if no host keys are configured, show as warning (yellow) that dane isn't
configured, and show instructions to enable it.
for issue #185 by morki, thanks for reporting!
the suggested dns records mention that these records are optional, but the
dnscheck makes it look serious. not helpful.
also remove unneeded whitespace in list of errors/warnings.
for issue #184 by morki, thanks for reporting!
the host tlsrpt record implied it was for the domain, but should have been for
the mail host.
some dns records were absolute, others weren't. now they all are for
consistency.
for issue #182 by mdavids, thanks for reporting!
there were a few test failures on the github runners. i can't reproduce it
locally. but i can see how they are happening: a gorouting running servectlcmd
could still be doing cleanup (removing files) while a next ctl command was
being run. with this change, we wait for servectlcmd to be done before starting
on a next test.
otherwise, in dark mode, the plain text iframe content would be black text on
the white background of the iframe as set by webmail. i can't find a way to set
the content text on the iframe that contains it.
some commands, like search, can specify any number of literals, of arbitrary
size. we already limited individual literals to 100kb. but you could specify
many of them, causing unbounded memory consumption. this change adds a limit of
1000 literals in a command, and a limit of 1mb of total combined memory for
literals. once the limits are exceeded, a TOOBIG error code is returned.
unbounded memory use could only be triggered on authenticated connections.
this addresses the same issue as CVE-2024-34055 for cyrus-imap, by damian
poddebniak.
per listener, you could enable the admin/account/webmail/webapi handlers. but
that would serve those services on their configured paths (/admin/, /,
/webmail/, /webapi/) on all domains mox would be webserving, including any
non-mail domains. so your www.example/admin/ would be serving the admin web
interface, with no way to disabled that.
with this change, the admin interface is only served on requests to (based on
Host header):
- ip addresses
- the listener host name (explicitly configured in the listener, with fallback
to global hostname)
- "localhost" (for ssh tunnel/forwarding scenario's)
the account/webmail/webapi interfaces are served on the same domains as the
admin interface, and additionally:
- the client settings domains, as optionally configured in each Domain in
domains.conf. typically "mail.<yourdomain>".
this means the internal services are no longer served on other domains
configured in the webserver, e.g. www.example.org/admin/ will not be handled
specially.
the order of evaluation of routes/services is also changed:
before this change, the internal handlers would always be evaluated first.
with this change, only the system handlers for
MTA-STS/autoconfig/ACME-validation will be evaluated first. then the webserver
handlers. and finally the internal services (admin/account/webmail/webapi).
this allows an admin to configure overrides for some of the domains (per
hostname-matching rules explained above) that would normally serve these
services.
webserver handlers can now be configured that pass the request to an internal
service: in addition to the existing static/redirect/forward config options,
there is now an "internal" config option, naming the service
(admin/account/webmail/webapi) for handling the request. this allows enabling
the internal services on custom domains.
for issue #160 by TragicLifeHu, thanks for reporting!
by setting the loglevel to debug in package mlog.
we restore the "info" logging in main.
except for "mox localserve", which still sets debug by default.
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.
the protocol is often: read one or more lines. only then return error. if we
would return an error after reading 1 line, parsing it and failing, the writer
(client connecting) may be busy writing more lines, not reading an error
response yet.
in quickstart, add troubleshooting hints.
in suggested dns records, explain the multiline long dkim record should
probably be converted into a single string.
the quickstart output is quite long already. i'm hoping for a "mox setup" in
the future where we help a user step-by-step to a fully working system. we'll
have more space to prevent hints and check the settings after a user made
changes. it's on the roadmap.
based on issues #158 and #164, thanks vipas84 and RobSlgm for reporting and
discussion.
recreating the account would resurface the old messages, certainly not what you'ld expect.
it's about time to just remove the files. we do ask admins to confirm that when
removing through admin interface. it's also in the "mox config account rm" help
output now.
for issue #162 by RobSlgm with feedback from x8x, thanks!
several fields in dmarc and tls reports have known string values. we have a Go
string type for them. sherpats (through sherpadoc) turns those strings into
typescript enums, and sherpats generates runtime-typechecking code (to enforce
correct types for incoming json, to prevent failing deeper in the code when we
get invalid data (much harder to debug)). the Go not-really-enum types allow
other values, and real-world reports have unknown/unspecified/invalid values.
this uses the sherpadoc -rename flag to turn those enums into regular untyped
strings, so sherpats doesn't generate enum-enforcing runtime type checking
code.
this required an update to sherpadoc, to properly handling renaming a type to a
basic type instead of another named type.
for issue #161 by RobSlgm, thanks for reporting!
- keep them on the right side of the window (more important now that we can resize)
- merge the close & cancel buttons into a close button, with a popup asking what to do for changes not saved as draft.
- the error wasn't caught because the parseInt() call wasn't evaluated inside the promise.
- we no longer require that the input (e.g. 2G) is the same as how we would format it (2g).
- tooltips and error message should now steer people to using these units.
feedback from pmarini-nc in #115, thanks!
this started with looking into the dark mode of PR #163 by mattfbacon. it's a
very good solution, especially for the amount of code. while looking into dark
mode, some common problems with inverting colors are:
- box-shadow start "glowing" which isn't great. likewise, semitransparent
layers would become brighter, not darker.
- while popups/overlays in light mode just stay the same white, in dark mode
they should become lighter than the regular content because box shadows don't
give enough contrast in dark mode.
while looking at adding explicit styles for dark mode, it turns out that's
easier when we work more with css rules/classes instead of inline styles (so we
can use the @media rule).
so we now also create css rules instead of working with inline styles a lot.
benefits:
- creating css rules is useful for items that repeat. they'll have a single css
class. changing a style on a css class is now reflected in all elements of that
kind (with that class)
- css class names are helpful when inspecting the DOM while developing: they
typically describe the function of the element.
most css classes are defined near where they are used, often while making the
element using the class (the css rule is created on first use).
this changes moves colors used for styling to a single place in webmail/lib.ts.
each property can get two values: one for regular/light mode, one for dark mode.
that should prevent forgetting one of them and makes it easy to configure both.
this change sets colors for the dark mode. i think the popups look better than
in PR #163, but in other ways it may be worse. this is a start, we can tweak
the styling.
if we can reduce the number of needed colors some more, we could make them
configurable in the webmail settings in the future. so this is also a step
towards making the ui looks configurable as discussed in issue #107.
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.
it's the responsibility of the sender to use unique fromid's.
we do check if that's the case, and return an error if not.
also make it more clear that "unique smtp mail from addresses" map to the
"FromIDLoginAddresses" account config field.
based on feedback from cuu508 for #31, thanks!
unless the address is the last member, then the admin must either remove the
alias first, or add new members. we don't want to accidentally remove an alias
address.
in the admin page for removing addresses, we warn the admin that the address
will be removed from any aliases.
before, the smtpserver that queued a dsn would set an empty senderaccount,
which was interpreted in a few places as the globally configured postmaster
cacount. the empty senderaccount would be used by the smtpserver that queued a
dsn with null return path. we now set the postmaster account when we add a
message to the queue. more code in the queue pretty much needs a non-empty
senderaccount, such as the filters when listing, and the suppression list.
since email addresses can contain multiple consecutive spaces.
this is a valid address: " "@localhost
and this is a different valid address: " "@localhost
webmail still todo
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.
