* tests: Ensure excessive FD limits are avoided
Processes that run as daemons (`postsrsd` and `fail2ban-server`) initialize by closing all FDs (File Descriptors).
This behaviour queries that maximum limit and iterates through the entire range even if only a few FDs are open. In some environments (Docker, limit configured by distro) this can be a range exceeding 1 billion (from kernel default of 1024 soft, 4096 hard), causing an 8 minute delay with heavy CPU activity.
`postsrsd` has since been updated to use `close_range()` syscall, and `fail2ban` will now iterate through `/proc/self/fd` (open FDs) which should resolve the performance hit. Until those updates reach our Docker image, we need to workaround it with `--ulimit` option.
NOTE: If `docker.service` on a distro sets `LimitNOFILE=` to approx 1 million or lower, it should not be an issue. On distros such as Fedora 36, it is `LimitNOFILE=infinity` (approx 1 billion) that causes excessive delays.
* chore: Use Docker host limits instead
Typically on modern distros with systemd, this should equate to 1024 (soft) and 512K (hard) limits. A distro may override the built-in global defaults systemd sets via setting `DefaultLimitNOFILE=` in `/etc/systemd/user.conf` and `/etc/systemd/system.conf`.
* tests(fix): Better prevent non-deterministic failures
- `no_containers.bats` tests the external script `setup.sh` (without `-c`). It's expected that no existing DMS container is running - otherwise it may attempt to use that container and fail. Detect this and fail early via `setup_file()` step.
- `mail_hostname.bats` had a odd timing failure with teardown due to the last tests bringing the containers down earlier (`docker stop` paired with the `docker run --rm`). Adding a moment of delay via `sleep` helps avoid that false positive scenario.
## Quick Summary
Resolves a `TODO` task with `addmailuser`.
## Overview
The main change is adding three new methods in `common.bash`, which replace the completion delay in `addmailuser` / `setup email add` command.
Other than that:
- I swapped `sh -c 'addmailuser ...'` to `setup email add ...`.
- Improved three tests in `setup-cli.bats` for `setup email add|update|del` (_logic remains effectively the same still_).
- Rewrote the `TODO` comment for `setup-cli.bats` test on `setup email del` to better clarify the concern, but the test itself was no longer affected due to changes prior to this PR, so I enabled the commented out assertion.
- Removed unnecessary waits. The two `skip` tests in `test/tests.bats` could be enabled again after this PR.
- Additional fixes to tests were made during the PR (see discussion comments for details), resolving race conditions.
Individual commit messages of the PR provide additional details if helpful.
---
## Relevant commit messages
* chore: Remove creation delay in `addmailuser`
This was apparently only for supporting tests that need to wait on account creation being ready to test against.
As per the removed inline docs, it should be fine to remove once tests are updated to work correctly without it.
* tests(feat): Add two new common helper methods
`wait_until_account_maildir_exists()` provides the same logic `addmailuser` command was carrying, to wait upon the account dir creation in `/var/mail`.
As this was specifically to support tests, it makes more sense as a test method.
`add_mail_account_then_wait_until_ready()` was added to handle the common pattern of creating account and waiting on it. An internal assert will ensure the account was successfully created first during the test before attempting to wait.
* tests(feat): Add common helper for waiting on change event to be processed
The current helper is more complicated for no real benefit, it only detects when a change is made that would trigger a change event in the `changedetector` service. Our usage of this in tests however is only interested in waiting out the completion of the change event.
Remove unnecessary change event waits. These waits should not be necessary if handled correctly.
* tests: `addmailuser` to `add_mail_account_then_wait_until_ready mail()`
This helper method is used where appropriate.
- A password is not relevant (optional).
- We need to wait on the creation on the account (Dovecot and `/var/mail` directory).
* tests: `setup-cli` revise `add`, `update`, `del` tests
The delete test was failing as the `/var/mail` directory did not yet exist.
There is now a proper delay imposed in the `add` test now shares the same account for both `update` and `del` tests resolving that failure.
Additionally tests use better asserts where appropriate and the wait + sleep logic in `add` has been improved (now takes 10 seconds to complete, approx half the time than before).
The `del` test TODO while not technically addressed is no longer relevant due to the tests being switched to `-c` option (there is a separate `no container` test file, but it doesn't provide a `del` test).
* tests(fix): Ensure Postfix is reachable after waiting on ClamAV
There is not much reason to check before waiting on ClamAV.
It is more helpful to debug failures from `nc` mail send commands if we know that nothing went wrong inbetween the ClamAV wait time.
Additionally added an assertion which should provide more information if this part of the test setup fails again.
