + d . Name ( )
2023-03-29 22:11:43 +03:00
m := map [ string ] string { }
2023-01-30 16:27:06 +03:00
c . withDynamicLock ( func ( ) {
for addr , ad := range c . accountDestinations {
if strings . HasSuffix ( addr , suffix ) {
2023-03-29 22:11:43 +03:00
if ad . Catchall {
m [ "" ] = ad . Account
} else {
m [ ad . Localpart . String ( ) ] = ad . Account
}
2023-01-30 16:27:06 +03:00
}
}
} )
return m
}
func ( c * Config ) Domain ( d dns . Domain ) ( dom config . Domain , ok bool ) {
c . withDynamicLock ( func ( ) {
dom , ok = c . Dynamic . Domains [ d . Name ( ) ]
} )
return
}
func ( c * Config ) Account ( name string ) ( acc config . Account , ok bool ) {
c . withDynamicLock ( func ( ) {
acc , ok = c . Dynamic . Accounts [ name ]
} )
return
}
func ( c * Config ) AccountDestination ( addr string ) ( accDests AccountDestination , ok bool ) {
c . withDynamicLock ( func ( ) {
accDests , ok = c . accountDestinations [ addr ]
} )
return
}
new feature: when delivering messages from the queue, make it possible to use a "transport"
the default transport is still just "direct delivery", where we connect to the
destination domain's MX servers.
other transports are:
- regular smtp without authentication, this is relaying to a smarthost.
- submission with authentication, e.g. to a third party email sending service.
- direct delivery, but with with connections going through a socks proxy. this
can be helpful if your ip is blocked, you need to get email out, and you have
another IP that isn't blocked.
keep in mind that for all of the above, appropriate SPF/DKIM settings have to
be configured. the "dnscheck" for a domain does a check for any SOCKS IP in the
SPF record. SPF for smtp/submission (ranges? includes?) and any DKIM
requirements cannot really be checked.
which transport is used can be configured through routes. routes can be set on
an account, a domain, or globally. the routes are evaluated in that order, with
the first match selecting the transport. these routes are evaluated for each
delivery attempt. common selection criteria are recipient domain and sender
domain, but also which delivery attempt this is. you could configured mox to
attempt sending through a 3rd party from the 4th attempt onwards.
routes and transports are optional. if no route matches, or an empty/zero
transport is selected, normal direct delivery is done.
we could already "submit" emails with 3rd party accounts with "sendmail". but
we now support more SASL authentication mechanisms with SMTP (not only PLAIN,
but also SCRAM-SHA-256, SCRAM-SHA-1 and CRAM-MD5), which sendmail now also
supports. sendmail will use the most secure mechanism supported by the server,
or the explicitly configured mechanism.
for issue #36 by dmikushin. also based on earlier discussion on hackernews.
2023-06-16 19:38:28 +03:00
func ( c * Config ) Routes ( accountName string , domain dns . Domain ) ( accountRoutes , domainRoutes , globalRoutes [ ] config . Route ) {
c . withDynamicLock ( func ( ) {
acc := c . Dynamic . Accounts [ accountName ]
accountRoutes = acc . Routes
dom := c . Dynamic . Domains [ domain . Name ( ) ]
domainRoutes = dom . Routes
globalRoutes = c . Dynamic . Routes
} )
return
}
2023-12-05 15:35:58 +03:00
func ( c * Config ) allowACMEHosts ( log mlog . Log , checkACMEHosts bool ) {
2023-01-30 16:27:06 +03:00
for _ , l := range c . Static . Listeners {
if l . TLS == nil || l . TLS . ACME == "" {
continue
}
2023-03-01 00:12:27 +03:00
2023-01-30 16:27:06 +03:00
m := c . Static . ACME [ l . TLS . ACME ] . Manager
2023-03-01 00:12:27 +03:00
hostnames := map [ dns . Domain ] struct { } { }
hostnames [ c . Static . HostnameDomain ] = struct { } { }
if l . HostnameDomain . ASCII != "" {
hostnames [ l . HostnameDomain ] = struct { } { }
}
2023-01-30 16:27:06 +03:00
for _ , dom := range c . Dynamic . Domains {
2024-01-26 21:51:23 +03:00
// Do not allow TLS certificates for domains for which we only accept DMARC/TLS
// reports as external party.
if dom . ReportsOnly {
2023-11-14 02:26:18 +03:00
continue
}
2023-02-25 13:28:15 +03:00
if l . AutoconfigHTTPS . Enabled && ! l . AutoconfigHTTPS . NonTLS {
2023-02-03 19:53:45 +03:00
if d , err := dns . ParseDomain ( "autoconfig." + dom . Domain . ASCII ) ; err != nil {
2023-12-05 15:35:58 +03:00
log . Errorx ( "parsing autoconfig domain" , err , slog . Any ( "domain" , dom . Domain ) )
2023-02-03 19:53:45 +03:00
} else {
2023-03-01 00:12:27 +03:00
hostnames [ d ] = struct { } { }
2023-02-03 19:53:45 +03:00
}
2023-01-30 16:27:06 +03:00
}
2023-02-25 13:28:15 +03:00
if l . MTASTSHTTPS . Enabled && dom . MTASTS != nil && ! l . MTASTSHTTPS . NonTLS {
2023-01-30 16:27:06 +03:00
d , err := dns . ParseDomain ( "mta-sts." + dom . Domain . ASCII )
if err != nil {
2023-12-05 15:35:58 +03:00
log . Errorx ( "parsing mta-sts domain" , err , slog . Any ( "domain" , dom . Domain ) )
2023-03-01 00:12:27 +03:00
} else {
hostnames [ d ] = struct { } { }
2023-01-30 16:27:06 +03:00
}
}
assume a dns cname record mail.<domain>, pointing to the hostname of the mail server, for clients to connect to
the autoconfig/autodiscover endpoints, and the printed client settings (in
quickstart, in the admin interface) now all point to the cname record (called
"client settings domain"). it is configurable per domain, and set to
"mail.<domain>" by default. for existing mox installs, the domain can be added
by editing the config file.
this makes it easier for a domain to migrate to another server in the future.
client settings don't have to be updated, the cname can just be changed.
before, the hostname of the mail server was configured in email clients.
migrating away would require changing settings in all clients.
if a client settings domain is configured, a TLS certificate for the name will
be requested through ACME, or must be configured manually.
2023-12-24 13:01:16 +03:00
if dom . ClientSettingsDomain != "" {
hostnames [ dom . ClientSettingsDNSDomain ] = struct { } { }
}
2023-01-30 16:27:06 +03:00
}
2023-03-01 00:12:27 +03:00
if l . WebserverHTTPS . Enabled {
improve webserver, add domain redirects (aliases), add tests and admin page ui to manage the config
- make builtin http handlers serve on specific domains, such as for mta-sts, so
e.g. /.well-known/mta-sts.txt isn't served on all domains.
- add logging of a few more fields in access logging.
- small tweaks/bug fixes in webserver request handling.
- add config option for redirecting entire domains to another (common enough).
- split httpserver metric into two: one for duration until writing header (i.e.
performance of server), another for duration until full response is sent to
client (i.e. performance as perceived by users).
- add admin ui, a new page for managing the configs. after making changes
and hitting "save", the changes take effect immediately. the page itself
doesn't look very well-designed (many input fields, makes it look messy). i
have an idea to improve it (explained in admin.html as todo) by making the
layout look just like the config file. not urgent though.
i've already changed my websites/webapps over.
the idea of adding a webserver is to take away a (the) reason for folks to want
to complicate their mox setup by running an other webserver on the same machine.
i think the current webserver implementation can already serve most common use
cases. with a few more tweaks (feedback needed!) we should be able to get to 95%
of the use cases. the reverse proxy can take care of the remaining 5%.
nevertheless, a next step is still to change the quickstart to make it easier
for folks to run with an existing webserver, with existing tls certs/keys.
that's how this relates to issue #5.
2023-03-02 20:15:54 +03:00
for from := range c . Dynamic . WebDNSDomainRedirects {
hostnames [ from ] = struct { } { }
}
2023-03-01 00:12:27 +03:00
for _ , wh := range c . Dynamic . WebHandlers {
hostnames [ wh . DNSDomain ] = struct { } { }
}
}
2023-08-11 11:13:17 +03:00
public := c . Static . Listeners [ "public" ]
ips := public . IPs
if len ( public . NATIPs ) > 0 {
ips = public . NATIPs
}
if public . IPsNATed {
ips = nil
}
2023-12-05 15:35:58 +03:00
m . SetAllowedHostnames ( log , dns . StrictResolver { Pkg : "autotls" , Log : log . Logger } , hostnames , ips , checkACMEHosts )
2023-01-30 16:27:06 +03:00
}
}
// todo future: write config parsing & writing code that can read a config and remembers the exact tokens including newlines and comments, and can write back a modified file. the goal is to be able to write a config file automatically (after changing fields through the ui), but not loose comments and whitespace, to still get useful diffs for storing the config in a version control system.
// must be called with lock held.
add a webapi and webhooks for a simple http/json-based api
for applications to compose/send messages, receive delivery feedback, and
maintain suppression lists.
this is an alternative to applications using a library to compose messages,
submitting those messages using smtp, and monitoring a mailbox with imap for
DSNs, which can be processed into the equivalent of suppression lists. but you
need to know about all these standards/protocols and find libraries. by using
the webapi & webhooks, you just need a http & json library.
unfortunately, there is no standard for these kinds of api, so mox has made up
yet another one...
matching incoming DSNs about deliveries to original outgoing messages requires
keeping history of "retired" messages (delivered from the queue, either
successfully or failed). this can be enabled per account. history is also
useful for debugging deliveries. we now also keep history of each delivery
attempt, accessible while still in the queue, and kept when a message is
retired. the queue webadmin pages now also have pagination, to show potentially
large history.
a queue of webhook calls is now managed too. failures are retried similar to
message deliveries. webhooks can also be saved to the retired list after
completing. also configurable per account.
messages can be sent with a "unique smtp mail from" address. this can only be
used if the domain is configured with a localpart catchall separator such as
"+". when enabled, a queued message gets assigned a random "fromid", which is
added after the separator when sending. when DSNs are returned, they can be
related to previously sent messages based on this fromid. in the future, we can
implement matching on the "envid" used in the smtp dsn extension, or on the
"message-id" of the message. using a fromid can be triggered by authenticating
with a login email address that is configured as enabling fromid.
suppression lists are automatically managed per account. if a delivery attempt
results in certain smtp errors, the destination address is added to the
suppression list. future messages queued for that recipient will immediately
fail without a delivery attempt. suppression lists protect your mail server
reputation.
submitted messages can carry "extra" data through the queue and webhooks for
outgoing deliveries. through webapi as a json object, through smtp submission
as message headers of the form "x-mox-extra-<key>: value".
to make it easy to test webapi/webhooks locally, the "localserve" mode actually
puts messages in the queue. when it's time to deliver, it still won't do a full
delivery attempt, but just delivers to the sender account. unless the recipient
address has a special form, simulating a failure to deliver.
admins now have more control over the queue. "hold rules" can be added to mark
newly queued messages as "on hold", pausing delivery. rules can be about
certain sender or recipient domains/addresses, or apply to all messages pausing
the entire queue. also useful for (local) testing.
new config options have been introduced. they are editable through the admin
and/or account web interfaces.
the webapi http endpoints are enabled for newly generated configs with the
quickstart, and in localserve. existing configurations must explicitly enable
the webapi in mox.conf.
gopherwatch.org was created to dogfood this code. it initially used just the
compose/smtpclient/imapclient mox packages to send messages and process
delivery feedback. it will get a config option to use the mox webapi/webhooks
instead. the gopherwatch code to use webapi/webhook is smaller and simpler, and
developing that shaped development of the mox webapi/webhooks.
for issue #31 by cuu508
2024-04-15 22:49:02 +03:00
// Returns ErrConfig if the configuration is not valid.
2023-12-05 15:35:58 +03:00
func writeDynamic ( ctx context . Context , log mlog . Log , c config . Dynamic ) error {
accDests , errs := prepareDynamicConfig ( ctx , log , ConfigDynamicPath , Conf . Static , & c )
2023-01-30 16:27:06 +03:00
if len ( errs ) > 0 {
add a webapi and webhooks for a simple http/json-based api
for applications to compose/send messages, receive delivery feedback, and
maintain suppression lists.
this is an alternative to applications using a library to compose messages,
submitting those messages using smtp, and monitoring a mailbox with imap for
DSNs, which can be processed into the equivalent of suppression lists. but you
need to know about all these standards/protocols and find libraries. by using
the webapi & webhooks, you just need a http & json library.
unfortunately, there is no standard for these kinds of api, so mox has made up
yet another one...
matching incoming DSNs about deliveries to original outgoing messages requires
keeping history of "retired" messages (delivered from the queue, either
successfully or failed). this can be enabled per account. history is also
useful for debugging deliveries. we now also keep history of each delivery
attempt, accessible while still in the queue, and kept when a message is
retired. the queue webadmin pages now also have pagination, to show potentially
large history.
a queue of webhook calls is now managed too. failures are retried similar to
message deliveries. webhooks can also be saved to the retired list after
completing. also configurable per account.
messages can be sent with a "unique smtp mail from" address. this can only be
used if the domain is configured with a localpart catchall separator such as
"+". when enabled, a queued message gets assigned a random "fromid", which is
added after the separator when sending. when DSNs are returned, they can be
related to previously sent messages based on this fromid. in the future, we can
implement matching on the "envid" used in the smtp dsn extension, or on the
"message-id" of the message. using a fromid can be triggered by authenticating
with a login email address that is configured as enabling fromid.
suppression lists are automatically managed per account. if a delivery attempt
results in certain smtp errors, the destination address is added to the
suppression list. future messages queued for that recipient will immediately
fail without a delivery attempt. suppression lists protect your mail server
reputation.
submitted messages can carry "extra" data through the queue and webhooks for
outgoing deliveries. through webapi as a json object, through smtp submission
as message headers of the form "x-mox-extra-<key>: value".
to make it easy to test webapi/webhooks locally, the "localserve" mode actually
puts messages in the queue. when it's time to deliver, it still won't do a full
delivery attempt, but just delivers to the sender account. unless the recipient
address has a special form, simulating a failure to deliver.
admins now have more control over the queue. "hold rules" can be added to mark
newly queued messages as "on hold", pausing delivery. rules can be about
certain sender or recipient domains/addresses, or apply to all messages pausing
the entire queue. also useful for (local) testing.
new config options have been introduced. they are editable through the admin
and/or account web interfaces.
the webapi http endpoints are enabled for newly generated configs with the
quickstart, and in localserve. existing configurations must explicitly enable
the webapi in mox.conf.
gopherwatch.org was created to dogfood this code. it initially used just the
compose/smtpclient/imapclient mox packages to send messages and process
delivery feedback. it will get a config option to use the mox webapi/webhooks
instead. the gopherwatch code to use webapi/webhook is smaller and simpler, and
developing that shaped development of the mox webapi/webhooks.
for issue #31 by cuu508
2024-04-15 22:49:02 +03:00
return fmt . Errorf ( "%w: %v" , ErrConfig , errs [ 0 ] )
2023-01-30 16:27:06 +03:00
}
var b bytes . Buffer
err := sconf . Write ( & b , c )
if err != nil {
return err
}
f , err := os . OpenFile ( ConfigDynamicPath , os . O_WRONLY , 0660 )
if err != nil {
return err
}
defer func ( ) {
if f != nil {
2023-02-16 15:22:00 +03:00
err := f . Close ( )
log . Check ( err , "closing file after error" )
2023-01-30 16:27:06 +03:00
}
} ( )
buf := b . Bytes ( )
if _ , err := f . Write ( buf ) ; err != nil {
return fmt . Errorf ( "write domains.conf: %v" , err )
}
if err := f . Truncate ( int64 ( len ( buf ) ) ) ; err != nil {
return fmt . Errorf ( "truncate domains.conf after write: %v" , err )
}
if err := f . Sync ( ) ; err != nil {
return fmt . Errorf ( "sync domains.conf after write: %v" , err )
}
2023-12-05 15:35:58 +03:00
if err := moxio . SyncDir ( log , filepath . Dir ( ConfigDynamicPath ) ) ; err != nil {
2023-01-30 16:27:06 +03:00
return fmt . Errorf ( "sync dir of domains.conf after write: %v" , err )
}
fi , err := f . Stat ( )
if err != nil {
return fmt . Errorf ( "stat after writing domains.conf: %v" , err )
}
2023-02-16 15:22:00 +03:00
if err := f . Close ( ) ; err != nil {
return fmt . Errorf ( "close written domains.conf: %v" , err )
}
f = nil
2023-01-30 16:27:06 +03:00
Conf . dynamicMtime = fi . ModTime ( )
Conf . DynamicLastCheck = time . Now ( )
Conf . Dynamic = c
Conf . accountDestinations = accDests
2023-12-05 15:35:58 +03:00
Conf . allowACMEHosts ( log , true )
2023-01-30 16:27:06 +03:00
return nil
}
// MustLoadConfig loads the config, quitting on errors.
2023-06-16 14:27:27 +03:00
func MustLoadConfig ( doLoadTLSKeyCerts , checkACMEHosts bool ) {
2023-12-05 15:35:58 +03:00
errs := LoadConfig ( context . Background ( ) , pkglog , doLoadTLSKeyCerts , checkACMEHosts )
2023-01-30 16:27:06 +03:00
if len ( errs ) > 1 {
2023-12-05 15:35:58 +03:00
pkglog . Error ( "loading config file: multiple errors" )
2023-01-30 16:27:06 +03:00
for _ , err := range errs {
2023-12-05 15:35:58 +03:00
pkglog . Errorx ( "config error" , err )
2023-01-30 16:27:06 +03:00
}
2023-12-05 15:35:58 +03:00
pkglog . Fatal ( "stopping after multiple config errors" )
2023-01-30 16:27:06 +03:00
} else if len ( errs ) == 1 {
2023-12-05 15:35:58 +03:00
pkglog . Fatalx ( "loading config file" , errs [ 0 ] )
2023-01-30 16:27:06 +03:00
}
}
// LoadConfig attempts to parse and load a config, returning any errors
// encountered.
2023-12-05 15:35:58 +03:00
func LoadConfig ( ctx context . Context , log mlog . Log , doLoadTLSKeyCerts , checkACMEHosts bool ) [ ] error {
2023-03-12 12:38:02 +03:00
Shutdown , ShutdownCancel = context . WithCancel ( context . Background ( ) )
Context , ContextCancel = context . WithCancel ( context . Background ( ) )
2023-12-05 15:35:58 +03:00
c , errs := ParseConfig ( ctx , log , ConfigStaticPath , false , doLoadTLSKeyCerts , checkACMEHosts )
2023-01-30 16:27:06 +03:00
if len ( errs ) > 0 {
return errs
}
mlog . SetConfig ( c . Log )
SetConfig ( c )
return nil
}
// SetConfig sets a new config. Not to be used during normal operation.
func SetConfig ( c * Config ) {
// Cannot just assign *c to Conf, it would copy the mutex.
