Example symptom, when deleting a message in the webmail (which moves to Trash):
l=error m="duplicating message in old mailbox for current sessions" err="link data/accounts/mjl/msg/I/368638 data/accounts/mjl/msg/J/368640: no such file or directory" pkg=webmail
Problem introduced a few weeks ago, where moving messages starting duplicating
the message first, and the copy is erased once all references (in IMAP
sessions) to the old mailbox have been removed.
The intent to remove the account is stored in the database. At startup, if
there are any such referenes, they are applied by removing the account
directories and the entry in the database. This ensures the account directory
is properly removed, even on incomplete shutdown.
Don't add an account when its directory already exits.
Initialize store and switchboard first, then open account, and close in reverse
order.
Replace all "CheckClosed" calls with "WaitClosed", future changings will keep
an account reference open for a bit after the last regular close, so we can't
know that an account should be closed during tests.
Remove one parameter from the (still too long) "start test server" function in
imapserver testing code.
We weren't appending the individual changes to the slice, but the entire slice.
Since "Change" is an "any", this isn't a type error. So make a Change a
non-empty interface (I had seen an issue like this coming, should have made it
an interface then, at least now we have a reasonable method, to get the modseq
of a change).
Found while working on an imap webpush prototype.
Since a recent change (likely since implementing MULTIAPPEND), the temporary
files weren't removed any more. When changing it, I must have had the wrong
mental model about the MessageAdd method, assuming it would remove the temp
file.
Noticed during tests.
In case the precis check failed, our return of a nil account cleared acc, and
we were then trying to close it, returning in a nil pointer dereference.
Rewrite the return statements so we don't overwrite the named return variables.
Just to be cautious. This hasn't happened yet in practice that I'm aware of.
But in theory, mox could crash after it has written the message file during
delivery, but before the database transaction was committed. In that case, a
message file for the "next message id" is already present. Any future delivery
attempts will get assigned the same message id by the database, but writing the
file will fail because there already is one, causing delivery to fail (until
the file is moved away).
When opening an account, we now check in the file system if newer files exist
than we expect based on the last existing message in the database. If so, we
adjust the message ID the database will assign next.
Keeping the message files around, and the message details in the database, is
useful for IMAP sessions that haven't seen/processed the removal of a message
yet and try to fetch it. Before, we would return errors. Similarly, a session
that has a mailbox selected that is removed can (at least in theory) still read
messages.
The mechanics to do this need keeping removed mailboxes around too. JMAP needs
that anyway, so we now keep modseq/createseq/expunged history for mailboxes
too. And while we're at it, for annotations as well.
For future JMAP support, we now also keep the mailbox parent id around for a
mailbox, with an upgrade step to set the field for existing mailboxes and
fixing up potential missing parents (which could possibly have happened in an
obscure corner case that I doubt anyone ran into).
DeliverMessage() is now MessageAdd(), and it takes a Mailbox object that it
modifies but doesn't write to the database (the caller must do it, and plenty
of times can do it more efficiently by doing it once for multiple messages).
The new AddOpts let the caller influence how many checks and how much of the
work MessageAdd() does. The zero-value AddOpts enable all checks and all the
work, but callers can take responsibility of some of the checks/work if it can
do it more efficiently itself.
This simplifies the code in most places, and makes it more efficient. The
checks to update per-mailbox keywords is a bit simpler too now.
We are also more careful to close the junk filter without saving it in case of
errors.
Still part of more upcoming changes.
In the common case, it's the same as the previous delivery. That means we don't
have to try to create the directory (fewer syscalls) and that we can sync the
dir to disk.
This also tweaks the defer handling in case of a late failure.
Fix up a test or two. Simplify the XOR logic when we train the junk filter:
Only if junk or nonjunk is set, but not when both (or none are set). i.e. when
the values aren't the same.
Locking the account when we do consistency checks prevents spurious test
failures that may have been introduced in the previous commit.
MULTIAPPEND modifies the existing APPEND command to allow multiple messages. it
is somewhat more involved than a regular append of a single message since the
operation (of adding multiple messages) must be atomic. either all are added,
or none are.
we check as early as possible if the messages won't cause an over-quota error.
not just /private. /shared/ is the more commonly implemented namespace, because
it is easier te implement: you don't need per-user/account storage of metadata.
i initially approached it from the other direction: we don't have a mechanism
to share metadata with other accounts, so everything is private, and i assumed
that would be what a user would prefer. but email clients make the decisions,
and they'll likely try the /shared/ namespace.
i added the metadata extension to the imapserver recently. then i wondered how
a client would efficiently find changed metadata. turns out the qresync rfc
mentions that metadata changes should set a new modseq on the mailbox.
shouldn't be hard, except that we were not explicitly keeping track of modseqs
per mailbox. we only kept them for messages, and we were just looking up the
latest message modseq when we needed the modseq (we keep db entries for
expunged messages, so this worked out fine). that approach isn't enough
anymore. so know we keep track of modseq & createseq for mailboxes, just as for
messages. and we also track modseq/createseq for annotations. there's a good
chance jmap is going to need it.