* tests(fix): Move health check to the top
This test is a bit fragile. It relies on defaults for the healthcheck with intervals of 30 seconds.
If the check occurs while Postfix is down due a change event from earlier tests and the healthcheck kicks in at that point, then if there is not enough time to refresh the health status from `unhealthy`, the test will fail with a false-positive as Postfix is actually working and up again..
* tests(fix): Wait on directory to be removed
Workaround that tries not to introduce heavier delays by waiting on a full change event to complete in the previous `email update` if possible.
There is a chance that the account has the folder deleted, but restored from an active change event (for password update, then the account delete).
The new version uses our `log.sh` helper to simplify logging
significantly. Moreover, the script was adjusted to the current style
and the GitHub workflow was streamlined. The workflow is ot providing
the version anymore (which was useless anyway), and has been compacted.
* outsourcing env variable setup
This commit contains major parts of the work of refactoring the setup
and usage of environment variables. It outsources the setup into its own
script and provides dedicated functions to be executed at a later point in time.
A **new** env variable was added: `USER_PROVISIONG` which provides a
better way of defining which method / protocol to use when it comes to
setting up users. This way, the `ENABLE_LDAP` variable is deprecated,
but all of this is backwards compatible due to a "compatibility layer", a function provided by the new variables script.
This is not a breaking change. It mostly refators internal scripts. The
only change facing the user-side is the deprecation of `ENABLE_LDAP`. We
can prolong the period of deprecation for this variable as long as we
want, because the new function that ensures backwards compatibility
provides a clean interface for the future.
Co-authored-by: Brennan Kinney <5098581+polarathene@users.noreply.github.com>
Co-authored-by: Casper <casperklein@users.noreply.github.com>
* tests: Update testing submodules (bats-assert, bats-support)
These two submodules were migrated to the `bats-core` organization, where they continued to receive updates.
* tests: Use tagged release of `bats-core/bats-support`
This is technically one commit backwards, but no relevant difference has been made since, other than moving the submodule to the `bats-core` organization.
* tests: Bump `bats-assert` to August 2022 (master)
No official release tag since Nov 2018, but a fair amount of changes since then.
* tests: Bump `bats-core` to `v1.7.0` release
* tests(fix): Correctly use assertions
Some tests were updated as the upgrade of bats submodules had `assert` methods raise awareness of incorrect usage.
This additionally revealed some existing tests that weren't meant to be using `run`, which swallowed failures from surfacing.
See the associated PR for more detailed commentary on specific changes.
### Commands refactored:
- User (**All:** add / list / update / del + _dovecot-master variants_)
- Quota (**All:** set / del)
- Virtual Alias (**All:** add / list /del)
- Relay (**All:** add-relayhost / add-sasl / exclude-domain)
### Overall changes involve:
- **Fairly common structure:**
- `_main` method at the top provides an overview of logical steps:
- After all methods are declared beneath it (_and imported from the new `helpers/database/db.sh`_), the `_main` is called at the bottom of the file.
- `delmailuser` additionally processes option support for `-y` prior to calling `_main`.
- `__usage` is now consistent with each of these commands, along with the `help` command.
- Most logic delegated to new helper scripts. Some duplicate content remains on the basis that it's low-risk to maintenance and avoids less hassle to jump between files to check a single line, usually this is arg validation.
- Error handling should be more consistent, along with var names (_no more `USER`/`EMAIL`/`FULL_EMAIL` to refer to the same expected value_).
- **Three new management scripts** (in `helpers/database/manage/`) using a common structure for managing changes to their respective "Database" config file.
- `postfix-accounts.sh` unified not only add and update commands, but also all the dovecot-master versions, a single password call for all 4 of them, with a 5th consumer of the password prompt from the relay command `addsaslpassword`.
- These scripts delegate actual writes to `helpers/database/db.sh` which provides a common API to support the changes made.
- This is more verbose/complex vs the current inline operations each command currently has, as it provides generic support instead of slightly different variations being maintained, along with handling some edge cases that existed and would lead to bugs (notably substring matches).
- Centralizing changes here seems wiser than scattered about. I've tried to make it easy to grok, hopefully it's not worse than the current situation.
- List operations were kept in their respective commands, `db.sh` is only really managing writes. I didn't see a nice way for removing the code duplication for list commands as the duplication was fairly minimal, especially for `listalias` and `listdovecotmasteruser` which were quite simple in their differences in the loop body.
- `listmailuser` and `delmailuser` also retain methods exclusive to respective commands, I wasn't sure if there was any benefit to move those, but they were refactored.
* chore: Create bare new test file `setup-cli.bats`
Bare minimum to setup a new test.