Conf = Config { c . Static , sync . Mutex { } , c . Log , sync . Mutex { } , c . Dynamic , c . dynamicMtime , c . DynamicLastCheck , c . accountDestinations }
2023-03-10 18:25:18 +03:00
// If we have non-standard CA roots, use them for all HTTPS requests.
if Conf . Static . TLS . CertPool != nil {
http . DefaultTransport . ( * http . Transport ) . TLSClientConfig = & tls . Config {
RootCAs : Conf . Static . TLS . CertPool ,
}
}
2023-03-12 17:16:01 +03:00
2023-12-05 23:13:57 +03:00
SetPedantic ( c . Static . Pedantic )
}
// Set pedantic in all packages.
func SetPedantic ( p bool ) {
dkim . Pedantic = p
dns . Pedantic = p
message . Pedantic = p
smtp . Pedantic = p
Pedantic = p
2023-01-30 16:27:06 +03:00
}
2023-03-04 02:49:02 +03:00
// ParseConfig parses the static config at path p. If checkOnly is true, no changes
2023-06-16 14:27:27 +03:00
// are made, such as registering ACME identities. If doLoadTLSKeyCerts is true,
// the TLS KeyCerts configuration is loaded and checked. This is used during the
2023-03-04 02:49:02 +03:00
// quickstart in the case the user is going to provide their own certificates.
2023-03-06 01:56:02 +03:00
// If checkACMEHosts is true, the hosts allowed for acme are compared with the
// explicitly configured ips we are listening on.
2023-12-05 15:35:58 +03:00
func ParseConfig ( ctx context . Context , log mlog . Log , p string , checkOnly , doLoadTLSKeyCerts , checkACMEHosts bool ) ( c * Config , errs [ ] error ) {
2023-01-30 16:27:06 +03:00
c = & Config {
Static : config . Static {
DataDir : "." ,
} ,
}
f , err := os . Open ( p )
if err != nil {
if os . IsNotExist ( err ) && os . Getenv ( "MOXCONF" ) == "" {
return nil , [ ] error { fmt . Errorf ( "open config file: %v (hint: use mox -config ... or set MOXCONF=...)" , err ) }
}
return nil , [ ] error { fmt . Errorf ( "open config file: %v" , err ) }
}
defer f . Close ( )
if err := sconf . Parse ( f , & c . Static ) ; err != nil {
2023-06-24 11:12:25 +03:00
return nil , [ ] error { fmt . Errorf ( "parsing %s%v" , p , err ) }
2023-01-30 16:27:06 +03:00
}
2023-12-05 15:35:58 +03:00
if xerrs := PrepareStaticConfig ( ctx , log , p , c , checkOnly , doLoadTLSKeyCerts ) ; len ( xerrs ) > 0 {
2023-01-30 16:27:06 +03:00
return nil , xerrs
}
pp := filepath . Join ( filepath . Dir ( p ) , "domains.conf" )
2023-12-05 15:35:58 +03:00
c . Dynamic , c . dynamicMtime , c . accountDestinations , errs = ParseDynamicConfig ( ctx , log , pp , c . Static )
2023-01-30 16:27:06 +03:00
if ! checkOnly {
2023-12-05 15:35:58 +03:00
c . allowACMEHosts ( log , checkACMEHosts )
2023-01-30 16:27:06 +03:00
}
return c , errs
}
// PrepareStaticConfig parses the static config file and prepares data structures
// for starting mox. If checkOnly is set no substantial changes are made, like
// creating an ACME registration.
2023-12-05 15:35:58 +03:00
func PrepareStaticConfig ( ctx context . Context , log mlog . Log , configFile string , conf * Config , checkOnly , doLoadTLSKeyCerts bool ) ( errs [ ] error ) {
2023-01-30 16:27:06 +03:00
addErrorf := func ( format string , args ... any ) {
errs = append ( errs , fmt . Errorf ( format , args ... ) )
}
new feature: when delivering messages from the queue, make it possible to use a "transport"
the default transport is still just "direct delivery", where we connect to the
destination domain's MX servers.
other transports are:
- regular smtp without authentication, this is relaying to a smarthost.
- submission with authentication, e.g. to a third party email sending service.
- direct delivery, but with with connections going through a socks proxy. this
can be helpful if your ip is blocked, you need to get email out, and you have
another IP that isn't blocked.
keep in mind that for all of the above, appropriate SPF/DKIM settings have to
be configured. the "dnscheck" for a domain does a check for any SOCKS IP in the
SPF record. SPF for smtp/submission (ranges? includes?) and any DKIM
requirements cannot really be checked.
which transport is used can be configured through routes. routes can be set on
an account, a domain, or globally. the routes are evaluated in that order, with
the first match selecting the transport. these routes are evaluated for each
delivery attempt. common selection criteria are recipient domain and sender
domain, but also which delivery attempt this is. you could configured mox to
attempt sending through a 3rd party from the 4th attempt onwards.
routes and transports are optional. if no route matches, or an empty/zero
transport is selected, normal direct delivery is done.
we could already "submit" emails with 3rd party accounts with "sendmail". but
we now support more SASL authentication mechanisms with SMTP (not only PLAIN,
but also SCRAM-SHA-256, SCRAM-SHA-1 and CRAM-MD5), which sendmail now also
supports. sendmail will use the most secure mechanism supported by the server,
or the explicitly configured mechanism.
for issue #36 by dmikushin. also based on earlier discussion on hackernews.
2023-06-16 19:38:28 +03:00
c := & conf . Static
2023-01-30 16:27:06 +03:00
// check that mailbox is in unicode NFC normalized form.
checkMailboxNormf := func ( mailbox string , format string , args ... any ) {
s := norm . NFC . String ( mailbox )
if mailbox != s {
msg := fmt . Sprintf ( format , args ... )
addErrorf ( "%s: mailbox %q is not in NFC normalized form, should be %q" , msg , mailbox , s )
}
}
// Post-process logging config.
if logLevel , ok := mlog . Levels [ c . LogLevel ] ; ok {
2023-12-05 15:35:58 +03:00
conf . Log = map [ string ] slog . Level { "" : logLevel }
2023-01-30 16:27:06 +03:00
} else {
addErrorf ( "invalid log level %q" , c . LogLevel )
}
for pkg , s := range c . PackageLogLevels {
if logLevel , ok := mlog . Levels [ s ] ; ok {
new feature: when delivering messages from the queue, make it possible to use a "transport"
the default transport is still just "direct delivery", where we connect to the
destination domain's MX servers.
other transports are:
- regular smtp without authentication, this is relaying to a smarthost.
- submission with authentication, e.g. to a third party email sending service.
- direct delivery, but with with connections going through a socks proxy. this
can be helpful if your ip is blocked, you need to get email out, and you have
another IP that isn't blocked.
keep in mind that for all of the above, appropriate SPF/DKIM settings have to
be configured. the "dnscheck" for a domain does a check for any SOCKS IP in the
SPF record. SPF for smtp/submission (ranges? includes?) and any DKIM
requirements cannot really be checked.
which transport is used can be configured through routes. routes can be set on
an account, a domain, or globally. the routes are evaluated in that order, with
the first match selecting the transport. these routes are evaluated for each
delivery attempt. common selection criteria are recipient domain and sender
domain, but also which delivery attempt this is. you could configured mox to
attempt sending through a 3rd party from the 4th attempt onwards.
routes and transports are optional. if no route matches, or an empty/zero
transport is selected, normal direct delivery is done.
we could already "submit" emails with 3rd party accounts with "sendmail". but
we now support more SASL authentication mechanisms with SMTP (not only PLAIN,
but also SCRAM-SHA-256, SCRAM-SHA-1 and CRAM-MD5), which sendmail now also
supports. sendmail will use the most secure mechanism supported by the server,
or the explicitly configured mechanism.
for issue #36 by dmikushin. also based on earlier discussion on hackernews.
2023-06-16 19:38:28 +03:00
conf . Log [ pkg ] = logLevel
2023-01-30 16:27:06 +03:00
} else {
addErrorf ( "invalid package log level %q" , s )
}
}
change mox to start as root, bind to network sockets, then drop to regular unprivileged mox user
makes it easier to run on bsd's, where you cannot (easily?) let non-root users
bind to ports <1024. starting as root also paves the way for future improvements
with privilege separation.
unfortunately, this requires changes to how you start mox. though mox will help
by automatically fix up dir/file permissions/ownership.
if you start mox from the systemd unit file, you should update it so it starts
as root and adds a few additional capabilities:
# first update the mox binary, then, as root:
./mox config printservice >mox.service
systemctl daemon-reload
systemctl restart mox
journalctl -f -u mox &
# you should see mox start up, with messages about fixing permissions on dirs/files.
if you used the recommended config/ and data/ directory, in a directory just for
mox, and with the mox user called "mox", this should be enough.
if you don't want mox to modify dir/file permissions, set "NoFixPermissions:
true" in mox.conf.
if you named the mox user something else than mox, e.g. "_mox", add "User: _mox"
to mox.conf.
if you created a shared service user as originally suggested, you may want to
get rid of that as it is no longer useful and may get in the way. e.g. if you
had /home/service/mox with a "service" user, that service user can no longer
access any files: only mox and root can.
this also adds scripts for building mox docker images for alpine-supported
platforms.
the "restart" subcommand has been removed. it wasn't all that useful and got in
the way.
and another change: when adding a domain while mtasts isn't enabled, don't add
the per-domain mtasts config, as it would cause failure to add the domain.
based on report from setting up mox on openbsd from mteege.
and based on issue #3. thanks for the feedback!
2023-02-27 14:19:55 +03:00
if c . User == "" {
c . User = "mox"
}
u , err := user . Lookup ( c . User )
make mox compile on windows, without "mox serve" but with working "mox localserve"
getting mox to compile required changing code in only a few places where
package "syscall" was used: for accessing file access times and for umask
handling. an open problem is how to start a process as an unprivileged user on
windows. that's why "mox serve" isn't implemented yet. and just finding a way
to implement it now may not be good enough in the near future: we may want to
starting using a more complete privilege separation approach, with a process
handling sensitive tasks (handling private keys, authentication), where we may
want to pass file descriptors between processes. how would that work on
windows?
anyway, getting mox to compile for windows doesn't mean it works properly on
windows. the largest issue: mox would normally open a file, rename or remove
it, and finally close it. this happens during message delivery. that doesn't
work on windows, the rename/remove would fail because the file is still open.
so this commit swaps many "remove" and "close" calls. renames are a longer
story: message delivery had two ways to deliver: with "consuming" the
(temporary) message file (which would rename it to its final destination), and
without consuming (by hardlinking the file, falling back to copying). the last
delivery to a recipient of a message (and the only one in the common case of a
single recipient) would consume the message, and the earlier recipients would
not. during delivery, the already open message file was used, to parse the
message. we still want to use that open message file, and the caller now stays
responsible for closing it, but we no longer try to rename (consume) the file.
we always hardlink (or copy) during delivery (this works on windows), and the
caller is responsible for closing and removing (in that order) the original
temporary file. this does cost one syscall more. but it makes the delivery code
(responsibilities) a bit simpler.
there is one more obvious issue: the file system path separator. mox already
used the "filepath" package to join paths in many places, but not everywhere.
and it still used strings with slashes for local file access. with this commit,
the code now uses filepath.FromSlash for path strings with slashes, uses
"filepath" in a few more places where it previously didn't. also switches from
"filepath" to regular "path" package when handling mailbox names in a few
places, because those always use forward slashes, regardless of local file
system conventions. windows can handle forward slashes when opening files, so
test code that passes path strings with forward slashes straight to go stdlib
file i/o functions are left unchanged to reduce code churn. the regular
non-test code, or test code that uses path strings in places other than
standard i/o functions, does have the paths converted for consistent paths
(otherwise we would end up with paths with mixed forward/backward slashes in
log messages).
windows cannot dup a listening socket. for "mox localserve", it isn't
important, and we can work around the issue. the current approach for "mox
serve" (forking a process and passing file descriptors of listening sockets on
"privileged" ports) won't work on windows. perhaps it isn't needed on windows,
and any user can listen on "privileged" ports? that would be welcome.
on windows, os.Open cannot open a directory, so we cannot call Sync on it after
message delivery. a cursory internet search indicates that directories cannot
be synced on windows. the story is probably much more nuanced than that, with
long deep technical details/discussions/disagreement/confusion, like on unix.
for "mox localserve" we can get away with making syncdir a no-op.
2023-10-14 11:54:07 +03:00
if err != nil {
change mox to start as root, bind to network sockets, then drop to regular unprivileged mox user
makes it easier to run on bsd's, where you cannot (easily?) let non-root users
bind to ports <1024. starting as root also paves the way for future improvements
with privilege separation.
unfortunately, this requires changes to how you start mox. though mox will help
by automatically fix up dir/file permissions/ownership.
if you start mox from the systemd unit file, you should update it so it starts
as root and adds a few additional capabilities:
# first update the mox binary, then, as root:
./mox config printservice >mox.service
systemctl daemon-reload
systemctl restart mox
journalctl -f -u mox &
# you should see mox start up, with messages about fixing permissions on dirs/files.
if you used the recommended config/ and data/ directory, in a directory just for
mox, and with the mox user called "mox", this should be enough.
if you don't want mox to modify dir/file permissions, set "NoFixPermissions:
true" in mox.conf.
if you named the mox user something else than mox, e.g. "_mox", add "User: _mox"
to mox.conf.
if you created a shared service user as originally suggested, you may want to
get rid of that as it is no longer useful and may get in the way. e.g. if you
had /home/service/mox with a "service" user, that service user can no longer
access any files: only mox and root can.
this also adds scripts for building mox docker images for alpine-supported
platforms.
the "restart" subcommand has been removed. it wasn't all that useful and got in
the way.
and another change: when adding a domain while mtasts isn't enabled, don't add
the per-domain mtasts config, as it would cause failure to add the domain.
based on report from setting up mox on openbsd from mteege.
and based on issue #3. thanks for the feedback!
2023-02-27 14:19:55 +03:00
uid , err := strconv . ParseUint ( c . User , 10 , 32 )
if err != nil {
2023-03-04 02:49:02 +03:00
addErrorf ( "parsing unknown user %s as uid: %v (hint: add user mox with \"useradd -d $PWD mox\" or specify a different username on the quickstart command-line)" , c . User , err )
change mox to start as root, bind to network sockets, then drop to regular unprivileged mox user
makes it easier to run on bsd's, where you cannot (easily?) let non-root users
bind to ports <1024. starting as root also paves the way for future improvements
with privilege separation.
unfortunately, this requires changes to how you start mox. though mox will help
by automatically fix up dir/file permissions/ownership.
if you start mox from the systemd unit file, you should update it so it starts
as root and adds a few additional capabilities:
# first update the mox binary, then, as root:
./mox config printservice >mox.service
systemctl daemon-reload
systemctl restart mox
journalctl -f -u mox &
# you should see mox start up, with messages about fixing permissions on dirs/files.
if you used the recommended config/ and data/ directory, in a directory just for
mox, and with the mox user called "mox", this should be enough.
if you don't want mox to modify dir/file permissions, set "NoFixPermissions:
true" in mox.conf.
if you named the mox user something else than mox, e.g. "_mox", add "User: _mox"
to mox.conf.
if you created a shared service user as originally suggested, you may want to
get rid of that as it is no longer useful and may get in the way. e.g. if you
had /home/service/mox with a "service" user, that service user can no longer
access any files: only mox and root can.
this also adds scripts for building mox docker images for alpine-supported
platforms.
the "restart" subcommand has been removed. it wasn't all that useful and got in
the way.
and another change: when adding a domain while mtasts isn't enabled, don't add
the per-domain mtasts config, as it would cause failure to add the domain.
based on report from setting up mox on openbsd from mteege.
and based on issue #3. thanks for the feedback!
2023-02-27 14:19:55 +03:00
} else {
// We assume the same gid as uid.
c . UID = uint32 ( uid )
c . GID = uint32 ( uid )
}
} else {
if uid , err := strconv . ParseUint ( u . Uid , 10 , 32 ) ; err != nil {
addErrorf ( "parsing uid %s: %v" , u . Uid , err )
} else {
c . UID = uint32 ( uid )
}
if gid , err := strconv . ParseUint ( u . Gid , 10 , 32 ) ; err != nil {
addErrorf ( "parsing gid %s: %v" , u . Gid , err )
} else {
c . GID = uint32 ( gid )
}
}
2023-01-30 16:27:06 +03:00
hostname , err := dns . ParseDomain ( c . Hostname )
if err != nil {
addErrorf ( "parsing hostname: %s" , err )
} else if hostname . Name ( ) != c . Hostname {
2024-01-24 12:48:44 +03:00
addErrorf ( "hostname must be in unicode form %q instead of %q" , hostname . Name ( ) , c . Hostname )
2023-01-30 16:27:06 +03:00
}
c . HostnameDomain = hostname
implement outgoing tls reports
we were already accepting, processing and displaying incoming tls reports. now
we start tracking TLS connection and security-policy-related errors for
outgoing message deliveries as well. we send reports once a day, to the
reporting addresses specified in TLSRPT records (rua) of a policy domain. these
reports are about MTA-STS policies and/or DANE policies, and about
STARTTLS-related failures.
sending reports is enabled by default, but can be disabled through setting
NoOutgoingTLSReports in mox.conf.
only at the end of the implementation process came the realization that the
TLSRPT policy domain for DANE (MX) hosts are separate from the TLSRPT policy
for the recipient domain, and that MTA-STS and DANE TLS/policy results are
typically delivered in separate reports. so MX hosts need their own TLSRPT
policies.
config for the per-host TLSRPT policy should be added to mox.conf for existing
installs, in field HostTLSRPT. it is automatically configured by quickstart for
new installs. with a HostTLSRPT config, the "dns records" and "dns check" admin
pages now suggest the per-host TLSRPT record. by creating that record, you're
requesting TLS reports about your MX host.
gathering all the TLS/policy results is somewhat tricky. the tentacles go
throughout the code. the positive result is that the TLS/policy-related code
had to be cleaned up a bit. for example, the smtpclient TLS modes now reflect
reality better, with independent settings about whether PKIX and/or DANE
verification has to be done, and/or whether verification errors have to be
ignored (e.g. for tls-required: no header). also, cached mtasts policies of
mode "none" are now cleaned up once the MTA-STS DNS record goes away.
