This change will force everyone to be logged out of Roundcube since the encryption key and cipher won't match anyone's already-set cookie, but this happens anyway after every Mail-in-a-Box update since we generate a new key each time already.
Fixes#1968.
* Stop generating RSASHA1-NSEC3-SHA1 keys on new installs since it is no longer recommended, but preserve the key on existing installs so that we continue to sign zones with existing keys to retain the chain of trust with existing DS records.
* Start generating ECDSAP256SHA256 keys during setup, the current best practice (in addition to RSASHA256 which is also ok). See https://www.iana.org/assignments/dns-sec-alg-numbers/dns-sec-alg-numbers.xhtml#dns-sec-alg-numbers-1 and https://www.cloudflare.com/dns/dnssec/ecdsa-and-dnssec/.
* Sign zones using all available keys rather than choosing just one based on the TLD to enable rotation/migration to the new key and to give the user some options since not every registrar/TLD supports every algorithm.
* Allow a user to drop a key from signing specific domains using DOMAINS= in our key configuration file. Signing the zones with extraneous keys may increase the size of DNS responses, which isn't ideal, although I don't know if this is a problem in practice. (Although a user can delete the RSASHA1-NSEC3-SHA1 key file, the other keys will be re-generated on upgrade.)
* When generating zonefiles, add a hash of all of the DNSSEC signing keys so that when the keys change the zone is definitely regenerated and re-signed.
* In status checks, if DNSSEC is not active (or not valid), offer to use all of the keys that have been generated (for RSASHA1-NSEC3-SHA1 on existing installs, RSASHA256, and now ECDSAP256SHA256) with all digest types, since not all registers support everything, but list them in an order that guides users to the best practice.
* In status checks, if the deployed DS record doesn't use a ECDSAP256SHA256 key, prompt the user to update their DS record.
* In status checks, if multiple DS records are set, only fail if none are valid. If some use ECDSAP256SHA256 and some don't, remind the user to delete the DS records that don't.
* Don't fail if the DS record uses the SHA384 digest (by pre-generating a DS record with that digest type) but don't recommend it because it is not in the IANA mandatory list yet (https://www.iana.org/assignments/ds-rr-types/ds-rr-types.xhtml).
See #1953
Configures opendmarc to send failure reports for domains that request them, including when p=none.
The emails are sent as the package default of package name and user@hostname: OpenDMARC Filter <opendmarc@box.example.com>
Note I have been running this for several months with a configuration I did not include in the PR to have reports BCC'd to me (FailureReportsBcc postmaster@example.com). Very low load for my personal server of rarely more than a dozen emails sent out per day.
I am not familiar with editing scripts, so apologies in advance and please feel free to correct me.
roundcube Bug Fixes:
Fix for Cross-Site Scripting (XSS) via HTML messages with malicious CSS content
General Improvements from roundcube's Issue Tracker
This reverts commit b1d703a5e7 and adds python3-setuptools per the first version of #1899 which fixes an installation error for the b2sdk Python package.
* Installing b2sdk for b2 support
* Added Duplicity PPA so the most recent version is used
* Implemented list_target_files for b2
* Implemented b2 in frontend
* removed python2 boto package
* add user interface for managing 2fa
* update user schema with 2fa columns
* implement two factor check during login
* Use pyotp for validating TOTP codes
* also implements resynchronisation support via `pyotp`'s `valid_window option
* Update API route naming, update setup page
* Rename /two-factor-auth/ => /2fa/
* Nest totp routes under /2fa/totp/
* Update ids and methods in panel to allow for different setup types
* Autofocus otp input when logging in, update layout
* Extract TOTPStrategy class to totp.py
* this decouples `TOTP` validation and storage logic from `auth` and moves it to `totp`
* reduce `pyotp.validate#valid_window` from `2` to `1`
* Update OpenApi docs, rename /2fa/ => /mfa/
* Decouple totp from users table by moving to totp_credentials table
* this allows implementation of other mfa schemes in the future (webauthn)
* also makes key management easier and enforces one totp credentials per user on db-level
* Add sqlite migration
* Rename internal validate_two_factor_secret => validate_two_factor_secret
* conn.close() if mru_token update can't .commit()
* Address review feedback, thanks @hija
* Use hmac.compare_digest() to compare mru_token
* Safeguard against empty mru_token column
* hmac.compare_digest() expects arguments of type string, make sure we don't pass None
* Currently, this cannot happen but we might not want to store `mru_token` during setup
* Do not log failed login attempts for MissingToken errors
* Due to the way that the /login UI works, this persists at least one failed login each time a user logs into the admin panel. This in turn triggers fail2ban at some point.