* chore: Transfer over relevant tests
* chore: `mail` container name to dynamic `${TEST_NAME}`
Only applied where it's relevant. Next commit will handle the config path correction.
* chore: Use `TEST_TMP_CONFIG` for referencing local config directory
Could technically use the existing function call. Some paths were using a hard-coded config location.
Both have been converted to `TEST_TMP_CONFIG` and related `grep` calls normalizing the quote mark usage, escaping doesn't seem necessary.
* tests(fix): Create container without providing extra args reference var
If a variable name (of an array) was not provided to reference, this would fail trying to reference `'`.
* chore: Remove `SYS_PTRACE` capability from docs and configs
* chore: Remove `SYS_PTRACE` capability from tests
Doesn't seem to be required. It was originally added when the original change detection feature PR apparently needed it to function.
This helper was to support an earlier ENV for SASL auth support. When extracting logic into individual helpers, it was assumed this was separate from relay support, which it appears was not the case.
---
The `SASL_PASSWD` ENV is specified in tests but no longer used. There is no `external-domain.com` relay configured or tested against anywhere in the project.
The ENV was likely used in tests prior to improved relay support that allowed for adding more than a single set of relay credentials.
---
It likewise has no real relevance anywhere else outside of `relay.sh` as it's the only portion of code to operate with it.
It's only relevant for SASL auth as an SMTP client, not the SMTP server (`smtpd`) SASL support that is delegated to Dovecot. Functionality has been completely migrated into `relay.sh` as a result.
Documentation is poor for this ENV, it is unlikely in wide use? Should consider for removal.
---
The ENV has been dependent upon `RELAY_HOST` to actually enable postfix to use `/etc/postfix/sasl_passwd`, thus not likely relevant in existing setups?
---
Migrate `/etc/postfix/sasl_passwd` check from `tests.bats` as it belongs to relay tests.
* chore: Fix typo
* chore: Apply explicit chroot default for `sender-cleanup`
The implicit default is set to `y` as a compatibility fallback, but otherwise it is [advised to set to `n` going forward](http://www.postfix.org/COMPATIBILITY_README.html#chroot).
Test was changed to catch any backwards-compatibility logs, not just those for `chroot=y`. `using` added as a prefix to avoid catching log message whenever a setting is changed that the default compatibility level is active.
* chore: Set `compatibility_level` in `main.cf`
We retain the level`2` value previously set via scripts. This avoids log noise that isn't helpful.
Applied review feedback to give maintainers some context with this setting and why we have it presently set to `2`.
* chore: Extract change-detection method to it's own helper
This doesn't really belong in `helpers/ssl.sh`. Moving to it's own helper script.
* chore: Co-locate related change-detection method from container startup
It seems relevant to migrate the related support during startup for the change detection feature into this helper.
I opted to move the call from `start-mailserver.sh` into the `_setup` call at the end for a more explicit/visible location.
* chore: Move `CHKSUM_FILE` into `helpers/change-detection.sh`
It belongs there, not in `helpers/index.sh`.
* chore: Revise inline documentation
* tests(fix): Ensure correct functionality
Presently `test/test_helper.bats` is using it's own `CHKSUM_FILE` instead of sourcing the var for the filepath.
`test_helper/common.bash` was calling a method to check for changes, but this helper may not correctly detect letsencrypt related changes as these are not ENV rely on, but global vars handled by `helpers/dns.sh`, so that should be run first like it is for `check-for-changes.sh`.
* tests(chore): Use `CHKSUM_FILE` var from helper
* chore: `addmailuser` should use `CHKSUM_FILE` var
* chore: Update `check-for-changes.sh` log message with correct path
* tests(fix): Increase some timeouts
Running tests locally via a VM these tests would fail sometimes due to the time from being queued and Amavis actually processing being roughly around 30 seconds.
There should be no harm in raising this to 60 seconds, other than delaying a failure case which will ripple through other time sensitive tests.
It's better to pass when functionality is actually correct but just needs a bit longer to complete.
* tests(fix): Don't setup an invalid hostname
During container startup `helpers/dns.sh` would panic with `hostname -f` failing.
Dropping `--domainname` for this container is fine and does not affect the point of it's test.
---
It's unclear why this does not occur in CI. Possibly changes within the docker daemon since as CI runs docker on Ubuntu 20.04? (2020).
For clarity, this may be equivalent to setting a hostname of `domain.com.domain.com`, or `--hostname` value truncated the NIS domain (`--domainname`) of the same value.
IIRC, it would still fail with both options using different values if `--hostname` was multi-label. I believe I've documented how non-deterministic these options can be across different environments.