2023-11-09 19:40:46 +03:00
if c . HostTLSRPT . Account != "" {
tlsrptLocalpart , err := smtp . ParseLocalpart ( c . HostTLSRPT . Localpart )
if err != nil {
addErrorf ( "invalid localpart %q for host tlsrpt: %v" , c . HostTLSRPT . Localpart , err )
} else if tlsrptLocalpart . IsInternational ( ) {
// Does not appear documented in ../rfc/8460, but similar to DMARC it makes sense
// to keep this ascii-only addresses.
addErrorf ( "host TLSRPT localpart %q is an internationalized address, only conventional ascii-only address allowed for interopability" , tlsrptLocalpart )
}
c . HostTLSRPT . ParsedLocalpart = tlsrptLocalpart
}
implement dnssec-awareness throughout code, and dane for incoming/outgoing mail delivery
the vendored dns resolver code is a copy of the go stdlib dns resolver, with
awareness of the "authentic data" (i.e. dnssec secure) added, as well as support
for enhanced dns errors, and looking up tlsa records (for dane). ideally it
would be upstreamed, but the chances seem slim.
dnssec-awareness is added to all packages, e.g. spf, dkim, dmarc, iprev. their
dnssec status is added to the Received message headers for incoming email.
but the main reason to add dnssec was for implementing dane. with dane, the
verification of tls certificates can be done through certificates/public keys
published in dns (in the tlsa records). this only makes sense (is trustworthy)
if those dns records can be verified to be authentic.
mox now applies dane to delivering messages over smtp. mox already implemented
mta-sts for webpki/pkix-verification of certificates against the (large) pool
of CA's, and still enforces those policies when present. but it now also checks
for dane records, and will verify those if present. if dane and mta-sts are
both absent, the regular opportunistic tls with starttls is still done. and the
fallback to plaintext is also still done.
mox also makes it easy to setup dane for incoming deliveries, so other servers
can deliver with dane tls certificate verification. the quickstart now
generates private keys that are used when requesting certificates with acme.
the private keys are pre-generated because they must be static and known during
setup, because their public keys must be published in tlsa records in dns.
autocert would generate private keys on its own, so had to be forked to add the
option to provide the private key when requesting a new certificate. hopefully
upstream will accept the change and we can drop the fork.
with this change, using the quickstart to setup a new mox instance, the checks
at internet.nl result in a 100% score, provided the domain is dnssec-signed and
the network doesn't have any issues.
2023-10-10 13:09:35 +03:00
// Return private key for host name for use with an ACME. Used to return the same
// private key as pre-generated for use with DANE, with its public key in DNS.
// We only use this key for Listener's that have this ACME configured, and for
// which the effective listener host name (either specific to the listener, or the
// global name) is requested. Other host names can get a fresh private key, they
// don't appear in DANE records.
//
// - run 0: only use listener with explicitly matching host name in listener
// (default quickstart config does not set it).
// - run 1: only look at public listener (and host matching mox host name)
// - run 2: all listeners (and host matching mox host name)
findACMEHostPrivateKey := func ( acmeName , host string , keyType autocert . KeyType , run int ) crypto . Signer {
for listenerName , l := range Conf . Static . Listeners {
if l . TLS == nil || l . TLS . ACME != acmeName {
continue
}
if run == 0 && host != l . HostnameDomain . ASCII {
continue
}
if run == 1 && listenerName != "public" || host != Conf . Static . HostnameDomain . ASCII {
continue
}
switch keyType {
case autocert . KeyRSA2048 :
if len ( l . TLS . HostPrivateRSA2048Keys ) == 0 {
continue
}
return l . TLS . HostPrivateRSA2048Keys [ 0 ]
case autocert . KeyECDSAP256 :
if len ( l . TLS . HostPrivateECDSAP256Keys ) == 0 {
continue
}
return l . TLS . HostPrivateECDSAP256Keys [ 0 ]
default :
return nil
}
}
return nil
}
// Make a function for an autocert.Manager.GetPrivateKey, using findACMEHostPrivateKey.
makeGetPrivateKey := func ( acmeName string ) func ( host string , keyType autocert . KeyType ) ( crypto . Signer , error ) {
return func ( host string , keyType autocert . KeyType ) ( crypto . Signer , error ) {
key := findACMEHostPrivateKey ( acmeName , host , keyType , 0 )
if key == nil {
key = findACMEHostPrivateKey ( acmeName , host , keyType , 1 )
}
if key == nil {
key = findACMEHostPrivateKey ( acmeName , host , keyType , 2 )
}
if key != nil {
2023-12-05 18:06:50 +03:00
log . Debug ( "found existing private key for certificate for host" ,
slog . String ( "acmename" , acmeName ) ,
slog . String ( "host" , host ) ,
slog . Any ( "keytype" , keyType ) )
implement dnssec-awareness throughout code, and dane for incoming/outgoing mail delivery
the vendored dns resolver code is a copy of the go stdlib dns resolver, with
awareness of the "authentic data" (i.e. dnssec secure) added, as well as support
for enhanced dns errors, and looking up tlsa records (for dane). ideally it
would be upstreamed, but the chances seem slim.
dnssec-awareness is added to all packages, e.g. spf, dkim, dmarc, iprev. their
dnssec status is added to the Received message headers for incoming email.
but the main reason to add dnssec was for implementing dane. with dane, the
verification of tls certificates can be done through certificates/public keys
published in dns (in the tlsa records). this only makes sense (is trustworthy)
if those dns records can be verified to be authentic.
mox now applies dane to delivering messages over smtp. mox already implemented
mta-sts for webpki/pkix-verification of certificates against the (large) pool
of CA's, and still enforces those policies when present. but it now also checks
for dane records, and will verify those if present. if dane and mta-sts are
both absent, the regular opportunistic tls with starttls is still done. and the
fallback to plaintext is also still done.
mox also makes it easy to setup dane for incoming deliveries, so other servers
can deliver with dane tls certificate verification. the quickstart now
generates private keys that are used when requesting certificates with acme.
the private keys are pre-generated because they must be static and known during
setup, because their public keys must be published in tlsa records in dns.
autocert would generate private keys on its own, so had to be forked to add the
option to provide the private key when requesting a new certificate. hopefully
upstream will accept the change and we can drop the fork.
with this change, using the quickstart to setup a new mox instance, the checks
at internet.nl result in a 100% score, provided the domain is dnssec-signed and
the network doesn't have any issues.
2023-10-10 13:09:35 +03:00
return key , nil
}
2023-12-05 18:06:50 +03:00
log . Debug ( "generating new private key for certificate for host" ,
slog . String ( "acmename" , acmeName ) ,
slog . String ( "host" , host ) ,
slog . Any ( "keytype" , keyType ) )
implement dnssec-awareness throughout code, and dane for incoming/outgoing mail delivery
the vendored dns resolver code is a copy of the go stdlib dns resolver, with
awareness of the "authentic data" (i.e. dnssec secure) added, as well as support
for enhanced dns errors, and looking up tlsa records (for dane). ideally it
would be upstreamed, but the chances seem slim.
dnssec-awareness is added to all packages, e.g. spf, dkim, dmarc, iprev. their
dnssec status is added to the Received message headers for incoming email.
but the main reason to add dnssec was for implementing dane. with dane, the
verification of tls certificates can be done through certificates/public keys
published in dns (in the tlsa records). this only makes sense (is trustworthy)
if those dns records can be verified to be authentic.
mox now applies dane to delivering messages over smtp. mox already implemented
mta-sts for webpki/pkix-verification of certificates against the (large) pool
of CA's, and still enforces those policies when present. but it now also checks
for dane records, and will verify those if present. if dane and mta-sts are
both absent, the regular opportunistic tls with starttls is still done. and the
fallback to plaintext is also still done.
mox also makes it easy to setup dane for incoming deliveries, so other servers
can deliver with dane tls certificate verification. the quickstart now
generates private keys that are used when requesting certificates with acme.
the private keys are pre-generated because they must be static and known during
setup, because their public keys must be published in tlsa records in dns.
autocert would generate private keys on its own, so had to be forked to add the
option to provide the private key when requesting a new certificate. hopefully
upstream will accept the change and we can drop the fork.
with this change, using the quickstart to setup a new mox instance, the checks
at internet.nl result in a 100% score, provided the domain is dnssec-signed and
the network doesn't have any issues.
2023-10-10 13:09:35 +03:00
switch keyType {
case autocert . KeyRSA2048 :
return rsa . GenerateKey ( cryptorand . Reader , 2048 )
case autocert . KeyECDSAP256 :
return ecdsa . GenerateKey ( elliptic . P256 ( ) , cryptorand . Reader )
default :
return nil , fmt . Errorf ( "unrecognized requested key type %v" , keyType )
}
}
}
2023-01-30 16:27:06 +03:00
for name , acme := range c . ACME {
2023-12-22 12:34:55 +03:00
var eabKeyID string
var eabKey [ ] byte
if acme . ExternalAccountBinding != nil {
eabKeyID = acme . ExternalAccountBinding . KeyID
p := configDirPath ( configFile , acme . ExternalAccountBinding . KeyFile )
buf , err := os . ReadFile ( p )
if err != nil {
addErrorf ( "reading external account binding key for acme provider %q: %s" , name , err )
} else {
dec := make ( [ ] byte , base64 . RawURLEncoding . DecodedLen ( len ( buf ) ) )
n , err := base64 . RawURLEncoding . Decode ( dec , buf )
if err != nil {
addErrorf ( "parsing external account binding key as base64 for acme provider %q: %s" , name , err )
} else {
eabKey = dec [ : n ]
}
}
}
2023-01-30 16:27:06 +03:00
if checkOnly {
continue
}
2023-12-22 12:34:55 +03:00
2023-01-30 16:27:06 +03:00
acmeDir := dataDirPath ( configFile , c . DataDir , "acme" )
os . MkdirAll ( acmeDir , 0770 )
2023-12-22 12:34:55 +03:00
manager , err := autotls . Load ( name , acmeDir , acme . ContactEmail , acme . DirectoryURL , eabKeyID , eabKey , makeGetPrivateKey ( name ) , Shutdown . Done ( ) )
2023-01-30 16:27:06 +03:00
if err != nil {
addErrorf ( "loading ACME identity for %q: %s" , name , err )
}
acme . Manager = manager
2023-12-21 17:16:30 +03:00
// Help configurations from older quickstarts.
if acme . IssuerDomainName == "" && acme . DirectoryURL == "https://acme-v02.api.letsencrypt.org/directory" {
acme . IssuerDomainName = "letsencrypt.org"
}
2023-01-30 16:27:06 +03:00
c . ACME [ name ] = acme
}
var haveUnspecifiedSMTPListener bool
for name , l := range c . Listeners {
if l . Hostname != "" {
d , err := dns . ParseDomain ( l . Hostname )
if err != nil {
addErrorf ( "bad listener hostname %q: %s" , l . Hostname , err )
}
l . HostnameDomain = d
}
if l . TLS != nil {
if l . TLS . ACME != "" && len ( l . TLS . KeyCerts ) != 0 {
addErrorf ( "listener %q: cannot have ACME and static key/certificates" , name )
} else if l . TLS . ACME != "" {
acme , ok := c . ACME [ l . TLS . ACME ]
if ! ok {
addErrorf ( "listener %q: unknown ACME provider %q" , name , l . TLS . ACME )
}
2023-02-22 01:06:11 +03:00
// If only checking or with missing ACME definition, we don't have an acme manager,
// so set an empty tls config to continue.
2023-01-30 16:27:06 +03:00
var tlsconfig * tls . Config
2023-02-22 01:06:11 +03:00
if checkOnly || acme . Manager == nil {
2023-01-30 16:27:06 +03:00
tlsconfig = & tls . Config { }
} else {
tlsconfig = acme . Manager . TLSConfig . Clone ( )
l . TLS . ACMEConfig = acme . Manager . ACMETLSConfig
// SMTP STARTTLS connections are commonly made without SNI, because certificates
implement dnssec-awareness throughout code, and dane for incoming/outgoing mail delivery
the vendored dns resolver code is a copy of the go stdlib dns resolver, with
awareness of the "authentic data" (i.e. dnssec secure) added, as well as support
for enhanced dns errors, and looking up tlsa records (for dane). ideally it
would be upstreamed, but the chances seem slim.
dnssec-awareness is added to all packages, e.g. spf, dkim, dmarc, iprev. their
dnssec status is added to the Received message headers for incoming email.
but the main reason to add dnssec was for implementing dane. with dane, the
verification of tls certificates can be done through certificates/public keys
published in dns (in the tlsa records). this only makes sense (is trustworthy)
if those dns records can be verified to be authentic.
mox now applies dane to delivering messages over smtp. mox already implemented
mta-sts for webpki/pkix-verification of certificates against the (large) pool
of CA's, and still enforces those policies when present. but it now also checks
for dane records, and will verify those if present. if dane and mta-sts are
both absent, the regular opportunistic tls with starttls is still done. and the
fallback to plaintext is also still done.
mox also makes it easy to setup dane for incoming deliveries, so other servers
can deliver with dane tls certificate verification. the quickstart now
generates private keys that are used when requesting certificates with acme.
the private keys are pre-generated because they must be static and known during
setup, because their public keys must be published in tlsa records in dns.
autocert would generate private keys on its own, so had to be forked to add the
option to provide the private key when requesting a new certificate. hopefully
upstream will accept the change and we can drop the fork.
with this change, using the quickstart to setup a new mox instance, the checks
at internet.nl result in a 100% score, provided the domain is dnssec-signed and
the network doesn't have any issues.
2023-10-10 13:09:35 +03:00
// often aren't verified.
2023-01-30 16:27:06 +03:00
hostname := c . HostnameDomain
if l . Hostname != "" {
hostname = l . HostnameDomain
}
getCert := tlsconfig . GetCertificate
tlsconfig . GetCertificate = func ( hello * tls . ClientHelloInfo ) ( * tls . Certificate , error ) {
if hello . ServerName == "" {
hello . ServerName = hostname . ASCII
}
return getCert ( hello )
}
}
l . TLS . Config = tlsconfig
} else if len ( l . TLS . KeyCerts ) != 0 {
2023-06-16 14:27:27 +03:00
if doLoadTLSKeyCerts {
2023-03-04 02:49:02 +03:00
if err := loadTLSKeyCerts ( configFile , "listener " + name , l . TLS ) ; err != nil {
addErrorf ( "%w" , err )
}
2023-01-30 16:27:06 +03:00
}
} else {
addErrorf ( "listener %q: cannot have TLS config without ACME and without static keys/certificates" , name )
}
implement dnssec-awareness throughout code, and dane for incoming/outgoing mail delivery
the vendored dns resolver code is a copy of the go stdlib dns resolver, with
awareness of the "authentic data" (i.e. dnssec secure) added, as well as support
for enhanced dns errors, and looking up tlsa records (for dane). ideally it
would be upstreamed, but the chances seem slim.
dnssec-awareness is added to all packages, e.g. spf, dkim, dmarc, iprev. their
dnssec status is added to the Received message headers for incoming email.
but the main reason to add dnssec was for implementing dane. with dane, the
verification of tls certificates can be done through certificates/public keys
published in dns (in the tlsa records). this only makes sense (is trustworthy)
if those dns records can be verified to be authentic.
mox now applies dane to delivering messages over smtp. mox already implemented
mta-sts for webpki/pkix-verification of certificates against the (large) pool
of CA's, and still enforces those policies when present. but it now also checks
for dane records, and will verify those if present. if dane and mta-sts are
both absent, the regular opportunistic tls with starttls is still done. and the
fallback to plaintext is also still done.
mox also makes it easy to setup dane for incoming deliveries, so other servers
can deliver with dane tls certificate verification. the quickstart now
generates private keys that are used when requesting certificates with acme.
the private keys are pre-generated because they must be static and known during
setup, because their public keys must be published in tlsa records in dns.
autocert would generate private keys on its own, so had to be forked to add the
option to provide the private key when requesting a new certificate. hopefully
upstream will accept the change and we can drop the fork.
with this change, using the quickstart to setup a new mox instance, the checks
at internet.nl result in a 100% score, provided the domain is dnssec-signed and
the network doesn't have any issues.
2023-10-10 13:09:35 +03:00
for _ , privKeyFile := range l . TLS . HostPrivateKeyFiles {
keyPath := configDirPath ( configFile , privKeyFile )
privKey , err := loadPrivateKeyFile ( keyPath )
if err != nil {
addErrorf ( "listener %q: parsing host private key for DANE and ACME certificates: %v" , name , err )
continue
}
switch k := privKey . ( type ) {
case * rsa . PrivateKey :
if k . N . BitLen ( ) != 2048 {
2023-12-05 18:06:50 +03:00
log . Error ( "need rsa key with 2048 bits, for host private key for DANE/ACME certificates, ignoring" ,
slog . String ( "listener" , name ) ,
slog . String ( "file" , keyPath ) ,
slog . Int ( "bits" , k . N . BitLen ( ) ) )
implement dnssec-awareness throughout code, and dane for incoming/outgoing mail delivery
the vendored dns resolver code is a copy of the go stdlib dns resolver, with
awareness of the "authentic data" (i.e. dnssec secure) added, as well as support
for enhanced dns errors, and looking up tlsa records (for dane). ideally it
would be upstreamed, but the chances seem slim.
dnssec-awareness is added to all packages, e.g. spf, dkim, dmarc, iprev. their
dnssec status is added to the Received message headers for incoming email.
but the main reason to add dnssec was for implementing dane. with dane, the
verification of tls certificates can be done through certificates/public keys
published in dns (in the tlsa records). this only makes sense (is trustworthy)
if those dns records can be verified to be authentic.
mox now applies dane to delivering messages over smtp. mox already implemented
mta-sts for webpki/pkix-verification of certificates against the (large) pool
of CA's, and still enforces those policies when present. but it now also checks
for dane records, and will verify those if present. if dane and mta-sts are
both absent, the regular opportunistic tls with starttls is still done. and the
fallback to plaintext is also still done.
mox also makes it easy to setup dane for incoming deliveries, so other servers
can deliver with dane tls certificate verification. the quickstart now
generates private keys that are used when requesting certificates with acme.
the private keys are pre-generated because they must be static and known during
setup, because their public keys must be published in tlsa records in dns.
autocert would generate private keys on its own, so had to be forked to add the
option to provide the private key when requesting a new certificate. hopefully
upstream will accept the change and we can drop the fork.
with this change, using the quickstart to setup a new mox instance, the checks
at internet.nl result in a 100% score, provided the domain is dnssec-signed and
the network doesn't have any issues.