* Add TOTP secret to user_key hash
thanks @downtownallday
* this invalidates all user_keys after TOTP status is changed for user
* after changing TOTP state, a login is required
* due to the forced login, we can't and don't need to store the code used for setup in `mru_code`
* Typo
* Reorganize the MFA backend methods
* Reorganize MFA front-end and add label column
* Fix handling of bad input when enabling mfa
* Update openAPI docs
* Remove unique key constraint on foreign key user_id in mfa table
* Don't expose mru_token and secret for enabled mfas over HTTP
* Only update mru_token for matched mfa row
* Exclude mru_token in user key hash
* Rename tools/mail.py to management/cli.py
* Add MFA list/disable to the management CLI so admins can restore access if MFA device is lost
Co-authored-by: Joshua Tauberer <jt@occams.info>
v0.50 (September 25, 2020)
--------------------------
Setup:
* When upgrading from versions before v0.40, setup will now warn that ownCloud/Nextcloud data cannot be migrated rather than failing the installation.
Mail:
* An MTA-STS policy for incoming mail is now published (in DNS and over HTTPS) when the primary hostname and email address domain both have a signed TLS certificate installed, allowing senders to know that an encrypted connection should be enforced.
* The per-IP connection limit to the IMAP server has been doubled to allow more devices to connect at once, especially with multiple users behind a NAT.
DNS:
* autoconfig and autodiscover subdomains and CalDAV/CardDAV SRV records are no longer generated for domains that don't have user accounts since they are unnecessary.
* IPv6 addresses can now be specified for secondary DNS nameservers in the control panel.
TLS:
* TLS certificates are now provisioned in groups by parent domain to limit easy domain enumeration and make provisioning more resilient to errors for particular domains.
Control Panel:
* The control panel API is now fully documented at https://mailinabox.email/api-docs.html.
* User passwords can now have spaces.
* Status checks for automatic subdomains have been moved into the section for the parent domain.
* Typo fixed.
Web:
* The default web page served on fresh installations now adds the `noindex` meta tag.
* The HSTS header is revised to also be sent on non-success responses.
* this allows implementation of other mfa schemes in the future (webauthn)
* also makes key management easier and enforces one totp credentials per user on db-level
- The MIAB version check regularly fails at 03:00, presumably because a
large portion of installations is checking mailinabox.email at the same
time.
- At installation time, the time of the nightly clock is configured to
run at a random minute after 03:00, but before 04:00.
- Users might expect the nightly tasks to be over at a certain time and
run their own custom tasks afterwards. This could thus interfere with
custom backup routines.
- This breaks reproducibility of the installation process.
- Users might also be surprised by the nightly task time changing after
updating MIAB.
The function apt_add_repository_to_unattended_upgrades is defined
but never called anywhere. It appears that automatic apt updates
are handled in system.sh where the file /etc/apt/apt.conf.d/02periodic
is created. The last call was removed in bbfa01f33a.
Co-authored-by: ddavis32 <dan@nthdegreesoftware.com>
Because Mailman reformats headers it breaks DKIM signatures. SPF also does
not apply in mailing lists. This together causes DMARC to fail and mark the
email as invalid. This fixes DKIM signatures for Mailman-based mailing lists
and makes sure DMARC test is passed.
A minimal Ubuntu server installation might not have universe enabled by
default. By adding it, we ensure we can install packages only available
in universe, such as python3-pip
Merges #1650.