this also adds consistency checks for modseq/createseq on mailboxes and
annotations to the account storage. it helped spot cases i missed where the
values need to be updated.
they are intended to be used by the server to automatically mark some messages
as important, based on server-defined heuristics. we don't have such heuristics
at the moment. perhaps in the future, but until then there are no plans.
we already supported special-use flags. settable through the webmail interface,
and new accounts already got standard mailboxes with special-use flags
predefined. but now the IMAP "CREATE" command implements creating mailboxes
with special-use flags.
it makes a new field available on stored messages. not when they they were
received (over smtp) or appended to the mailbox (over imap), but when they were
last "saved" in the mailbox. copy/move of a message (eg to the trash) resets
the "savedate" value. this helps implement "remove messages from trash after X
days".
this allows setting per-mailbox and per-server annotations (metadata). we have
a fixed maximum for total number of annotations (1000) and their total size
(1000000 bytes). this size isn't held against the regular quota for simplicity.
we send unsolicited metadata responses when a connection is in the idle
command and a change to a metadata item is made.
we currently only implement the /private/ namespace. we should implement the
/shared/ namespace, for mox-global metadata annotations. only the admin should
be able to configure those, probably through the config file, cli, or admin web
interface.
for issue #290
accounts with this option enabled can only generate get a new randomly
generated password. this prevents password reuse across services and weak
passwords. existing accounts keep their current ability to set custom
passwords. only admins can change this setting for an account.
related to issue #286 by skyguy
and show them in the account and admin interfaces. this should help with
debugging, to find misconfigured clients, and potentially find attackers trying
to login.
we include details like login name, account name, protocol, authentication
mechanism, ip addresses, tls connection properties, user-agent. and of course
the result.
we group entries by their details. repeat connections don't cause new records
in the database, they just increase the count on the existing record.
we keep data for at most 30 days. and we keep at most 10k entries per account.
to prevent unbounded growth. for successful login attempts, we store them all
for 30d. if a bad user causes so many entries this becomes a problem, it will
be time to talk to the user...
there is no pagination/searching yet in the admin/account interfaces. so the
list may be long. we only show the 10 most recent login attempts by default.
the rest is only shown on a separate page.
there is no way yet to disable this. may come later, either as global setting
or per account.
to facilitate migrations from/to other mail setups.
a domain can be added in "disabled" mode (or can be disabled/enabled later on).
you can configure a disabled domain, but incoming/outgoing messages involving
the domain are rejected with temporary error codes (as this may occur during a
migration, remote servers will try again, hopefully to the correct machine or
after this machine has been configured correctly). also, no acme tls certs will
be requested for disabled domains (the autoconfig/mta-sts dns records may still
point to the current/previous machine). accounts with addresses at disabled
domains can still login, unless logins are disabled for their accounts.
an account now has an option to disable logins. you can specify an error
message to show. this will be shown in smtp, imap and the web interfaces. it
could contain a message about migrations, and possibly a URL to a page with
information about how to migrate. incoming/outgoing email involving accounts
with login disabled are still accepted/delivered as normal (unless the domain
involved in the messages is disabled too). account operations by the admin,
such as importing/exporting messages still works.
in the admin web interface, listings of domains/accounts show if they are disabled.
domains & accounts can be enabled/disabled through the config file, cli
commands and admin web interface.
for issue #175 by RobSlgm
these settings are applied anywhere the webmail is open. the settings are for
showing keyboard shortcuts in the lower right after a mouse interaction, and
showing additional headers. the shorcuts were configurable in the "help" popup
before. the additional headers were only configurable through the developer
console before.
the "mailto:" (un)register buttons are now in the settings popup too.
bstore was updated to v0.0.6 to add this logging.
this simplifies some of the db-handling code in mtastsdb,tlsrptdb,dmarcdb. we
now call the package-level Init() and Close() in all tests properly.
store/threads_test.go opens an account, starts the threading upgrade, waits for
it to finish, runs some tests, and closes the account at the end, verifying all
references are gone. the "thread upgrade" goroutine has its own account
reference. it closes its account after having signaled completion of the
upgrade. in between that time, all checks from the tests could run, its account
closed and its no-more-account-references check would fail. the fix is
hopefully to mark the thread upgrade process finished after closing the
account. hard to verify, but this only happens very rarely.
the members must currently all be addresses of local accounts.
a message sent to an alias is accepted if at least one of the members accepts
it. if no members accepts it (e.g. due to bad reputation of sender), the
message is rejected.
if a message is submitted to both an alias addresses and to recipients that are
members of the alias in an smtp transaction, the message will be delivered to
such members only once. the same applies if the address in the message
from-header is the address of a member: that member won't receive the message
(they sent it). this prevents duplicate messages.
aliases have three configuration options:
- PostPublic: whether anyone can send through the alias, or only members.
members-only lists can be useful inside organizations for internal
communication. public lists can be useful for support addresses.