2023-10-10 13:09:35 +03:00
continue
}
l . TLS . HostPrivateRSA2048Keys = append ( l . TLS . HostPrivateRSA2048Keys , k )
case * ecdsa . PrivateKey :
if k . Curve != elliptic . P256 ( ) {
2023-12-05 15:35:58 +03:00
log . Error ( "unrecognized ecdsa curve for host private key for DANE/ACME certificates, ignoring" , slog . String ( "listener" , name ) , slog . String ( "file" , keyPath ) )
implement dnssec-awareness throughout code, and dane for incoming/outgoing mail delivery
the vendored dns resolver code is a copy of the go stdlib dns resolver, with
awareness of the "authentic data" (i.e. dnssec secure) added, as well as support
for enhanced dns errors, and looking up tlsa records (for dane). ideally it
would be upstreamed, but the chances seem slim.
dnssec-awareness is added to all packages, e.g. spf, dkim, dmarc, iprev. their
dnssec status is added to the Received message headers for incoming email.
but the main reason to add dnssec was for implementing dane. with dane, the
verification of tls certificates can be done through certificates/public keys
published in dns (in the tlsa records). this only makes sense (is trustworthy)
if those dns records can be verified to be authentic.
mox now applies dane to delivering messages over smtp. mox already implemented
mta-sts for webpki/pkix-verification of certificates against the (large) pool
of CA's, and still enforces those policies when present. but it now also checks
for dane records, and will verify those if present. if dane and mta-sts are
both absent, the regular opportunistic tls with starttls is still done. and the
fallback to plaintext is also still done.
mox also makes it easy to setup dane for incoming deliveries, so other servers
can deliver with dane tls certificate verification. the quickstart now
generates private keys that are used when requesting certificates with acme.
the private keys are pre-generated because they must be static and known during
setup, because their public keys must be published in tlsa records in dns.
autocert would generate private keys on its own, so had to be forked to add the
option to provide the private key when requesting a new certificate. hopefully
upstream will accept the change and we can drop the fork.
with this change, using the quickstart to setup a new mox instance, the checks
at internet.nl result in a 100% score, provided the domain is dnssec-signed and
the network doesn't have any issues.
2023-10-10 13:09:35 +03:00
continue
}
l . TLS . HostPrivateECDSAP256Keys = append ( l . TLS . HostPrivateECDSAP256Keys , k )
default :
2023-12-05 18:06:50 +03:00
log . Error ( "unrecognized key type for host private key for DANE/ACME certificates, ignoring" ,
slog . String ( "listener" , name ) ,
slog . String ( "file" , keyPath ) ,
slog . String ( "keytype" , fmt . Sprintf ( "%T" , privKey ) ) )
implement dnssec-awareness throughout code, and dane for incoming/outgoing mail delivery
the vendored dns resolver code is a copy of the go stdlib dns resolver, with
awareness of the "authentic data" (i.e. dnssec secure) added, as well as support
for enhanced dns errors, and looking up tlsa records (for dane). ideally it
would be upstreamed, but the chances seem slim.
dnssec-awareness is added to all packages, e.g. spf, dkim, dmarc, iprev. their
dnssec status is added to the Received message headers for incoming email.
but the main reason to add dnssec was for implementing dane. with dane, the
verification of tls certificates can be done through certificates/public keys
published in dns (in the tlsa records). this only makes sense (is trustworthy)
if those dns records can be verified to be authentic.
mox now applies dane to delivering messages over smtp. mox already implemented
mta-sts for webpki/pkix-verification of certificates against the (large) pool
of CA's, and still enforces those policies when present. but it now also checks
for dane records, and will verify those if present. if dane and mta-sts are
both absent, the regular opportunistic tls with starttls is still done. and the
fallback to plaintext is also still done.
mox also makes it easy to setup dane for incoming deliveries, so other servers
can deliver with dane tls certificate verification. the quickstart now
generates private keys that are used when requesting certificates with acme.
the private keys are pre-generated because they must be static and known during
setup, because their public keys must be published in tlsa records in dns.
autocert would generate private keys on its own, so had to be forked to add the
option to provide the private key when requesting a new certificate. hopefully
upstream will accept the change and we can drop the fork.
with this change, using the quickstart to setup a new mox instance, the checks
at internet.nl result in a 100% score, provided the domain is dnssec-signed and
the network doesn't have any issues.
2023-10-10 13:09:35 +03:00
continue
}
}
if l . TLS . ACME != "" && ( len ( l . TLS . HostPrivateRSA2048Keys ) == 0 ) != ( len ( l . TLS . HostPrivateECDSAP256Keys ) == 0 ) {
log . Error ( "warning: uncommon configuration with either only an RSA 2048 or ECDSA P256 host private key for DANE/ACME certificates; this ACME implementation can retrieve certificates for both type of keys, it is recommended to set either both or none; continuing" )
}
2023-01-30 16:27:06 +03:00
// TLS 1.2 was introduced in 2008. TLS <1.2 was deprecated by ../rfc/8996:31 and ../rfc/8997:66 in 2021.
var minVersion uint16 = tls . VersionTLS12
if l . TLS . MinVersion != "" {
versions := map [ string ] uint16 {
"TLSv1.0" : tls . VersionTLS10 ,
"TLSv1.1" : tls . VersionTLS11 ,
"TLSv1.2" : tls . VersionTLS12 ,
"TLSv1.3" : tls . VersionTLS13 ,
}
v , ok := versions [ l . TLS . MinVersion ]
if ! ok {
addErrorf ( "listener %q: unknown TLS mininum version %q" , name , l . TLS . MinVersion )
}
minVersion = v
}
if l . TLS . Config != nil {
l . TLS . Config . MinVersion = minVersion
}
if l . TLS . ACMEConfig != nil {
l . TLS . ACMEConfig . MinVersion = minVersion
}
2023-02-27 23:42:27 +03:00
} else {
var needsTLS [ ] string
needtls := func ( s string , v bool ) {
if v {
needsTLS = append ( needsTLS , s )
}
}
needtls ( "IMAPS" , l . IMAPS . Enabled )
needtls ( "SMTP" , l . SMTP . Enabled && ! l . SMTP . NoSTARTTLS )
needtls ( "Submissions" , l . Submissions . Enabled )
needtls ( "Submission" , l . Submission . Enabled && ! l . Submission . NoRequireSTARTTLS )
needtls ( "AccountHTTPS" , l . AccountHTTPS . Enabled )
needtls ( "AdminHTTPS" , l . AdminHTTPS . Enabled )
needtls ( "AutoconfigHTTPS" , l . AutoconfigHTTPS . Enabled && ! l . AutoconfigHTTPS . NonTLS )
needtls ( "MTASTSHTTPS" , l . MTASTSHTTPS . Enabled && ! l . MTASTSHTTPS . NonTLS )
2023-03-01 00:12:27 +03:00
needtls ( "WebserverHTTPS" , l . WebserverHTTPS . Enabled )
2023-02-27 23:42:27 +03:00
if len ( needsTLS ) > 0 {
addErrorf ( "listener %q does not specify tls config, but requires tls for %s" , name , strings . Join ( needsTLS , ", " ) )
}
2023-01-30 16:27:06 +03:00
}
2023-02-25 13:28:15 +03:00
if l . AutoconfigHTTPS . Enabled && l . MTASTSHTTPS . Enabled && l . AutoconfigHTTPS . Port == l . MTASTSHTTPS . Port && l . AutoconfigHTTPS . NonTLS != l . MTASTSHTTPS . NonTLS {
addErrorf ( "listener %q tries to enable autoconfig and mta-sts enabled on same port but with both http and https" , name )
}
2023-01-30 16:27:06 +03:00
if l . SMTP . Enabled {
if len ( l . IPs ) == 0 {
haveUnspecifiedSMTPListener = true
}
for _ , ipstr := range l . IPs {
ip := net . ParseIP ( ipstr )
if ip == nil {
addErrorf ( "listener %q has invalid IP %q" , name , ipstr )
continue
}
if ip . IsUnspecified ( ) {
haveUnspecifiedSMTPListener = true
break
}
if len ( c . SpecifiedSMTPListenIPs ) >= 2 {
haveUnspecifiedSMTPListener = true
} else if len ( c . SpecifiedSMTPListenIPs ) > 0 && ( c . SpecifiedSMTPListenIPs [ 0 ] . To4 ( ) == nil ) == ( ip . To4 ( ) == nil ) {
haveUnspecifiedSMTPListener = true
} else {
c . SpecifiedSMTPListenIPs = append ( c . SpecifiedSMTPListenIPs , ip )
}
}
}
for _ , s := range l . SMTP . DNSBLs {
d , err := dns . ParseDomain ( s )
if err != nil {
addErrorf ( "listener %q has invalid DNSBL zone %q" , name , s )
continue
}
l . SMTP . DNSBLZones = append ( l . SMTP . DNSBLZones , d )
}
2023-08-11 11:13:17 +03:00
if l . IPsNATed && len ( l . NATIPs ) > 0 {
addErrorf ( "listener %q has both IPsNATed and NATIPs (remove deprecated IPsNATed)" , name )
}
for _ , ipstr := range l . NATIPs {
ip := net . ParseIP ( ipstr )
if ip == nil {
addErrorf ( "listener %q has invalid ip %q" , name , ipstr )
} else if ip . IsUnspecified ( ) || ip . IsLoopback ( ) {
addErrorf ( "listener %q has NAT ip that is the unspecified or loopback address %s" , name , ipstr )
}
}
2023-03-12 13:52:15 +03:00
checkPath := func ( kind string , enabled bool , path string ) {
if enabled && path != "" && ! strings . HasPrefix ( path , "/" ) {
addErrorf ( "listener %q has %s with path %q that must start with a slash" , name , kind , path )
}
}
checkPath ( "AccountHTTP" , l . AccountHTTP . Enabled , l . AccountHTTP . Path )
checkPath ( "AccountHTTPS" , l . AccountHTTPS . Enabled , l . AccountHTTPS . Path )
checkPath ( "AdminHTTP" , l . AdminHTTP . Enabled , l . AdminHTTP . Path )
checkPath ( "AdminHTTPS" , l . AdminHTTPS . Enabled , l . AdminHTTPS . Path )
2023-01-30 16:27:06 +03:00
c . Listeners [ name ] = l
}
if haveUnspecifiedSMTPListener {
c . SpecifiedSMTPListenIPs = nil
}
2023-08-09 10:31:23 +03:00
var zerouse config . SpecialUseMailboxes
if len ( c . DefaultMailboxes ) > 0 && ( c . InitialMailboxes . SpecialUse != zerouse || len ( c . InitialMailboxes . Regular ) > 0 ) {
addErrorf ( "cannot have both DefaultMailboxes and InitialMailboxes" )
}
// DefaultMailboxes is deprecated.
2023-01-30 16:27:06 +03:00
for _ , mb := range c . DefaultMailboxes {
checkMailboxNormf ( mb , "default mailbox" )
}
2023-08-09 10:31:23 +03:00
checkSpecialUseMailbox := func ( nameOpt string ) {
if nameOpt != "" {
checkMailboxNormf ( nameOpt , "special-use initial mailbox" )
if strings . EqualFold ( nameOpt , "inbox" ) {
addErrorf ( "initial mailbox cannot be set to Inbox (Inbox is always created)" )
}
}
}
checkSpecialUseMailbox ( c . InitialMailboxes . SpecialUse . Archive )
checkSpecialUseMailbox ( c . InitialMailboxes . SpecialUse . Draft )
checkSpecialUseMailbox ( c . InitialMailboxes . SpecialUse . Junk )
checkSpecialUseMailbox ( c . InitialMailboxes . SpecialUse . Sent )
checkSpecialUseMailbox ( c . InitialMailboxes . SpecialUse . Trash )
for _ , name := range c . InitialMailboxes . Regular {
checkMailboxNormf ( name , "regular initial mailbox" )
if strings . EqualFold ( name , "inbox" ) {
addErrorf ( "initial regular mailbox cannot be set to Inbox (Inbox is always created)" )
}
}
2023-01-30 16:27:06 +03:00
new feature: when delivering messages from the queue, make it possible to use a "transport"
the default transport is still just "direct delivery", where we connect to the
destination domain's MX servers.
other transports are:
- regular smtp without authentication, this is relaying to a smarthost.
- submission with authentication, e.g. to a third party email sending service.
- direct delivery, but with with connections going through a socks proxy. this
can be helpful if your ip is blocked, you need to get email out, and you have
another IP that isn't blocked.
keep in mind that for all of the above, appropriate SPF/DKIM settings have to
be configured. the "dnscheck" for a domain does a check for any SOCKS IP in the
SPF record. SPF for smtp/submission (ranges? includes?) and any DKIM
requirements cannot really be checked.
which transport is used can be configured through routes. routes can be set on
an account, a domain, or globally. the routes are evaluated in that order, with
the first match selecting the transport. these routes are evaluated for each
delivery attempt. common selection criteria are recipient domain and sender
domain, but also which delivery attempt this is. you could configured mox to
attempt sending through a 3rd party from the 4th attempt onwards.
routes and transports are optional. if no route matches, or an empty/zero
transport is selected, normal direct delivery is done.
we could already "submit" emails with 3rd party accounts with "sendmail". but
we now support more SASL authentication mechanisms with SMTP (not only PLAIN,
but also SCRAM-SHA-256, SCRAM-SHA-1 and CRAM-MD5), which sendmail now also
supports. sendmail will use the most secure mechanism supported by the server,
or the explicitly configured mechanism.
for issue #36 by dmikushin. also based on earlier discussion on hackernews.
2023-06-16 19:38:28 +03:00
checkTransportSMTP := func ( name string , isTLS bool , t * config . TransportSMTP ) {
var err error
t . DNSHost , err = dns . ParseDomain ( t . Host )
if err != nil {
addErrorf ( "transport %s: bad host %s: %v" , name , t . Host , err )
}
if isTLS && t . STARTTLSInsecureSkipVerify {
addErrorf ( "transport %s: cannot have STARTTLSInsecureSkipVerify with immediate TLS" )
}
if isTLS && t . NoSTARTTLS {
addErrorf ( "transport %s: cannot have NoSTARTTLS with immediate TLS" )
}
if t . Auth == nil {
return
}
seen := map [ string ] bool { }
for _ , m := range t . Auth . Mechanisms {
if seen [ m ] {
addErrorf ( "transport %s: duplicate authentication mechanism %s" , name , m )
}
seen [ m ] = true
switch m {
2023-12-24 01:07:21 +03:00
case "SCRAM-SHA-256-PLUS" :
new feature: when delivering messages from the queue, make it possible to use a "transport"
the default transport is still just "direct delivery", where we connect to the
destination domain's MX servers.
other transports are:
- regular smtp without authentication, this is relaying to a smarthost.
- submission with authentication, e.g. to a third party email sending service.
- direct delivery, but with with connections going through a socks proxy. this
can be helpful if your ip is blocked, you need to get email out, and you have
another IP that isn't blocked.
keep in mind that for all of the above, appropriate SPF/DKIM settings have to
be configured. the "dnscheck" for a domain does a check for any SOCKS IP in the
SPF record. SPF for smtp/submission (ranges? includes?) and any DKIM
requirements cannot really be checked.
which transport is used can be configured through routes. routes can be set on
an account, a domain, or globally. the routes are evaluated in that order, with
the first match selecting the transport. these routes are evaluated for each
delivery attempt. common selection criteria are recipient domain and sender
domain, but also which delivery attempt this is. you could configured mox to
attempt sending through a 3rd party from the 4th attempt onwards.
routes and transports are optional. if no route matches, or an empty/zero
transport is selected, normal direct delivery is done.
we could already "submit" emails with 3rd party accounts with "sendmail". but
we now support more SASL authentication mechanisms with SMTP (not only PLAIN,
but also SCRAM-SHA-256, SCRAM-SHA-1 and CRAM-MD5), which sendmail now also
supports. sendmail will use the most secure mechanism supported by the server,
or the explicitly configured mechanism.
for issue #36 by dmikushin. also based on earlier discussion on hackernews.
2023-06-16 19:38:28 +03:00
case "SCRAM-SHA-256" :
2023-12-24 01:07:21 +03:00
case "SCRAM-SHA-1-PLUS" :
new feature: when delivering messages from the queue, make it possible to use a "transport"
the default transport is still just "direct delivery", where we connect to the
destination domain's MX servers.
other transports are:
- regular smtp without authentication, this is relaying to a smarthost.
- submission with authentication, e.g. to a third party email sending service.
- direct delivery, but with with connections going through a socks proxy. this
can be helpful if your ip is blocked, you need to get email out, and you have
another IP that isn't blocked.
keep in mind that for all of the above, appropriate SPF/DKIM settings have to
be configured. the "dnscheck" for a domain does a check for any SOCKS IP in the
SPF record. SPF for smtp/submission (ranges? includes?) and any DKIM
requirements cannot really be checked.
which transport is used can be configured through routes. routes can be set on
an account, a domain, or globally. the routes are evaluated in that order, with
the first match selecting the transport. these routes are evaluated for each
delivery attempt. common selection criteria are recipient domain and sender
domain, but also which delivery attempt this is. you could configured mox to
attempt sending through a 3rd party from the 4th attempt onwards.
routes and transports are optional. if no route matches, or an empty/zero
transport is selected, normal direct delivery is done.
we could already "submit" emails with 3rd party accounts with "sendmail". but
we now support more SASL authentication mechanisms with SMTP (not only PLAIN,
but also SCRAM-SHA-256, SCRAM-SHA-1 and CRAM-MD5), which sendmail now also
supports. sendmail will use the most secure mechanism supported by the server,
or the explicitly configured mechanism.
for issue #36 by dmikushin. also based on earlier discussion on hackernews.
2023-06-16 19:38:28 +03:00
case "SCRAM-SHA-1" :
case "CRAM-MD5" :
case "PLAIN" :
default :
addErrorf ( "transport %s: unknown authentication mechanism %s" , name , m )
}
}
t . Auth . EffectiveMechanisms = t . Auth . Mechanisms
if len ( t . Auth . EffectiveMechanisms ) == 0 {
2023-12-24 01:07:21 +03:00
t . Auth . EffectiveMechanisms = [ ] string { "SCRAM-SHA-256-PLUS" , "SCRAM-SHA-256" , "SCRAM-SHA-1-PLUS" , "SCRAM-SHA-1" , "CRAM-MD5" }
new feature: when delivering messages from the queue, make it possible to use a "transport"
the default transport is still just "direct delivery", where we connect to the
destination domain's MX servers.
other transports are:
- regular smtp without authentication, this is relaying to a smarthost.
- submission with authentication, e.g. to a third party email sending service.