* Download and verify Nextcloud download before deleting old install directory
* Changed install logic to look at config.php and not version.php for database version number. When restoring from a backup, config.php in STORAGE_ROOT will hold the Nextcloud version that corresponds to the user's database and version.php in /usr/local won't even exist, so we were missing Nextcloud migration steps. In other cases they should be the same.
* Update to Nextcloud 15.0.7, Contacts to 3.1.1, and Calendar to 1.6.5
* Enabled localhost-only insecure IMAP login for localhost Nextcloud auth
* Add package php-imagick and BigInt conversion
* added support for /cloud/oc[sm]-provider/ endpoint
* Upgraded Nextcloud from 13.0.6 to 14.0.6.
* Upgraded Contacts from 2.1.5 to 2.1.8.
* Upgraded Calendar from 1.6.1 to 1.6.4.
* Cleanup unsupported version upgrades: Since an upgrade to v0.30 is mandatory before moving upward, I removed the checks for Nextcloud prior version 12.
* Fix the storage root path.
* Add missing indices. Thx @yodax for your feedback.
* drop the ondrej/php PPA since PHP 7.x is available directly from Ubuntu 18.04
* intall PHP 7.2 which is just the "php" package in Ubuntu 18.04
* some package names changed, some unnecessary packages are no longer provided
* update paths
@joshdata squashed pull request #1398, removed some comments, and added these notes:
* The old init.d script for the management daemon is replaced with a systemd service.
* A systemd service configuration is added to configure permissions for munin on startup.
* nginx SSL settings are updated because nginx's options and defaults have changed, and we now enable http2.
* Automatic SSHFP record generation is updated to know that 22 is the default SSH daemon port, since it is no longer explicit in sshd_config.
* The dovecot-lucene package is dropped because the Mail-in-a-Box PPA where we built the package has not been updated for Ubuntu 18.04.
* The stock postgrey package is installed instead of the one from our PPA (which we no longer support), which loses the automatic whitelisting of DNSWL.org-whitelisted senders.
* Drop memcached and the status check for memcached, which we used to use with ownCloud long ago but are no longer installing.
* Other minor changes.
fixes#1409
This reverts commit 82844ca651 ("make certbot auto-agree to TOS if NONINTERACTIVE=1 env var is set (#1399)") and instead *always* auto-agree. If we don't auto-agree, certbot asks the user interactively, but our "curl | bash" setup line does not permit interactive prompts, so certbot failed to register and all certificate things were broken until the command was re-run interactively.
Starting with 2.4, Z-Push no longer provides tarballs on their download server. The only options are getting the code from their git repository or using one of their distribution packages. Their Ubuntu 18.04 packaes don't seem to actually work in Ubuntu 18.04, so thinking ahead that's currently a bad choice. In 78d1c9be6e we switched from doing a git clone to using wget on their downloads server because of a problem with something related to stash.z-hub.io's SSL certificate. But wget also seems to work on their source code repository, so we can use that.
git clone (which uses curl) underneath was failing. Curiously, the same
git clone command would work on my macos host machine.
From the screenshot it looks like curl was somehow not able to negotiate
the connection. Might have been a missing CA certificate for Comodo, but
I was not able to determine if that was the issue.
fixes#1393closes#1387closes#1400
Our wget_verify function uses wget to download a file and then check
the file's hash. If wget fails, i.e. because of a 404 or other HTTP
or network error, we exited setup without displaying any output because
normally there are no errors and -q keeps the setup output clean.
Wrapping wget with our hide_output function, and dropping -q, captures
wget's output and shows it and exits setup just if wget fails.
see #1297
In 0088fb4553 I changed the management daemon's startup
script from a symlink to a Python script to a bash script that activated the new virtualenv
and then launched Python. As a result, the init.d script that starts the daemon would
write the pid of bash to the pidfile, and when trying to kill it, it would kill bash but
not the Python process.
Using exec to start Python fixes this problem by making the Python process have the pid
that the init.d script knows about.
fixes#1339