- ListMembers: whether members can see the addresses of other members. this can
be seen in the account web interface. in the future, we could export this in
other ways, so clients can expand the list.
- AllowMsgFrom: whether messages can be sent through the alias with the alias
address used in the message from-header. the webmail knows it can use that
address, and will use it as from-address when replying to a message sent to
that address.
ideas for the future:
- allow external addresses as members. still with some restrictions, such as
requiring a valid dkim-signature so delivery has a chance to succeed. will
also need configuration of an admin that can receive any bounces.
- allow specifying specific members who can sent through the list (instead of
all members).
for github issue #57 by hmfaysal.
also relevant for #99 by naturalethic.
thanks to damir & marin from sartura for discussing requirements/features.
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.
when sending a message with bcc's, prepend the bcc header to the message we
store in the sent folder. still not in the message we send to the recipients.
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
an é (e with accent) can also be written as e+\u0301. the first form is NFC,
the second NFD. when logging in, we transform usernames (email addresses) to
NFC. so both forms will be accepted. if a client is using NFD, they can log
in too.
for passwords, we apply the PRECIS "opaquestring", which (despite the name)
transforms the value too: unicode spaces are replaced with ascii spaces. the
string is also normalized to NFC. PRECIS may reject confusing passwords when
you set a password.
both when parsing our configs, and for incoming on smtp or in messages.
so we properly compare things like é and e+accent as equal, and accept the
different encodings of that same address.
the http basic auth we had was very simple to reason about, and to implement.
but it has a major downside:
there is no way to logout, browsers keep sending credentials. ideally, browsers
themselves would show a button to stop sending credentials.
a related downside: the http auth mechanism doesn't indicate for which server
paths the credentials are.
another downside: the original password is sent to the server with each
request. though sending original passwords to web servers seems to be
considered normal.
our new approach uses session cookies, along with csrf values when we can. the
sessions are server-side managed, automatically extended on each use. this
makes it easy to invalidate sessions and keeps the frontend simpler (than with
long- vs short-term sessions and refreshing). the cookies are httponly,
samesite=strict, scoped to the path of the web interface. cookies are set
"secure" when set over https. the cookie is set by a successful call to Login.
a call to Logout invalidates a session. changing a password invalidates all
sessions for a user, but keeps the session with which the password was changed
alive. the csrf value is also random, and associated with the session cookie.
the csrf must be sent as header for api calls, or as parameter for direct form
posts (where we cannot set a custom header). rest-like calls made directly by
the browser, e.g. for images, don't have a csrf protection. the csrf value is
returned by the Login api call and stored in localstorage.
api calls without credentials return code "user:noAuth", and with bad
credentials return "user:badAuth". the api client recognizes this and triggers
a login. after a login, all auth-failed api calls are automatically retried.
only for "user:badAuth" is an error message displayed in the login form (e.g.
session expired).
in an ideal world, browsers would take care of most session management. a
server would indicate authentication is needed (like http basic auth), and the
browsers uses trusted ui to request credentials for the server & path. the
browser could use safer mechanism than sending original passwords to the
server, such as scram, along with a standard way to create sessions. for now,
web developers have to do authentication themselves: from showing the login
prompt, ensuring the right session/csrf cookies/localstorage/headers/etc are
sent with each request.
webauthn is a newer way to do authentication, perhaps we'll implement it in the
future. though hardware tokens aren't an attractive option for many users, and
it may be overkill as long as we still do old-fashioned authentication in smtp
& imap where passwords can be sent to the server.
for issue #58
the imapserver started with imap4rev2-only and utf8=only. to prevent potential
issues with imaputf7, which makes "&" special, we refused any mailbox with an
"&" in the name. we already tried decoding utf7, falling back to using a
mailbox name verbatim. that behaviour wasn't great. we now treat the enabled
extensions IMAP4rev2 and/or UTF8=ACCEPT as indication whether mailbox names are
in imaputf7. if they are, the encoding must be correct.
we now also send mailbox names in imaputf7 when imap4rev2/utf8=accept isn't
enabled.
and we now allow "*" and "%" (wildcard characters for matching) in mailbox
names. not ideal for IMAP LIST with patterns, but not enough reason to refuse
them in mailbox names. people that migrate may run into this, possibly as
blocker.
we also allow "#" in mailbox names, but not as first character, to prevent
potential clashes with IMAP namespaces in the future.
based on report from Damian Poddebniak using
https://github.com/duesee/imap-flow and issue #110, thanks for reporting!
so a single user cannot fill up the disk.
by default, there is (still) no limit. a default can be set in the config file
for all accounts, and a per-account max size can be set that would override any
global setting.
this does not take into account disk usage of the index database. and also not
of any file system overhead.