- direct delivery, but with with connections going through a socks proxy. this
can be helpful if your ip is blocked, you need to get email out, and you have
another IP that isn't blocked.
keep in mind that for all of the above, appropriate SPF/DKIM settings have to
be configured. the "dnscheck" for a domain does a check for any SOCKS IP in the
SPF record. SPF for smtp/submission (ranges? includes?) and any DKIM
requirements cannot really be checked.
which transport is used can be configured through routes. routes can be set on
an account, a domain, or globally. the routes are evaluated in that order, with
the first match selecting the transport. these routes are evaluated for each
delivery attempt. common selection criteria are recipient domain and sender
domain, but also which delivery attempt this is. you could configured mox to
attempt sending through a 3rd party from the 4th attempt onwards.
routes and transports are optional. if no route matches, or an empty/zero
transport is selected, normal direct delivery is done.
we could already "submit" emails with 3rd party accounts with "sendmail". but
we now support more SASL authentication mechanisms with SMTP (not only PLAIN,
but also SCRAM-SHA-256, SCRAM-SHA-1 and CRAM-MD5), which sendmail now also
supports. sendmail will use the most secure mechanism supported by the server,
or the explicitly configured mechanism.
for issue #36 by dmikushin. also based on earlier discussion on hackernews.
2023-06-16 19:38:28 +03:00
}
}
checkTransportSocks := func ( name string , t * config . TransportSocks ) {
_ , _ , err := net . SplitHostPort ( t . Address )
if err != nil {
addErrorf ( "transport %s: bad address %s: %v" , name , t . Address , err )
}
for _ , ipstr := range t . RemoteIPs {
ip := net . ParseIP ( ipstr )
if ip == nil {
addErrorf ( "transport %s: bad ip %s" , name , ipstr )
} else {
t . IPs = append ( t . IPs , ip )
}
}
t . Hostname , err = dns . ParseDomain ( t . RemoteHostname )
if err != nil {
addErrorf ( "transport %s: bad hostname %s: %v" , name , t . RemoteHostname , err )
}
}
2024-04-08 22:50:30 +03:00
checkTransportDirect := func ( name string , t * config . TransportDirect ) {
if t . DisableIPv4 && t . DisableIPv6 {
addErrorf ( "transport %s: both IPv4 and IPv6 are disabled, enable at least one" , name )
}
t . IPFamily = "ip"
if t . DisableIPv4 {
t . IPFamily = "ip6"
}
if t . DisableIPv6 {
t . IPFamily = "ip4"
}
}
new feature: when delivering messages from the queue, make it possible to use a "transport"
the default transport is still just "direct delivery", where we connect to the
destination domain's MX servers.
other transports are:
- regular smtp without authentication, this is relaying to a smarthost.
- submission with authentication, e.g. to a third party email sending service.
- direct delivery, but with with connections going through a socks proxy. this
can be helpful if your ip is blocked, you need to get email out, and you have
another IP that isn't blocked.
keep in mind that for all of the above, appropriate SPF/DKIM settings have to
be configured. the "dnscheck" for a domain does a check for any SOCKS IP in the
SPF record. SPF for smtp/submission (ranges? includes?) and any DKIM
requirements cannot really be checked.
which transport is used can be configured through routes. routes can be set on
an account, a domain, or globally. the routes are evaluated in that order, with
the first match selecting the transport. these routes are evaluated for each
delivery attempt. common selection criteria are recipient domain and sender
domain, but also which delivery attempt this is. you could configured mox to
attempt sending through a 3rd party from the 4th attempt onwards.
routes and transports are optional. if no route matches, or an empty/zero
transport is selected, normal direct delivery is done.
we could already "submit" emails with 3rd party accounts with "sendmail". but
we now support more SASL authentication mechanisms with SMTP (not only PLAIN,
but also SCRAM-SHA-256, SCRAM-SHA-1 and CRAM-MD5), which sendmail now also
supports. sendmail will use the most secure mechanism supported by the server,
or the explicitly configured mechanism.
for issue #36 by dmikushin. also based on earlier discussion on hackernews.
2023-06-16 19:38:28 +03:00
for name , t := range c . Transports {
n := 0
if t . Submissions != nil {
n ++
checkTransportSMTP ( name , true , t . Submissions )
}
if t . Submission != nil {
n ++
checkTransportSMTP ( name , false , t . Submission )
}
if t . SMTP != nil {
n ++
checkTransportSMTP ( name , false , t . SMTP )
}
if t . Socks != nil {
n ++
checkTransportSocks ( name , t . Socks )
}
2024-04-08 22:50:30 +03:00
if t . Direct != nil {
n ++
checkTransportDirect ( name , t . Direct )
}
new feature: when delivering messages from the queue, make it possible to use a "transport"
the default transport is still just "direct delivery", where we connect to the
destination domain's MX servers.
other transports are:
- regular smtp without authentication, this is relaying to a smarthost.
- submission with authentication, e.g. to a third party email sending service.
- direct delivery, but with with connections going through a socks proxy. this
can be helpful if your ip is blocked, you need to get email out, and you have
another IP that isn't blocked.
keep in mind that for all of the above, appropriate SPF/DKIM settings have to
be configured. the "dnscheck" for a domain does a check for any SOCKS IP in the
SPF record. SPF for smtp/submission (ranges? includes?) and any DKIM
requirements cannot really be checked.
which transport is used can be configured through routes. routes can be set on
an account, a domain, or globally. the routes are evaluated in that order, with
the first match selecting the transport. these routes are evaluated for each
delivery attempt. common selection criteria are recipient domain and sender
domain, but also which delivery attempt this is. you could configured mox to
attempt sending through a 3rd party from the 4th attempt onwards.
routes and transports are optional. if no route matches, or an empty/zero
transport is selected, normal direct delivery is done.
we could already "submit" emails with 3rd party accounts with "sendmail". but
we now support more SASL authentication mechanisms with SMTP (not only PLAIN,
but also SCRAM-SHA-256, SCRAM-SHA-1 and CRAM-MD5), which sendmail now also
supports. sendmail will use the most secure mechanism supported by the server,
or the explicitly configured mechanism.
for issue #36 by dmikushin. also based on earlier discussion on hackernews.
2023-06-16 19:38:28 +03:00
if n > 1 {
addErrorf ( "transport %s: cannot have multiple methods in a transport" , name )
}
}
2023-01-30 16:27:06 +03:00
// Load CA certificate pool.
if c . TLS . CA != nil {
if c . TLS . CA . AdditionalToSystem {
var err error
c . TLS . CertPool , err = x509 . SystemCertPool ( )
if err != nil {
addErrorf ( "fetching system CA cert pool: %v" , err )
}
} else {
c . TLS . CertPool = x509 . NewCertPool ( )
}
for _ , certfile := range c . TLS . CA . CertFiles {
p := configDirPath ( configFile , certfile )
pemBuf , err := os . ReadFile ( p )
if err != nil {
addErrorf ( "reading TLS CA cert file: %v" , err )
continue
} else if ! c . TLS . CertPool . AppendCertsFromPEM ( pemBuf ) {
// todo: can we check more fully if we're getting some useful data back?
addErrorf ( "no CA certs added from %q" , p )
}
}
}
return
}
// PrepareDynamicConfig parses the dynamic config file given a static file.
2023-12-05 15:35:58 +03:00
func ParseDynamicConfig ( ctx context . Context , log mlog . Log , dynamicPath string , static config . Static ) ( c config . Dynamic , mtime time . Time , accDests map [ string ] AccountDestination , errs [ ] error ) {
2023-01-30 16:27:06 +03:00
addErrorf := func ( format string , args ... any ) {
errs = append ( errs , fmt . Errorf ( format , args ... ) )
}
f , err := os . Open ( dynamicPath )
if err != nil {
addErrorf ( "parsing domains config: %v" , err )
return
}
defer f . Close ( )
fi , err := f . Stat ( )
if err != nil {
addErrorf ( "stat domains config: %v" , err )
}
if err := sconf . Parse ( f , & c ) ; err != nil {
addErrorf ( "parsing dynamic config file: %v" , err )
return
}
2023-12-05 15:35:58 +03:00
accDests , errs = prepareDynamicConfig ( ctx , log , dynamicPath , static , & c )
2023-01-30 16:27:06 +03:00
return c , fi . ModTime ( ) , accDests , errs
}
2023-12-05 15:35:58 +03:00
func prepareDynamicConfig ( ctx context . Context , log mlog . Log , dynamicPath string , static config . Static , c * config . Dynamic ) ( accDests map [ string ] AccountDestination , errs [ ] error ) {
2023-01-30 16:27:06 +03:00
addErrorf := func ( format string , args ... any ) {
errs = append ( errs , fmt . Errorf ( format , args ... ) )
}
2023-03-29 22:11:43 +03:00
// Check that mailbox is in unicode NFC normalized form.
2023-01-30 16:27:06 +03:00
checkMailboxNormf := func ( mailbox string , format string , args ... any ) {
s := norm . NFC . String ( mailbox )
if mailbox != s {
msg := fmt . Sprintf ( format , args ... )
addErrorf ( "%s: mailbox %q is not in NFC normalized form, should be %q" , msg , mailbox , s )
}
}
// Validate postmaster account exists.
if _ , ok := c . Accounts [ static . Postmaster . Account ] ; ! ok {
addErrorf ( "postmaster account %q does not exist" , static . Postmaster . Account )
}
checkMailboxNormf ( static . Postmaster . Mailbox , "postmaster mailbox" )
implement outgoing tls reports
we were already accepting, processing and displaying incoming tls reports. now
we start tracking TLS connection and security-policy-related errors for
outgoing message deliveries as well. we send reports once a day, to the
reporting addresses specified in TLSRPT records (rua) of a policy domain. these
reports are about MTA-STS policies and/or DANE policies, and about
STARTTLS-related failures.
sending reports is enabled by default, but can be disabled through setting
NoOutgoingTLSReports in mox.conf.
only at the end of the implementation process came the realization that the
TLSRPT policy domain for DANE (MX) hosts are separate from the TLSRPT policy
for the recipient domain, and that MTA-STS and DANE TLS/policy results are
typically delivered in separate reports. so MX hosts need their own TLSRPT
policies.
config for the per-host TLSRPT policy should be added to mox.conf for existing
installs, in field HostTLSRPT. it is automatically configured by quickstart for
new installs. with a HostTLSRPT config, the "dns records" and "dns check" admin
pages now suggest the per-host TLSRPT record. by creating that record, you're
requesting TLS reports about your MX host.
gathering all the TLS/policy results is somewhat tricky. the tentacles go
throughout the code. the positive result is that the TLS/policy-related code
had to be cleaned up a bit. for example, the smtpclient TLS modes now reflect
reality better, with independent settings about whether PKIX and/or DANE
verification has to be done, and/or whether verification errors have to be
ignored (e.g. for tls-required: no header). also, cached mtasts policies of
mode "none" are now cleaned up once the MTA-STS DNS record goes away.
2023-11-09 19:40:46 +03:00
accDests = map [ string ] AccountDestination { }
// Validate host TLSRPT account/address.
if static . HostTLSRPT . Account != "" {
if _ , ok := c . Accounts [ static . HostTLSRPT . Account ] ; ! ok {
addErrorf ( "host tlsrpt account %q does not exist" , static . HostTLSRPT . Account )
}
checkMailboxNormf ( static . HostTLSRPT . Mailbox , "host tlsrpt mailbox" )
// Localpart has been parsed already.
addrFull := smtp . NewAddress ( static . HostTLSRPT . ParsedLocalpart , static . HostnameDomain ) . String ( )
dest := config . Destination {
Mailbox : static . HostTLSRPT . Mailbox ,
HostTLSReports : true ,
}
accDests [ addrFull ] = AccountDestination { false , static . HostTLSRPT . ParsedLocalpart , static . HostTLSRPT . Account , dest }
}
improve webserver, add domain redirects (aliases), add tests and admin page ui to manage the config
- make builtin http handlers serve on specific domains, such as for mta-sts, so
e.g. /.well-known/mta-sts.txt isn't served on all domains.
- add logging of a few more fields in access logging.
- small tweaks/bug fixes in webserver request handling.
- add config option for redirecting entire domains to another (common enough).
- split httpserver metric into two: one for duration until writing header (i.e.
performance of server), another for duration until full response is sent to
client (i.e. performance as perceived by users).
- add admin ui, a new page for managing the configs. after making changes
and hitting "save", the changes take effect immediately. the page itself
doesn't look very well-designed (many input fields, makes it look messy). i
have an idea to improve it (explained in admin.html as todo) by making the
layout look just like the config file. not urgent though.
i've already changed my websites/webapps over.
the idea of adding a webserver is to take away a (the) reason for folks to want
to complicate their mox setup by running an other webserver on the same machine.
i think the current webserver implementation can already serve most common use
cases. with a few more tweaks (feedback needed!) we should be able to get to 95%
of the use cases. the reverse proxy can take care of the remaining 5%.
nevertheless, a next step is still to change the quickstart to make it easier
for folks to run with an existing webserver, with existing tls certs/keys.
that's how this relates to issue #5.
2023-03-02 20:15:54 +03:00
var haveSTSListener , haveWebserverListener bool
2023-01-30 16:27:06 +03:00
for _ , l := range static . Listeners {
if l . MTASTSHTTPS . Enabled {
haveSTSListener = true
improve webserver, add domain redirects (aliases), add tests and admin page ui to manage the config
- make builtin http handlers serve on specific domains, such as for mta-sts, so
e.g. /.well-known/mta-sts.txt isn't served on all domains.
- add logging of a few more fields in access logging.
- small tweaks/bug fixes in webserver request handling.
- add config option for redirecting entire domains to another (common enough).
- split httpserver metric into two: one for duration until writing header (i.e.
performance of server), another for duration until full response is sent to
client (i.e. performance as perceived by users).
- add admin ui, a new page for managing the configs. after making changes
and hitting "save", the changes take effect immediately. the page itself
doesn't look very well-designed (many input fields, makes it look messy). i
have an idea to improve it (explained in admin.html as todo) by making the
layout look just like the config file. not urgent though.
i've already changed my websites/webapps over.
the idea of adding a webserver is to take away a (the) reason for folks to want
to complicate their mox setup by running an other webserver on the same machine.
i think the current webserver implementation can already serve most common use
cases. with a few more tweaks (feedback needed!) we should be able to get to 95%
of the use cases. the reverse proxy can take care of the remaining 5%.
nevertheless, a next step is still to change the quickstart to make it easier
for folks to run with an existing webserver, with existing tls certs/keys.
that's how this relates to issue #5.
2023-03-02 20:15:54 +03:00
}
if l . WebserverHTTP . Enabled || l . WebserverHTTPS . Enabled {
haveWebserverListener = true
2023-01-30 16:27:06 +03:00
}
}
new feature: when delivering messages from the queue, make it possible to use a "transport"
the default transport is still just "direct delivery", where we connect to the
destination domain's MX servers.
other transports are:
- regular smtp without authentication, this is relaying to a smarthost.
- submission with authentication, e.g. to a third party email sending service.
- direct delivery, but with with connections going through a socks proxy. this
can be helpful if your ip is blocked, you need to get email out, and you have
another IP that isn't blocked.
keep in mind that for all of the above, appropriate SPF/DKIM settings have to
be configured. the "dnscheck" for a domain does a check for any SOCKS IP in the
SPF record. SPF for smtp/submission (ranges? includes?) and any DKIM
requirements cannot really be checked.
which transport is used can be configured through routes. routes can be set on
an account, a domain, or globally. the routes are evaluated in that order, with
the first match selecting the transport. these routes are evaluated for each
delivery attempt. common selection criteria are recipient domain and sender
domain, but also which delivery attempt this is. you could configured mox to
attempt sending through a 3rd party from the 4th attempt onwards.
routes and transports are optional. if no route matches, or an empty/zero
transport is selected, normal direct delivery is done.
we could already "submit" emails with 3rd party accounts with "sendmail". but
we now support more SASL authentication mechanisms with SMTP (not only PLAIN,
but also SCRAM-SHA-256, SCRAM-SHA-1 and CRAM-MD5), which sendmail now also
supports. sendmail will use the most secure mechanism supported by the server,
or the explicitly configured mechanism.
for issue #36 by dmikushin. also based on earlier discussion on hackernews.
2023-06-16 19:38:28 +03:00
checkRoutes := func ( descr string , routes [ ] config . Route ) {
parseRouteDomains := func ( l [ ] string ) [ ] string {
var r [ ] string
for _ , e := range l {
if e == "." {
r = append ( r , e )
continue
}
prefix := ""
if strings . HasPrefix ( e , "." ) {
prefix = "."
e = e [ 1 : ]
}
d , err := dns . ParseDomain ( e )
if err != nil {
addErrorf ( "%s: invalid domain %s: %v" , descr , e , err )
}
r = append ( r , prefix + d . ASCII )
}
return r
}
for i := range routes {
routes [ i ] . FromDomainASCII = parseRouteDomains ( routes [ i ] . FromDomain )
routes [ i ] . ToDomainASCII = parseRouteDomains ( routes [ i ] . ToDomain )
var ok bool
routes [ i ] . ResolvedTransport , ok = static . Transports [ routes [ i ] . Transport ]
if ! ok {
addErrorf ( "%s: route references undefined transport %s" , descr , routes [ i ] . Transport )
}
}
}
checkRoutes ( "global routes" , c . Routes )
2023-01-30 16:27:06 +03:00
// Validate domains.
for d , domain := range c . Domains {
dnsdomain , err := dns . ParseDomain ( d )
if err != nil {
addErrorf ( "bad domain %q: %s" , d , err )
} else if dnsdomain . Name ( ) != d {
2024-01-24 12:48:44 +03:00
addErrorf ( "domain %s must be specified in unicode form, %s" , d , dnsdomain . Name ( ) )
2023-01-30 16:27:06 +03:00
}
domain . Domain = dnsdomain
assume a dns cname record mail.<domain>, pointing to the hostname of the mail server, for clients to connect to
the autoconfig/autodiscover endpoints, and the printed client settings (in
quickstart, in the admin interface) now all point to the cname record (called
"client settings domain"). it is configurable per domain, and set to
"mail.<domain>" by default. for existing mox installs, the domain can be added
by editing the config file.
this makes it easier for a domain to migrate to another server in the future.
client settings don't have to be updated, the cname can just be changed.
before, the hostname of the mail server was configured in email clients.
migrating away would require changing settings in all clients.
if a client settings domain is configured, a TLS certificate for the name will
be requested through ACME, or must be configured manually.