`--hostname` should be preferred. There doesn't seem to be any reason to actually need `--domainname` (which is NIS domain name, unrelated to the DNS domain name). We still need to properly investigate reworking our ENV support that `dns.sh` manages.
---
Containers were also not removing themselves after failures either (missing teardown). Which would cause problems when running tests again.
* chore: Normalize white-space
Sets a consistent indent size of 2 spaces. Previously this varied a fair bit, sometimes with tabs or mixed tabs and spaces.
Some formatting with blank lines.
Easier to review with white-space in diff ignored. Some minor edits besides blank lines, but no change in functionality.
* fix: `setup.sh` target container under test
Some of the `setup.sh` commands did not specify the container which was problematic if another `docker-mailserver` container was running, causing test failures.
This probably doesn't help with `test/no_container.bats`, but at least prevents `test/tests.bats` failing at this point.
Dovecot master accounts can now be configured in DMS via `setup.sh`.
A master account is useful for administration purposes, or to perform mailbox backups of every user account over IMAP.
Upstream Docs: https://doc.dovecot.org/configuration_manual/authentication/master_users/
Co-authored-by: Casper <casperklein@users.noreply.github.com>
Co-authored-by: Brennan Kinney <5098581+polarathene@users.noreply.github.com>
* chore: Extract letsencrypt logic into methods
This allows other scripts to share the functionality to discover the correct letsencrypt folder from the 3 possible locations (where specific order is important).
As these methods should now return a string value, the `return 1` after a panic is now dropped.
* chore: Update comments
The todo is resolved with this PR, `_setup_ssl` will be called by both cert conditional statements with purpose for each better documented to maintainers at the start of the logic block.
* refactor: Defer most logic to helper/ssl.sh
The loop is no longer required, extraction is delegated to `_setup_ssl` now.
For the change event prevention, we retrieve the relevant FQDN via the new helper method, beyond that it's just indentation diff.
`check-for-changes.sh` adjusted to allow locally scoped var declarations by wrapping a function. Presently no loop control flow is needed so this seems fine. Made it clear that `CHANGED` is local and `CHKSUM_FILE` is not.
Panic scope doesn't require `SSL_TYPE` for context, it's clearly`letsencrypt`.
* fix: Correctly match wildcard results
Now that the service configs are properly updated, when the services restart they will return a cert with the SAN `DNS:*.example.test`, which is valid for `mail.example.test`, however the test function did not properly account for this in the regexp query.
Resolved by truncating the left-most DNS label from FQDN and adding a third check to match a returned wildcard DNS result.
Extracted out the common logic to create the regexp query and renamed the methods to communicate more clearly that they check the FQDN is supported, not necessarily explicitly listed by the cert.
* tests(letsencrypt): Enable remaining tests
These will now pass. Adjusted comments accordingly.
Added an additional test on a fake FQDN that should still be valid to a wildcard cert (SNI validation in a proper setup would reject the connection afterwards).
Co-authored-by: Georg Lauterbach <44545919+georglauterbach@users.noreply.github.com>
* first adjustments to use Fail2Ban with nftables
* replace `iptables` -> `nftables` and adjust tests
nftables lists IPs a bit differently , so the order was adjusted for the
tests to be more flexible.
* line correction in mailserver.env
* change from `.conf` -> `.local` and remove redundant config
* revert HEREDOC to `echo`
Co-authored-by: Casper <casperklein@users.noreply.github.com>
* refactored `check-for-changes.sh`
I refactored `check-for-changes.sh` and used the new log. `_notify` can
therefore be deleted as it is used no more.
I opted to source `/etc/dms-settings` as a whole to
future-proof the script. When the DNS adjustments PRs (that do not exist
by now but will exit in the future) are done, we can then remove
`_obtain_hostname_and_domainname` because we're already writing the
variables to `/etc/dms-settings`. I left instructions in the script in
the form of TODO comments.
Because we now log the date for all messages of the changedetector, we
need to `tail` a bit more log than before.
* disabled unreliable test
The "quota exceeded" test is unreliable and failed too often lately for
my taste. Therefore, I'd like to disable it because there is no use in
having such a test.
* corrected PR id in URL
* refactored scripts located under `target/bin/`
The scripts under `target/bin/` now use the new log and I replaced some
`""` with `''` on the way. The functionality stays the same, this mostly
style and log.
* corrected fail2ban (script and tests)
* corrected OpenDKIM log output in tests
* reverted (some) changes to `sedfile`
Moreover, a few messages for BATS were streamlined and a regression in
the linting script reverted.
* apple PR feedback
* improve log output from `fail2ban` script
The new output has a single, clear message with the '[ ERROR ] '
prefix, and then output that explains the error afterwards. This is
coherent with the logging style which should be used while providing
more information than just a single line about IPTables not functioning.