2023-12-24 13:01:16 +03:00
if domain . ClientSettingsDomain != "" {
csd , err := dns . ParseDomain ( domain . ClientSettingsDomain )
if err != nil {
addErrorf ( "bad client settings domain %q: %s" , domain . ClientSettingsDomain , err )
}
domain . ClientSettingsDNSDomain = csd
}
2023-01-30 16:27:06 +03:00
for _ , sign := range domain . DKIM . Sign {
if _ , ok := domain . DKIM . Selectors [ sign ] ; ! ok {
2023-03-09 22:18:34 +03:00
addErrorf ( "selector %s for signing is missing in domain %s" , sign , d )
2023-01-30 16:27:06 +03:00
}
}
for name , sel := range domain . DKIM . Selectors {
seld , err := dns . ParseDomain ( name )
if err != nil {
addErrorf ( "bad selector %q: %s" , name , err )
} else if seld . Name ( ) != name {
2024-01-24 12:48:44 +03:00
addErrorf ( "selector %q must be specified in unicode form, %q" , name , seld . Name ( ) )
2023-01-30 16:27:06 +03:00
}
sel . Domain = seld
if sel . Expiration != "" {
exp , err := time . ParseDuration ( sel . Expiration )
if err != nil {
addErrorf ( "selector %q has invalid expiration %q: %v" , name , sel . Expiration , err )
} else {
sel . ExpirationSeconds = int ( exp / time . Second )
}
}
sel . HashEffective = sel . Hash
switch sel . HashEffective {
case "" :
sel . HashEffective = "sha256"
case "sha1" :
log . Error ( "using sha1 with DKIM is deprecated as not secure enough, switch to sha256" )
case "sha256" :
default :
2023-03-09 22:18:34 +03:00
addErrorf ( "unsupported hash %q for selector %q in domain %s" , sel . HashEffective , name , d )
2023-01-30 16:27:06 +03:00
}
pemBuf , err := os . ReadFile ( configDirPath ( dynamicPath , sel . PrivateKeyFile ) )
if err != nil {
2023-03-09 22:18:34 +03:00
addErrorf ( "reading private key for selector %s in domain %s: %s" , name , d , err )
2023-01-30 16:27:06 +03:00
continue
}
p , _ := pem . Decode ( pemBuf )
if p == nil {
2023-03-09 22:18:34 +03:00
addErrorf ( "private key for selector %s in domain %s has no PEM block" , name , d )
2023-01-30 16:27:06 +03:00
continue
}
key , err := x509 . ParsePKCS8PrivateKey ( p . Bytes )
if err != nil {
2023-03-09 22:18:34 +03:00
addErrorf ( "parsing private key for selector %s in domain %s: %s" , name , d , err )
2023-01-30 16:27:06 +03:00
continue
}
switch k := key . ( type ) {
case * rsa . PrivateKey :
if k . N . BitLen ( ) < 1024 {
// ../rfc/6376:757
// Let's help user do the right thing.
addErrorf ( "rsa keys should be >= 1024 bits" )
}
sel . Key = k
2024-04-19 11:23:53 +03:00
sel . Algorithm = fmt . Sprintf ( "rsa-%d" , k . N . BitLen ( ) )
2023-01-30 16:27:06 +03:00
case ed25519 . PrivateKey :
if sel . HashEffective != "sha256" {
addErrorf ( "hash algorithm %q is not supported with ed25519, only sha256 is" , sel . HashEffective )
}
sel . Key = k
2024-04-19 11:23:53 +03:00
sel . Algorithm = "ed25519"
2023-01-30 16:27:06 +03:00
default :
2023-03-09 22:18:34 +03:00
addErrorf ( "private key type %T not yet supported, at selector %s in domain %s" , key , name , d )
2023-01-30 16:27:06 +03:00
}
if len ( sel . Headers ) == 0 {
// ../rfc/6376:2139
// ../rfc/6376:2203
// ../rfc/6376:2212
// By default we seal signed headers, and we sign user-visible headers to
// prevent/limit reuse of previously signed messages: All addressing fields, date
// and subject, message-referencing fields, parsing instructions (content-type).
sel . HeadersEffective = strings . Split ( "From,To,Cc,Bcc,Reply-To,References,In-Reply-To,Subject,Date,Message-Id,Content-Type" , "," )
} else {
var from bool
for _ , h := range sel . Headers {
from = from || strings . EqualFold ( h , "From" )
// ../rfc/6376:2269
if strings . EqualFold ( h , "DKIM-Signature" ) || strings . EqualFold ( h , "Received" ) || strings . EqualFold ( h , "Return-Path" ) {
log . Error ( "DKIM-signing header %q is recommended against as it may be modified in transit" )
}
}
if ! from {
addErrorf ( "From-field must always be DKIM-signed" )
}
sel . HeadersEffective = sel . Headers
}
domain . DKIM . Selectors [ name ] = sel
}
if domain . MTASTS != nil {
if ! haveSTSListener {
addErrorf ( "MTA-STS enabled for domain %q, but there is no listener for MTASTS" , d )
}
sts := domain . MTASTS
if sts . PolicyID == "" {
addErrorf ( "invalid empty MTA-STS PolicyID" )
}
switch sts . Mode {
case mtasts . ModeNone , mtasts . ModeTesting , mtasts . ModeEnforce :
default :
addErrorf ( "invalid mtasts mode %q" , sts . Mode )
}
}
new feature: when delivering messages from the queue, make it possible to use a "transport"
the default transport is still just "direct delivery", where we connect to the
destination domain's MX servers.
other transports are:
- regular smtp without authentication, this is relaying to a smarthost.
- submission with authentication, e.g. to a third party email sending service.
- direct delivery, but with with connections going through a socks proxy. this
can be helpful if your ip is blocked, you need to get email out, and you have
another IP that isn't blocked.
keep in mind that for all of the above, appropriate SPF/DKIM settings have to
be configured. the "dnscheck" for a domain does a check for any SOCKS IP in the
SPF record. SPF for smtp/submission (ranges? includes?) and any DKIM
requirements cannot really be checked.
which transport is used can be configured through routes. routes can be set on
an account, a domain, or globally. the routes are evaluated in that order, with
the first match selecting the transport. these routes are evaluated for each
delivery attempt. common selection criteria are recipient domain and sender
domain, but also which delivery attempt this is. you could configured mox to
attempt sending through a 3rd party from the 4th attempt onwards.
routes and transports are optional. if no route matches, or an empty/zero
transport is selected, normal direct delivery is done.
we could already "submit" emails with 3rd party accounts with "sendmail". but
we now support more SASL authentication mechanisms with SMTP (not only PLAIN,
but also SCRAM-SHA-256, SCRAM-SHA-1 and CRAM-MD5), which sendmail now also
supports. sendmail will use the most secure mechanism supported by the server,
or the explicitly configured mechanism.
for issue #36 by dmikushin. also based on earlier discussion on hackernews.
2023-06-16 19:38:28 +03:00
checkRoutes ( "routes for domain" , domain . Routes )
2023-01-30 16:27:06 +03:00
c . Domains [ d ] = domain
}
2024-01-26 21:51:23 +03:00
// To determine ReportsOnly.
domainHasAddress := map [ string ] bool { }
2023-03-29 11:55:05 +03:00
// Validate email addresses.
2023-01-30 16:27:06 +03:00
for accName , acc := range c . Accounts {
var err error
acc . DNSDomain , err = dns . ParseDomain ( acc . Domain )
if err != nil {
2023-03-09 22:18:34 +03:00
addErrorf ( "parsing domain %s for account %q: %s" , acc . Domain , accName , err )
2023-01-30 16:27:06 +03:00
}
if strings . EqualFold ( acc . RejectsMailbox , "Inbox" ) {
improve training of junk filter
before, we used heuristics to decide when to train/untrain a message as junk or
nonjunk: the message had to be seen, be in certain mailboxes. then if a message
was marked as junk, it was junk. and otherwise it was nonjunk. this wasn't good
enough: you may want to keep some messages around as neither junk or nonjunk.
and that wasn't possible.
ideally, we would just look at the imap $Junk and $NotJunk flags. the problem
is that mail clients don't set these flags, or don't make it easy. thunderbird
can set the flags based on its own bayesian filter. it has a shortcut for
marking Junk and moving it to the junk folder (good), but the counterpart of
notjunk only marks a message as notjunk without showing in the UI that it was
marked as notjunk. there is also no "move and mark as notjunk" mechanism. e.g.
"archive" does not mark a message as notjunk. ios mail and mutt don't appear to
have any way to see or change the $Junk and $NotJunk flags.
what email clients do have is the ability to move messages to other
mailboxes/folders. so mox now has a mechanism that allows you to configure
mailboxes that automatically set $Junk or $NotJunk (or clear both) when a
message is moved/copied/delivered to that folder. e.g. a mailbox called junk or
spam or rejects marks its messags as junk. inbox, postmaster, dmarc, tlsrpt,
neutral* mark their messages as neither junk or notjunk. other folders mark
their messages as notjunk. e.g. list/*, archive. this functionality is
optional, but enabled with the quickstart and for new accounts.
also, mox now keeps track of the previous training of a message and will only
untrain/train if needed. before, there probably have been duplicate or missing
(un)trainings.
this also includes a new subcommand "retrain" to recreate the junkfilter for an
account. you should run it after updating to this version. and you should
probably also modify your account config to include the AutomaticJunkFlags.
2023-02-12 01:00:12 +03:00
addErrorf ( "account %q: cannot set RejectsMailbox to inbox, messages will be removed automatically from the rejects mailbox" , accName )
2023-01-30 16:27:06 +03:00
}
checkMailboxNormf ( acc . RejectsMailbox , "account %q" , accName )
improve training of junk filter
before, we used heuristics to decide when to train/untrain a message as junk or
nonjunk: the message had to be seen, be in certain mailboxes. then if a message
was marked as junk, it was junk. and otherwise it was nonjunk. this wasn't good
enough: you may want to keep some messages around as neither junk or nonjunk.
and that wasn't possible.
ideally, we would just look at the imap $Junk and $NotJunk flags. the problem
is that mail clients don't set these flags, or don't make it easy. thunderbird
can set the flags based on its own bayesian filter. it has a shortcut for
marking Junk and moving it to the junk folder (good), but the counterpart of
notjunk only marks a message as notjunk without showing in the UI that it was
marked as notjunk. there is also no "move and mark as notjunk" mechanism. e.g.
"archive" does not mark a message as notjunk. ios mail and mutt don't appear to
have any way to see or change the $Junk and $NotJunk flags.
what email clients do have is the ability to move messages to other
mailboxes/folders. so mox now has a mechanism that allows you to configure
mailboxes that automatically set $Junk or $NotJunk (or clear both) when a
message is moved/copied/delivered to that folder. e.g. a mailbox called junk or
spam or rejects marks its messags as junk. inbox, postmaster, dmarc, tlsrpt,
neutral* mark their messages as neither junk or notjunk. other folders mark
their messages as notjunk. e.g. list/*, archive. this functionality is
optional, but enabled with the quickstart and for new accounts.
also, mox now keeps track of the previous training of a message and will only
untrain/train if needed. before, there probably have been duplicate or missing
(un)trainings.
this also includes a new subcommand "retrain" to recreate the junkfilter for an
account. you should run it after updating to this version. and you should
probably also modify your account config to include the AutomaticJunkFlags.
2023-02-12 01:00:12 +03:00
if acc . AutomaticJunkFlags . JunkMailboxRegexp != "" {
r , err := regexp . Compile ( acc . AutomaticJunkFlags . JunkMailboxRegexp )
if err != nil {
addErrorf ( "invalid JunkMailboxRegexp regular expression: %v" , err )
}
acc . JunkMailbox = r
}
if acc . AutomaticJunkFlags . NeutralMailboxRegexp != "" {
r , err := regexp . Compile ( acc . AutomaticJunkFlags . NeutralMailboxRegexp )
if err != nil {
addErrorf ( "invalid NeutralMailboxRegexp regular expression: %v" , err )
}
acc . NeutralMailbox = r
}
if acc . AutomaticJunkFlags . NotJunkMailboxRegexp != "" {
r , err := regexp . Compile ( acc . AutomaticJunkFlags . NotJunkMailboxRegexp )
if err != nil {
addErrorf ( "invalid NotJunkMailboxRegexp regular expression: %v" , err )
}
acc . NotJunkMailbox = r
}
add a webapi and webhooks for a simple http/json-based api
for applications to compose/send messages, receive delivery feedback, and
maintain suppression lists.
this is an alternative to applications using a library to compose messages,
submitting those messages using smtp, and monitoring a mailbox with imap for
DSNs, which can be processed into the equivalent of suppression lists. but you
need to know about all these standards/protocols and find libraries. by using
the webapi & webhooks, you just need a http & json library.
unfortunately, there is no standard for these kinds of api, so mox has made up
yet another one...
matching incoming DSNs about deliveries to original outgoing messages requires
keeping history of "retired" messages (delivered from the queue, either
successfully or failed). this can be enabled per account. history is also
useful for debugging deliveries. we now also keep history of each delivery
attempt, accessible while still in the queue, and kept when a message is
retired. the queue webadmin pages now also have pagination, to show potentially
large history.
a queue of webhook calls is now managed too. failures are retried similar to
message deliveries. webhooks can also be saved to the retired list after
completing. also configurable per account.
messages can be sent with a "unique smtp mail from" address. this can only be
used if the domain is configured with a localpart catchall separator such as
"+". when enabled, a queued message gets assigned a random "fromid", which is
added after the separator when sending. when DSNs are returned, they can be
related to previously sent messages based on this fromid. in the future, we can
implement matching on the "envid" used in the smtp dsn extension, or on the
"message-id" of the message. using a fromid can be triggered by authenticating
with a login email address that is configured as enabling fromid.
suppression lists are automatically managed per account. if a delivery attempt
results in certain smtp errors, the destination address is added to the
suppression list. future messages queued for that recipient will immediately
fail without a delivery attempt. suppression lists protect your mail server
reputation.
submitted messages can carry "extra" data through the queue and webhooks for
outgoing deliveries. through webapi as a json object, through smtp submission
as message headers of the form "x-mox-extra-<key>: value".
to make it easy to test webapi/webhooks locally, the "localserve" mode actually
puts messages in the queue. when it's time to deliver, it still won't do a full
delivery attempt, but just delivers to the sender account. unless the recipient
address has a special form, simulating a failure to deliver.
admins now have more control over the queue. "hold rules" can be added to mark
newly queued messages as "on hold", pausing delivery. rules can be about
certain sender or recipient domains/addresses, or apply to all messages pausing
the entire queue. also useful for (local) testing.
new config options have been introduced. they are editable through the admin
and/or account web interfaces.
the webapi http endpoints are enabled for newly generated configs with the
quickstart, and in localserve. existing configurations must explicitly enable
the webapi in mox.conf.
gopherwatch.org was created to dogfood this code. it initially used just the
compose/smtpclient/imapclient mox packages to send messages and process
delivery feedback. it will get a config option to use the mox webapi/webhooks
instead. the gopherwatch code to use webapi/webhook is smaller and simpler, and
developing that shaped development of the mox webapi/webhooks.
for issue #31 by cuu508
2024-04-15 22:49:02 +03:00
acc . ParsedFromIDLoginAddresses = make ( [ ] smtp . Address , len ( acc . FromIDLoginAddresses ) )
for i , s := range acc . FromIDLoginAddresses {
a , err := smtp . ParseAddress ( s )
if err != nil {
addErrorf ( "invalid fromid login address %q in account %q: %v" , s , accName , err )
}
// We check later on if address belongs to account.
dom , ok := c . Domains [ a . Domain . Name ( ) ]
if ! ok {
addErrorf ( "unknown domain in fromid login address %q for account %q" , s , accName )
} else if dom . LocalpartCatchallSeparator == "" {
addErrorf ( "localpart catchall separator not configured for domain for fromid login address %q for account %q" , s , accName )
}
acc . ParsedFromIDLoginAddresses [ i ] = a
}
improve training of junk filter
before, we used heuristics to decide when to train/untrain a message as junk or
nonjunk: the message had to be seen, be in certain mailboxes. then if a message
was marked as junk, it was junk. and otherwise it was nonjunk. this wasn't good
enough: you may want to keep some messages around as neither junk or nonjunk.
and that wasn't possible.
ideally, we would just look at the imap $Junk and $NotJunk flags. the problem
is that mail clients don't set these flags, or don't make it easy. thunderbird
can set the flags based on its own bayesian filter. it has a shortcut for
marking Junk and moving it to the junk folder (good), but the counterpart of
notjunk only marks a message as notjunk without showing in the UI that it was
marked as notjunk. there is also no "move and mark as notjunk" mechanism. e.g.
"archive" does not mark a message as notjunk. ios mail and mutt don't appear to
have any way to see or change the $Junk and $NotJunk flags.
what email clients do have is the ability to move messages to other
mailboxes/folders. so mox now has a mechanism that allows you to configure
mailboxes that automatically set $Junk or $NotJunk (or clear both) when a
message is moved/copied/delivered to that folder. e.g. a mailbox called junk or
spam or rejects marks its messags as junk. inbox, postmaster, dmarc, tlsrpt,
neutral* mark their messages as neither junk or notjunk. other folders mark
their messages as notjunk. e.g. list/*, archive. this functionality is
optional, but enabled with the quickstart and for new accounts.
also, mox now keeps track of the previous training of a message and will only
untrain/train if needed. before, there probably have been duplicate or missing
(un)trainings.
this also includes a new subcommand "retrain" to recreate the junkfilter for an
account. you should run it after updating to this version. and you should
probably also modify your account config to include the AutomaticJunkFlags.
2023-02-12 01:00:12 +03:00
c . Accounts [ accName ] = acc
add a webapi and webhooks for a simple http/json-based api
for applications to compose/send messages, receive delivery feedback, and
maintain suppression lists.
this is an alternative to applications using a library to compose messages,
submitting those messages using smtp, and monitoring a mailbox with imap for
DSNs, which can be processed into the equivalent of suppression lists. but you
need to know about all these standards/protocols and find libraries. by using
the webapi & webhooks, you just need a http & json library.
unfortunately, there is no standard for these kinds of api, so mox has made up
yet another one...
matching incoming DSNs about deliveries to original outgoing messages requires
keeping history of "retired" messages (delivered from the queue, either
successfully or failed). this can be enabled per account. history is also
useful for debugging deliveries. we now also keep history of each delivery
attempt, accessible while still in the queue, and kept when a message is
retired. the queue webadmin pages now also have pagination, to show potentially
large history.
a queue of webhook calls is now managed too. failures are retried similar to
message deliveries. webhooks can also be saved to the retired list after
completing. also configurable per account.
messages can be sent with a "unique smtp mail from" address. this can only be
used if the domain is configured with a localpart catchall separator such as
"+". when enabled, a queued message gets assigned a random "fromid", which is
added after the separator when sending. when DSNs are returned, they can be
related to previously sent messages based on this fromid. in the future, we can
implement matching on the "envid" used in the smtp dsn extension, or on the
"message-id" of the message. using a fromid can be triggered by authenticating
with a login email address that is configured as enabling fromid.
suppression lists are automatically managed per account. if a delivery attempt
results in certain smtp errors, the destination address is added to the
suppression list. future messages queued for that recipient will immediately
fail without a delivery attempt. suppression lists protect your mail server
reputation.
submitted messages can carry "extra" data through the queue and webhooks for
outgoing deliveries. through webapi as a json object, through smtp submission
as message headers of the form "x-mox-extra-<key>: value".
to make it easy to test webapi/webhooks locally, the "localserve" mode actually
puts messages in the queue. when it's time to deliver, it still won't do a full
delivery attempt, but just delivers to the sender account. unless the recipient
address has a special form, simulating a failure to deliver.
admins now have more control over the queue. "hold rules" can be added to mark
newly queued messages as "on hold", pausing delivery. rules can be about
certain sender or recipient domains/addresses, or apply to all messages pausing
the entire queue. also useful for (local) testing.
new config options have been introduced. they are editable through the admin
and/or account web interfaces.
the webapi http endpoints are enabled for newly generated configs with the
quickstart, and in localserve. existing configurations must explicitly enable
the webapi in mox.conf.
gopherwatch.org was created to dogfood this code. it initially used just the
compose/smtpclient/imapclient mox packages to send messages and process
delivery feedback. it will get a config option to use the mox webapi/webhooks
instead. the gopherwatch code to use webapi/webhook is smaller and simpler, and
developing that shaped development of the mox webapi/webhooks.
for issue #31 by cuu508
2024-04-15 22:49:02 +03:00
if acc . OutgoingWebhook != nil {
u , err := url . Parse ( acc . OutgoingWebhook . URL )
if err == nil && ( u . Scheme != "http" && u . Scheme != "https" ) {
err = errors . New ( "scheme must be http or https" )
}
if err != nil {
addErrorf ( "parsing outgoing hook url %q in account %q: %v" , acc . OutgoingWebhook . URL , accName , err )
}
// note: outgoing hook events are in ../queue/hooks.go, ../mox-/config.go, ../queue.go and ../webapi/gendoc.sh. keep in sync.
outgoingHookEvents := [ ] string { "delivered" , "suppressed" , "delayed" , "failed" , "relayed" , "expanded" , "canceled" , "unrecognized" }
for _ , e := range acc . OutgoingWebhook . Events {
if ! slices . Contains ( outgoingHookEvents , e ) {
addErrorf ( "unknown outgoing hook event %q" , e )
}
}
}
if acc . IncomingWebhook != nil {
u , err := url . Parse ( acc . IncomingWebhook . URL )
if err == nil && ( u . Scheme != "http" && u . Scheme != "https" ) {
err = errors . New ( "scheme must be http or https" )
}
if err != nil {
addErrorf ( "parsing incoming hook url %q in account %q: %v" , acc . IncomingWebhook . URL , accName , err )
}
}
2023-03-10 00:07:37 +03:00
// todo deprecated: only localpart as keys for Destinations, we are replacing them with full addresses. if domains.conf is written, we won't have to do this again.
replaceLocalparts := map [ string ] string { }
2023-01-30 16:27:06 +03:00
for addrName , dest := range acc . Destinations {
checkMailboxNormf ( dest . Mailbox , "account %q, destination %q" , accName , addrName )
for i , rs := range dest . Rulesets {
checkMailboxNormf ( rs . Mailbox , "account %q, destination %q, ruleset %d" , accName , addrName , i + 1 )
n := 0
if rs . SMTPMailFromRegexp != "" {
n ++
r , err := regexp . Compile ( rs . SMTPMailFromRegexp )
if err != nil {
addErrorf ( "invalid SMTPMailFrom regular expression: %v" , err )
}
c . Accounts [ accName ] . Destinations [ addrName ] . Rulesets [ i ] . SMTPMailFromRegexpCompiled = r
}
webmail: when moving a single message out of/to the inbox, ask if user wants to create a rule to automatically do that server-side for future deliveries
if the message has a list-id header, we assume this is a (mailing) list
message, and we require a dkim/spf-verified domain (we prefer the shortest that
is a suffix of the list-id value). the rule we would add will mark such
messages as from a mailing list, changing filtering rules on incoming messages
(not enforcing dmarc policies). messages will be matched on list-id header and
will only match if they have the same dkim/spf-verified domain.
if the message doesn't have a list-id header, we'll ask to match based on
"message from" address.
we don't ask the user in several cases:
- if the destination/source mailbox is a special-use mailbox (e.g.
trash,archive,sent,junk; inbox isn't included)
- if the rule already exist (no point in adding it again).
- if the user said "no, not for this list-id/from-address" in the past.
- if the user said "no, not for messages moved to this mailbox" in the past.
we'll add the rule if the message was moved out of the inbox.
if the message was moved to the inbox, we check if there is a matching rule
that we can remove.
we now remember the "no" answers (for list-id, msg-from-addr and mailbox) in
the account database.
to implement the msgfrom rules, this adds support to rulesets for matching on
message "from" address. before, we could match on smtp from address (and other
fields). rulesets now also have a field for comments. webmail adds a note that
it created the rule, with the date.
manual editing of the rulesets is still in the webaccount page. this webmail
functionality is just a convenient way to add/remove common rules.
2024-04-21 18:01:50 +03:00
if rs . MsgFromRegexp != "" {
n ++
r , err := regexp . Compile ( rs . MsgFromRegexp )
if err != nil {
addErrorf ( "invalid MsgFrom regular expression: %v" , err )
}
c . Accounts [ accName ] . Destinations [ addrName ] . Rulesets [ i ] . MsgFromRegexpCompiled = r
}
2023-01-30 16:27:06 +03:00
if rs . VerifiedDomain != "" {
n ++
d , err := dns . ParseDomain ( rs . VerifiedDomain )
if err != nil {
addErrorf ( "invalid VerifiedDomain: %v" , err )
}
c . Accounts [ accName ] . Destinations [ addrName ] . Rulesets [ i ] . VerifiedDNSDomain = d
}
var hdr [ ] [ 2 ] * regexp . Regexp
for k , v := range rs . HeadersRegexp {
n ++
if strings . ToLower ( k ) != k {
addErrorf ( "header field %q must only have lower case characters" , k )
}
if strings . ToLower ( v ) != v {
addErrorf ( "header value %q must only have lower case characters" , v )
}
rk , err := regexp . Compile ( k )
if err != nil {
addErrorf ( "invalid rule header regexp %q: %v" , k , err )
}
rv , err := regexp . Compile ( v )
if err != nil {
addErrorf ( "invalid rule header regexp %q: %v" , v , err )
}
hdr = append ( hdr , [ ... ] * regexp . Regexp { rk , rv } )
}
c . Accounts [ accName ] . Destinations [ addrName ] . Rulesets [ i ] . HeadersRegexpCompiled = hdr
if n == 0 {
addErrorf ( "ruleset must have at least one rule" )
}
2023-08-09 23:31:37 +03:00
if rs . IsForward && rs . ListAllowDomain != "" {
addErrorf ( "ruleset cannot have both IsForward and ListAllowDomain" )
}
if rs . IsForward {
if rs . SMTPMailFromRegexp == "" || rs . VerifiedDomain == "" {
addErrorf ( "ruleset with IsForward must have both SMTPMailFromRegexp and VerifiedDomain too" )
}
}
2023-01-30 16:27:06 +03:00
if rs . ListAllowDomain != "" {
d , err := dns . ParseDomain ( rs . ListAllowDomain )
if err != nil {
addErrorf ( "invalid ListAllowDomain %q: %v" , rs . ListAllowDomain , err )
}
c . Accounts [ accName ] . Destinations [ addrName ] . Rulesets [ i ] . ListAllowDNSDomain = d
}
2023-08-09 23:31:37 +03:00
checkMailboxNormf ( rs . AcceptRejectsToMailbox , "account %q, destination %q, ruleset %d, rejects mailbox" , accName , addrName , i + 1 )
if strings . EqualFold ( rs . AcceptRejectsToMailbox , "inbox" ) {
addErrorf ( "account %q, destination %q, ruleset %d: AcceptRejectsToMailbox cannot be set to Inbox" , accName , addrName , i + 1 )
}
2023-01-30 16:27:06 +03:00
}
2023-03-29 22:11:43 +03:00
// Catchall destination for domain.
if strings . HasPrefix ( addrName , "@" ) {
d , err := dns . ParseDomain ( addrName [ 1 : ] )
if err != nil {
addErrorf ( "parsing domain %q in account %q" , addrName [ 1 : ] , accName )
continue
} else if _ , ok := c . Domains [ d . Name ( ) ] ; ! ok {
addErrorf ( "unknown domain for address %q in account %q" , addrName , accName )
continue
}
2024-01-26 21:51:23 +03:00
domainHasAddress [ d . Name ( ) ] = true
2023-03-29 22:11:43 +03:00
addrFull := "@" + d . Name ( )
if _ , ok := accDests [ addrFull ] ; ok {
addErrorf ( "duplicate canonicalized catchall destination address %s" , addrFull )
}
accDests [ addrFull ] = AccountDestination { true , "" , accName , dest }
continue
}
2023-03-10 00:07:37 +03:00
// todo deprecated: remove support for parsing destination as just a localpart instead full address.
2023-01-30 16:27:06 +03:00
var address smtp . Address
2023-03-29 22:11:43 +03:00
if localpart , err := smtp . ParseLocalpart ( addrName ) ; err != nil && errors . Is ( err , smtp . ErrBadLocalpart ) {
2023-01-30 16:27:06 +03:00
address , err = smtp . ParseAddress ( addrName )
if err != nil {
addErrorf ( "invalid email address %q in account %q" , addrName , accName )
continue
} else if _ , ok := c . Domains [ address . Domain . Name ( ) ] ; ! ok {
addErrorf ( "unknown domain for address %q in account %q" , addrName , accName )
continue
}
} else {
if err != nil {
addErrorf ( "invalid localpart %q in account %q" , addrName , accName )
continue
}
address = smtp . NewAddress ( localpart , acc . DNSDomain )
if _ , ok := c . Domains [ acc . DNSDomain . Name ( ) ] ; ! ok {
2023-03-09 22:18:34 +03:00
addErrorf ( "unknown domain %s for account %q" , acc . DNSDomain . Name ( ) , accName )
2023-01-30 16:27:06 +03:00
continue
}
2023-03-10 00:07:37 +03:00
replaceLocalparts [ addrName ] = address . Pack ( true )
2023-01-30 16:27:06 +03:00
}
2023-03-29 11:55:05 +03:00
2023-03-29 22:11:43 +03:00
origLP := address . Localpart
2023-03-29 11:55:05 +03:00
dc := c . Domains [ address . Domain . Name ( ) ]
2024-01-26 21:51:23 +03:00
domainHasAddress [ address . Domain . Name ( ) ] = true
2023-03-29 11:55:05 +03:00
if lp , err := CanonicalLocalpart ( address . Localpart , dc ) ; err != nil {
addErrorf ( "canonicalizing localpart %s: %v" , address . Localpart , err )
} else if dc . LocalpartCatchallSeparator != "" && strings . Contains ( string ( address . Localpart ) , dc . LocalpartCatchallSeparator ) {
addErrorf ( "localpart of address %s includes domain catchall separator %s" , address , dc . LocalpartCatchallSeparator )
} else {
address . Localpart = lp
}
2023-01-30 16:27:06 +03:00
addrFull := address . Pack ( true )
if _ , ok := accDests [ addrFull ] ; ok {
2023-03-29 11:55:05 +03:00
addErrorf ( "duplicate canonicalized destination address %s" , addrFull )
2023-01-30 16:27:06 +03:00
}
2023-03-29 22:11:43 +03:00
accDests [ addrFull ] = AccountDestination { false , origLP , accName , dest }
2023-01-30 16:27:06 +03:00
}
2023-03-10 00:07:37 +03:00
for lp , addr := range replaceLocalparts {
dest , ok := acc . Destinations [ lp ]
if ! ok {
addErrorf ( "could not find localpart %q to replace with address in destinations" , lp )
} else {
2023-12-14 22:26:06 +03:00
log . Warn ( ` deprecation warning: support for account destination addresses specified as just localpart ("username") instead of full email address will be removed in the future; update domains.conf, for each Account, for each Destination, ensure each key is an email address by appending "@" and the default domain for the account ` ,
2023-12-05 18:06:50 +03:00
slog . Any ( "localpart" , lp ) ,
slog . Any ( "address" , addr ) ,
slog . String ( "account" , accName ) )
2023-03-10 00:07:37 +03:00
acc . Destinations [ addr ] = dest
delete ( acc . Destinations , lp )
}
}
new feature: when delivering messages from the queue, make it possible to use a "transport"
the default transport is still just "direct delivery", where we connect to the
destination domain's MX servers.
other transports are:
- regular smtp without authentication, this is relaying to a smarthost.
- submission with authentication, e.g. to a third party email sending service.
- direct delivery, but with with connections going through a socks proxy. this
can be helpful if your ip is blocked, you need to get email out, and you have
another IP that isn't blocked.
keep in mind that for all of the above, appropriate SPF/DKIM settings have to
be configured. the "dnscheck" for a domain does a check for any SOCKS IP in the
SPF record. SPF for smtp/submission (ranges? includes?) and any DKIM
requirements cannot really be checked.
which transport is used can be configured through routes. routes can be set on
an account, a domain, or globally. the routes are evaluated in that order, with
the first match selecting the transport. these routes are evaluated for each
delivery attempt. common selection criteria are recipient domain and sender
domain, but also which delivery attempt this is. you could configured mox to
attempt sending through a 3rd party from the 4th attempt onwards.
routes and transports are optional. if no route matches, or an empty/zero
transport is selected, normal direct delivery is done.
we could already "submit" emails with 3rd party accounts with "sendmail". but
we now support more SASL authentication mechanisms with SMTP (not only PLAIN,
but also SCRAM-SHA-256, SCRAM-SHA-1 and CRAM-MD5), which sendmail now also
supports. sendmail will use the most secure mechanism supported by the server,
or the explicitly configured mechanism.
for issue #36 by dmikushin. also based on earlier discussion on hackernews.
2023-06-16 19:38:28 +03:00
add a webapi and webhooks for a simple http/json-based api
for applications to compose/send messages, receive delivery feedback, and
maintain suppression lists.
this is an alternative to applications using a library to compose messages,
submitting those messages using smtp, and monitoring a mailbox with imap for
DSNs, which can be processed into the equivalent of suppression lists. but you
need to know about all these standards/protocols and find libraries. by using
the webapi & webhooks, you just need a http & json library.
unfortunately, there is no standard for these kinds of api, so mox has made up
yet another one...
matching incoming DSNs about deliveries to original outgoing messages requires
keeping history of "retired" messages (delivered from the queue, either
successfully or failed). this can be enabled per account. history is also
useful for debugging deliveries. we now also keep history of each delivery
attempt, accessible while still in the queue, and kept when a message is
retired. the queue webadmin pages now also have pagination, to show potentially
large history.
a queue of webhook calls is now managed too. failures are retried similar to
message deliveries. webhooks can also be saved to the retired list after
completing. also configurable per account.
messages can be sent with a "unique smtp mail from" address. this can only be
used if the domain is configured with a localpart catchall separator such as
"+". when enabled, a queued message gets assigned a random "fromid", which is
added after the separator when sending. when DSNs are returned, they can be
related to previously sent messages based on this fromid. in the future, we can
implement matching on the "envid" used in the smtp dsn extension, or on the
"message-id" of the message. using a fromid can be triggered by authenticating
with a login email address that is configured as enabling fromid.
suppression lists are automatically managed per account. if a delivery attempt
results in certain smtp errors, the destination address is added to the
suppression list. future messages queued for that recipient will immediately
fail without a delivery attempt. suppression lists protect your mail server
reputation.
submitted messages can carry "extra" data through the queue and webhooks for
outgoing deliveries. through webapi as a json object, through smtp submission
as message headers of the form "x-mox-extra-<key>: value".
to make it easy to test webapi/webhooks locally, the "localserve" mode actually
puts messages in the queue. when it's time to deliver, it still won't do a full
delivery attempt, but just delivers to the sender account. unless the recipient
address has a special form, simulating a failure to deliver.
admins now have more control over the queue. "hold rules" can be added to mark
newly queued messages as "on hold", pausing delivery. rules can be about
certain sender or recipient domains/addresses, or apply to all messages pausing
the entire queue. also useful for (local) testing.
new config options have been introduced. they are editable through the admin
and/or account web interfaces.
the webapi http endpoints are enabled for newly generated configs with the
quickstart, and in localserve. existing configurations must explicitly enable
the webapi in mox.conf.
gopherwatch.org was created to dogfood this code. it initially used just the
compose/smtpclient/imapclient mox packages to send messages and process
delivery feedback. it will get a config option to use the mox webapi/webhooks
instead. the gopherwatch code to use webapi/webhook is smaller and simpler, and
developing that shaped development of the mox webapi/webhooks.
for issue #31 by cuu508
2024-04-15 22:49:02 +03:00
// Now that all addresses are parsed, check if all fromid login addresses match
// configured addresses.
for i , a := range acc . ParsedFromIDLoginAddresses {
// For domain catchall.
if _ , ok := accDests [ "@" + a . Domain . Name ( ) ] ; ok {
continue
}
dc := c . Domains [ a . Domain . Name ( ) ]
lp , err := CanonicalLocalpart ( a . Localpart , dc )
if err != nil {
addErrorf ( "canonicalizing localpart for fromid login address %q in account %q: %v" , acc . FromIDLoginAddresses [ i ] , accName , err )
continue
}
a . Localpart = lp
if _ , ok := accDests [ a . Pack ( true ) ] ; ! ok {
addErrorf ( "fromid login address %q for account %q does not match its destination addresses" , acc . FromIDLoginAddresses [ i ] , accName )
}
}
new feature: when delivering messages from the queue, make it possible to use a "transport"
the default transport is still just "direct delivery", where we connect to the
destination domain's MX servers.
other transports are:
- regular smtp without authentication, this is relaying to a smarthost.
- submission with authentication, e.g. to a third party email sending service.
- direct delivery, but with with connections going through a socks proxy. this
can be helpful if your ip is blocked, you need to get email out, and you have
another IP that isn't blocked.
keep in mind that for all of the above, appropriate SPF/DKIM settings have to
be configured. the "dnscheck" for a domain does a check for any SOCKS IP in the
SPF record. SPF for smtp/submission (ranges? includes?) and any DKIM
requirements cannot really be checked.
which transport is used can be configured through routes. routes can be set on
an account, a domain, or globally. the routes are evaluated in that order, with
the first match selecting the transport. these routes are evaluated for each
delivery attempt. common selection criteria are recipient domain and sender
domain, but also which delivery attempt this is. you could configured mox to
attempt sending through a 3rd party from the 4th attempt onwards.
routes and transports are optional. if no route matches, or an empty/zero
transport is selected, normal direct delivery is done.
we could already "submit" emails with 3rd party accounts with "sendmail". but
we now support more SASL authentication mechanisms with SMTP (not only PLAIN,
but also SCRAM-SHA-256, SCRAM-SHA-1 and CRAM-MD5), which sendmail now also
supports. sendmail will use the most secure mechanism supported by the server,
or the explicitly configured mechanism.
for issue #36 by dmikushin. also based on earlier discussion on hackernews.
2023-06-16 19:38:28 +03:00
checkRoutes ( "routes for account" , acc . Routes )
2023-01-30 16:27:06 +03:00
}
// Set DMARC destinations.
for d , domain := range c . Domains {
dmarc := domain . DMARC
if dmarc == nil {
continue
}
if _ , ok := c . Accounts [ dmarc . Account ] ; ! ok {
addErrorf ( "DMARC account %q does not exist" , dmarc . Account )
}
lp , err := smtp . ParseLocalpart ( dmarc . Localpart )
if err != nil {
addErrorf ( "invalid DMARC localpart %q: %s" , dmarc . Localpart , err )
}
if lp . IsInternational ( ) {
// ../rfc/8616:234
addErrorf ( "DMARC localpart %q is an internationalized address, only conventional ascii-only address possible for interopability" , lp )
}
2023-08-23 15:27:21 +03:00
addrdom := domain . Domain
if dmarc . Domain != "" {
addrdom , err = dns . ParseDomain ( dmarc . Domain )
if err != nil {
addErrorf ( "DMARC domain %q: %s" , dmarc . Domain , err )
} else if _ , ok := c . Domains [ addrdom . Name ( ) ] ; ! ok {
2024-01-26 21:51:23 +03:00
addErrorf ( "unknown domain %q for DMARC address in domain %q" , addrdom , d )
2023-08-23 15:27:21 +03:00
}
}
2024-01-26 21:51:23 +03:00
if addrdom == domain . Domain {
domainHasAddress [ addrdom . Name ( ) ] = true
}
2023-08-23 15:27:21 +03:00
2023-01-30 16:27:06 +03:00
domain . DMARC . ParsedLocalpart = lp
2023-08-23 15:27:21 +03:00
domain . DMARC . DNSDomain = addrdom
2023-01-30 16:27:06 +03:00
c . Domains [ d ] = domain
2023-08-23 15:27:21 +03:00
addrFull := smtp . NewAddress ( lp , addrdom ) . String ( )
2023-01-30 16:27:06 +03:00
dest := config . Destination {
Mailbox : dmarc . Mailbox ,
DMARCReports : true ,
}
checkMailboxNormf ( dmarc . Mailbox , "DMARC mailbox for account %q" , dmarc . Account )
2023-03-29 22:11:43 +03:00
accDests [ addrFull ] = AccountDestination { false , lp , dmarc . Account , dest }
2023-01-30 16:27:06 +03:00
}
// Set TLSRPT destinations.
for d , domain := range c . Domains {
tlsrpt := domain . TLSRPT
if tlsrpt == nil {
continue
}
if _ , ok := c . Accounts [ tlsrpt . Account ] ; ! ok {
addErrorf ( "TLSRPT account %q does not exist" , tlsrpt . Account )
}
lp , err := smtp . ParseLocalpart ( tlsrpt . Localpart )
if err != nil {
addErrorf ( "invalid TLSRPT localpart %q: %s" , tlsrpt . Localpart , err )
}
if lp . IsInternational ( ) {
// Does not appear documented in ../rfc/8460, but similar to DMARC it makes sense
// to keep this ascii-only addresses.
addErrorf ( "TLSRPT localpart %q is an internationalized address, only conventional ascii-only address allowed for interopability" , lp )
}
2023-08-23 15:27:21 +03:00
addrdom := domain . Domain
if tlsrpt . Domain != "" {
addrdom , err = dns . ParseDomain ( tlsrpt . Domain )
if err != nil {
addErrorf ( "TLSRPT domain %q: %s" , tlsrpt . Domain , err )
} else if _ , ok := c . Domains [ addrdom . Name ( ) ] ; ! ok {
addErrorf ( "unknown domain %q for TLSRPT address in domain %q" , tlsrpt . Domain , d )
}
}
2024-01-26 21:51:23 +03:00
if addrdom == domain . Domain {
domainHasAddress [ addrdom . Name ( ) ] = true
}
2023-08-23 15:27:21 +03:00
2023-01-30 16:27:06 +03:00
domain . TLSRPT . ParsedLocalpart = lp
2023-08-23 15:27:21 +03:00
domain . TLSRPT . DNSDomain = addrdom
2023-01-30 16:27:06 +03:00
c . Domains [ d ] = domain
2023-08-23 15:27:21 +03:00
addrFull := smtp . NewAddress ( lp , addrdom ) . String ( )
2023-01-30 16:27:06 +03:00
dest := config . Destination {
implement outgoing tls reports
we were already accepting, processing and displaying incoming tls reports. now
we start tracking TLS connection and security-policy-related errors for
outgoing message deliveries as well. we send reports once a day, to the
reporting addresses specified in TLSRPT records (rua) of a policy domain. these
reports are about MTA-STS policies and/or DANE policies, and about
STARTTLS-related failures.
sending reports is enabled by default, but can be disabled through setting
NoOutgoingTLSReports in mox.conf.
only at the end of the implementation process came the realization that the
TLSRPT policy domain for DANE (MX) hosts are separate from the TLSRPT policy
for the recipient domain, and that MTA-STS and DANE TLS/policy results are
typically delivered in separate reports. so MX hosts need their own TLSRPT
policies.
config for the per-host TLSRPT policy should be added to mox.conf for existing
installs, in field HostTLSRPT. it is automatically configured by quickstart for
new installs. with a HostTLSRPT config, the "dns records" and "dns check" admin
pages now suggest the per-host TLSRPT record. by creating that record, you're
requesting TLS reports about your MX host.
gathering all the TLS/policy results is somewhat tricky. the tentacles go
throughout the code. the positive result is that the TLS/policy-related code
had to be cleaned up a bit. for example, the smtpclient TLS modes now reflect
reality better, with independent settings about whether PKIX and/or DANE
verification has to be done, and/or whether verification errors have to be
ignored (e.g. for tls-required: no header). also, cached mtasts policies of
mode "none" are now cleaned up once the MTA-STS DNS record goes away.
2023-11-09 19:40:46 +03:00
Mailbox : tlsrpt . Mailbox ,
DomainTLSReports : true ,
2023-01-30 16:27:06 +03:00
}
checkMailboxNormf ( tlsrpt . Mailbox , "TLSRPT mailbox for account %q" , tlsrpt . Account )
2023-03-29 22:11:43 +03:00
accDests [ addrFull ] = AccountDestination { false , lp , tlsrpt . Account , dest }
2023-01-30 16:27:06 +03:00
}
2023-03-01 00:12:27 +03:00
2024-01-26 21:51:23 +03:00
// Set ReportsOnly for domains, based on whether we have seen addresses (possibly
// from DMARC or TLS reporting).
for d , domain := range c . Domains {
domain . ReportsOnly = ! domainHasAddress [ domain . Domain . Name ( ) ]
c . Domains [ d ] = domain
}
2023-03-01 00:12:27 +03:00
// Check webserver configs.
improve webserver, add domain redirects (aliases), add tests and admin page ui to manage the config
- make builtin http handlers serve on specific domains, such as for mta-sts, so
e.g. /.well-known/mta-sts.txt isn't served on all domains.
- add logging of a few more fields in access logging.
- small tweaks/bug fixes in webserver request handling.
- add config option for redirecting entire domains to another (common enough).
- split httpserver metric into two: one for duration until writing header (i.e.
performance of server), another for duration until full response is sent to
client (i.e. performance as perceived by users).
- add admin ui, a new page for managing the configs. after making changes
and hitting "save", the changes take effect immediately. the page itself
doesn't look very well-designed (many input fields, makes it look messy). i
have an idea to improve it (explained in admin.html as todo) by making the
layout look just like the config file. not urgent though.
i've already changed my websites/webapps over.
the idea of adding a webserver is to take away a (the) reason for folks to want
to complicate their mox setup by running an other webserver on the same machine.
i think the current webserver implementation can already serve most common use
cases. with a few more tweaks (feedback needed!) we should be able to get to 95%
of the use cases. the reverse proxy can take care of the remaining 5%.
nevertheless, a next step is still to change the quickstart to make it easier
for folks to run with an existing webserver, with existing tls certs/keys.
that's how this relates to issue #5.
2023-03-02 20:15:54 +03:00
if ( len ( c . WebDomainRedirects ) > 0 || len ( c . WebHandlers ) > 0 ) && ! haveWebserverListener {
addErrorf ( "WebDomainRedirects or WebHandlers configured but no listener with WebserverHTTP or WebserverHTTPS enabled" )
}
c . WebDNSDomainRedirects = map [ dns . Domain ] dns . Domain { }
for from , to := range c . WebDomainRedirects {
fromdom , err := dns . ParseDomain ( from )
if err != nil {
addErrorf ( "parsing domain for redirect %s: %v" , from , err )
}
todom , err := dns . ParseDomain ( to )
if err != nil {
addErrorf ( "parsing domain for redirect %s: %v" , to , err )
} else if fromdom == todom {
addErrorf ( "will not redirect domain %s to itself" , todom )
}
var zerodom dns . Domain
if _ , ok := c . WebDNSDomainRedirects [ fromdom ] ; ok && fromdom != zerodom {
addErrorf ( "duplicate redirect domain %s" , from )
}
c . WebDNSDomainRedirects [ fromdom ] = todom
}
2023-03-01 00:12:27 +03:00
for i := range c . WebHandlers {
wh := & c . WebHandlers [ i ]
if wh . LogName == "" {
wh . Name = fmt . Sprintf ( "%d" , i )
} else {
wh . Name = wh . LogName
}
dom , err := dns . ParseDomain ( wh . Domain )
if err != nil {
addErrorf ( "webhandler %s %s: parsing domain: %v" , wh . Domain , wh . PathRegexp , err )
}
wh . DNSDomain = dom
if ! strings . HasPrefix ( wh . PathRegexp , "^" ) {
addErrorf ( "webhandler %s %s: path regexp must start with a ^" , wh . Domain , wh . PathRegexp )
}
re , err := regexp . Compile ( wh . PathRegexp )
if err != nil {
addErrorf ( "webhandler %s %s: compiling regexp: %v" , wh . Domain , wh . PathRegexp , err )
}
wh . Path = re
var n int
if wh . WebStatic != nil {
n ++
ws := wh . WebStatic
if ws . StripPrefix != "" && ! strings . HasPrefix ( ws . StripPrefix , "/" ) {
addErrorf ( "webstatic %s %s: prefix to strip %s must start with a slash" , wh . Domain , wh . PathRegexp , ws . StripPrefix )
}
for k := range ws . ResponseHeaders {
xk := k
k := strings . TrimSpace ( xk )
if k != xk || k == "" {
addErrorf ( "webstatic %s %s: bad header %q" , wh . Domain , wh . PathRegexp , xk )
}
}
}
if wh . WebRedirect != nil {
n ++
wr := wh . WebRedirect
if wr . BaseURL != "" {
u , err := url . Parse ( wr . BaseURL )
if err != nil {
addErrorf ( "webredirect %s %s: parsing redirect url %s: %v" , wh . Domain , wh . PathRegexp , wr . BaseURL , err )
}
switch u . Path {
case "" , "/" :
u . Path = "/"
default :
addErrorf ( "webredirect %s %s: BaseURL must have empty path" , wh . Domain , wh . PathRegexp , wr . BaseURL )
}
wr . URL = u
}
if wr . OrigPathRegexp != "" && wr . ReplacePath != "" {
re , err := regexp . Compile ( wr . OrigPathRegexp )
if err != nil {
addErrorf ( "webredirect %s %s: compiling regexp %s: %v" , wh . Domain , wh . PathRegexp , wr . OrigPathRegexp , err )
}
wr . OrigPath = re
} else if wr . OrigPathRegexp != "" || wr . ReplacePath != "" {
addErrorf ( "webredirect %s %s: must have either both OrigPathRegexp and ReplacePath, or neither" , wh . Domain , wh . PathRegexp )
} else if wr . BaseURL == "" {
addErrorf ( "webredirect %s %s: must at least one of BaseURL and OrigPathRegexp+ReplacePath" , wh . Domain , wh . PathRegexp )
}
if wr . StatusCode != 0 && ( wr . StatusCode < 300 || wr . StatusCode >= 400 ) {
addErrorf ( "webredirect %s %s: invalid redirect status code %d" , wh . Domain , wh . PathRegexp , wr . StatusCode )
}
}
if wh . WebForward != nil {
n ++
wf := wh . WebForward
u , err := url . Parse ( wf . URL )
if err != nil {
addErrorf ( "webforward %s %s: parsing url %s: %v" , wh . Domain , wh . PathRegexp , wf . URL , err )
}
wf . TargetURL = u
for k := range wf . ResponseHeaders {
xk := k
k := strings . TrimSpace ( xk )
if k != xk || k == "" {
addErrorf ( "webforward %s %s: bad header %q" , wh . Domain , wh . PathRegexp , xk )
}
}
}
if n != 1 {
addErrorf ( "webhandler %s %s: must have exactly one handler, not %d" , wh . Domain , wh . PathRegexp , n )
}
}
2024-03-07 13:26:53 +03:00
c . MonitorDNSBLZones = nil
2024-03-05 18:30:38 +03:00
for _ , s := range c . MonitorDNSBLs {
d , err := dns . ParseDomain ( s )
if err != nil {
addErrorf ( "invalid monitor dnsbl zone %s: %v" , s , err )
continue
}
if slices . Contains ( c . MonitorDNSBLZones , d ) {
addErrorf ( "duplicate zone %s in monitor dnsbl zones" , d )
continue
}
c . MonitorDNSBLZones = append ( c . MonitorDNSBLZones , d )
}
2023-01-30 16:27:06 +03:00
return
}
implement dnssec-awareness throughout code, and dane for incoming/outgoing mail delivery
the vendored dns resolver code is a copy of the go stdlib dns resolver, with
awareness of the "authentic data" (i.e. dnssec secure) added, as well as support
for enhanced dns errors, and looking up tlsa records (for dane). ideally it
would be upstreamed, but the chances seem slim.
dnssec-awareness is added to all packages, e.g. spf, dkim, dmarc, iprev. their
dnssec status is added to the Received message headers for incoming email.
but the main reason to add dnssec was for implementing dane. with dane, the
verification of tls certificates can be done through certificates/public keys
published in dns (in the tlsa records). this only makes sense (is trustworthy)
if those dns records can be verified to be authentic.
mox now applies dane to delivering messages over smtp. mox already implemented
mta-sts for webpki/pkix-verification of certificates against the (large) pool
of CA's, and still enforces those policies when present. but it now also checks
for dane records, and will verify those if present. if dane and mta-sts are
both absent, the regular opportunistic tls with starttls is still done. and the
fallback to plaintext is also still done.
mox also makes it easy to setup dane for incoming deliveries, so other servers
can deliver with dane tls certificate verification. the quickstart now
generates private keys that are used when requesting certificates with acme.
the private keys are pre-generated because they must be static and known during
setup, because their public keys must be published in tlsa records in dns.
autocert would generate private keys on its own, so had to be forked to add the
option to provide the private key when requesting a new certificate. hopefully
upstream will accept the change and we can drop the fork.
with this change, using the quickstart to setup a new mox instance, the checks
at internet.nl result in a 100% score, provided the domain is dnssec-signed and
the network doesn't have any issues.
2023-10-10 13:09:35 +03:00
func loadPrivateKeyFile ( keyPath string ) ( crypto . Signer , error ) {
keyBuf , err := os . ReadFile ( keyPath )
if err != nil {
return nil , fmt . Errorf ( "reading host private key: %v" , err )
}
b , _ := pem . Decode ( keyBuf )
if b == nil {
return nil , fmt . Errorf ( "parsing pem block for private key: %v" , err )
}
var privKey any
switch b . Type {
case "PRIVATE KEY" :
privKey , err = x509 . ParsePKCS8PrivateKey ( b . Bytes )
case "RSA PRIVATE KEY" :
privKey , err = x509 . ParsePKCS1PrivateKey ( b . Bytes )
case "EC PRIVATE KEY" :
privKey , err = x509 . ParseECPrivateKey ( b . Bytes )
default :
err = fmt . Errorf ( "unknown pem type %q" , b . Type )
}
if err != nil {
return nil , fmt . Errorf ( "parsing private key: %v" , err )
}
if k , ok := privKey . ( crypto . Signer ) ; ok {
return k , nil
}
return nil , fmt . Errorf ( "parsed private key not a crypto.Signer, but %T" , privKey )
}
2023-01-30 16:27:06 +03:00
func loadTLSKeyCerts ( configFile , kind string , ctls * config . TLS ) error {
certs := [ ] tls . Certificate { }
for _ , kp := range ctls . KeyCerts {
certPath := configDirPath ( configFile , kp . CertFile )
keyPath := configDirPath ( configFile , kp . KeyFile )
2023-05-31 15:09:53 +03:00
cert , err := loadX509KeyPairPrivileged ( certPath , keyPath )
2023-01-30 16:27:06 +03:00
if err != nil {
return fmt . Errorf ( "tls config for %q: parsing x509 key pair: %v" , kind , err )
}
certs = append ( certs , cert )
}
ctls . Config = & tls . Config {
Certificates : certs ,
}
return nil
}
2023-05-31 15:09:53 +03:00
// load x509 key/cert files from file descriptor possibly passed in by privileged
// process.
func loadX509KeyPairPrivileged ( certPath , keyPath string ) ( tls . Certificate , error ) {
certBuf , err := readFilePrivileged ( certPath )
if err != nil {
return tls . Certificate { } , fmt . Errorf ( "reading tls certificate: %v" , err )
}
keyBuf , err := readFilePrivileged ( keyPath )
if err != nil {
return tls . Certificate { } , fmt . Errorf ( "reading tls key: %v" , err )
}
return tls . X509KeyPair ( certBuf , keyBuf )
}
// like os.ReadFile, but open privileged file possibly passed in by root process.
func readFilePrivileged ( path string ) ( [ ] byte , error ) {
f , err := OpenPrivileged ( path )
if err != nil {
return nil , err
}
defer f . Close ( )
return io . ReadAll ( f )
}