* simplified `setquota` script
* consistently named the `__usage` function
Before, scripts located under `target/bin/` were using `usage` or
`__usage`. Now, they're using `__usage` as they should.
* improved `sedfile`
With `sedfile`, we cannot use the helper functions in a nice way because
it is used early in the Dockerfile at a stage where the helper scripts
are not yet copied. The script has been adjusted to be canonical with
all the other scripts under `target/bin/`.
* fixed tests
* removed `__usage` from places where it does not belong
`__usage` is to be used on wrong user input, not on other failures as
well. This was fixed in `delquota` and `setquota`.
* apply PR review feedback
* added new `_log` function
With `_log`, the `_notify` method wa rendered obsolete. `_notify` was
not completely removed due to test failures in `check-for-changes.sh`.
The new `_log` function properly uses log levels such as `trace`,
`debug`, `info`, `warn` and `error`. It provides a cleaner solution
and renders `DMS_DEBUG` obsolete too (as only `_notify` depends on it).
* converted all helper script to new `_log` function
* converted all startup stacks to new `log` function
* `start-mailserver.sh` now uses new `_log` function
* final test and misc small script adjustments
* updated documentation
The new setup will now set env variables on one place and on one place
only. The old setup used two separate places wich is not DRY and
confusing.
Some default values changed:
1. PFLOGSUMM_TRIGGER: logrotate => none
2. REPORT_SENDER: mailserver-report@HOSTNAME => mailserver-report@DOMAIN
3. REPORT_RECIPIENT: "0" => POSTMASTER_ADDRESS
One env variable was renamed: REPORT_INTERVAL => LOGROTATE_INTERVAL
I believe these defaults to be more sensible, especially the REPORT_RECIPIENT
address. The PFLOGSUMM_TRIGGER value was changed to `none` because otherwise
people would start getting daily Postfix log summary reports automatically.
Now, this is opt-in, and reports are sent only when enabled properly.
Some of the variables changed were marked as deprecated. I removed the note,
as the variables now bear some (sane) defaults again for other variables
(i.e.) REPORT_RECIPIENT is now default for other recipient addresses.
Co-authored-by: Brennan Kinney <5098581+polarathene@users.noreply.github.com>
Co-authored-by: Casper <casperklein@users.noreply.github.com>
* get rid of subshell + exec
The new way of executing `sha512sum` should work as well as the old way
but without the clutter and possible problems the usage of subshells +
exec incurs.
Moreover, there was a misconception about array expansion. Using `""`
around an expanding array (`${ARRAY[@]}`) is quite fine (and actually
the preffered way), not because it makes the expansion _one_ string
(this would be `${ARRAY[*]}`), but it makes sure when elements are
expanded, each element has `""` around them so to speak, i.e. there is
no re-splitting of these elements.
* removed old concerns in comments
* increase test and check for changes sleep duration
The `DYNAMIC_FILES` var was quote wrapped, treating all filepaths to create checksums for as a single string that would be ignored instead of processed individually.
Removed the quotes, and changed the for loop to an array which accomplishes the same goal.
* fix: Prevent unnecessary change detection event
`acme.json` change would extract new cert files, which would then be hashed after restarting services and considered a change event, running through the logic again and restarting services once more when that was not required.
The checksum entries for those cert files are now replaced with new entries containing updated checksum hashes, after `acme.json` extraction.
Removes duplicate logic from `check-for-changes.sh` that is used/maintained elsewhere to avoid risk of problems, as this code is already starting to diverge / rot.
---
Previously the change detection support has had code added for rebuilding config upon change detection which is the same as code run during startup scripts. Unfortunately over time this has fallen out of sync. Mostly the startup scripts would get maintenance and the contributor and reviewers may not have been aware of the duplicate code handled by `check-for-changes.sh`.
That code was starting to diverge in addition to some changes in structure (_eg: relay host logic seems interleaved here vs separated out in startup scripts_). I wanted to address this before it risks becoming a much bigger headache.
Rather than bloat `helper-functions.sh` further, I've added a `helpers/` folder extracting relevant common logic between startup scripts and `changedetector`. If you want to follow that process I've kept scoped commits to make those diffs easier. Some minor changes/improvements were added but nothing significant.
---
- chore: Extract relay host logic to new `relay.sh` helper
- chore: Extract `/etc/postfix/sasl_passwd` logic to new `sasl.sh` helper
- chore: Extract `postfix-accounts.cf` logic to new `accounts.sh` helper
- chore: Extract `/etc/aliases` logic to new `aliases.sh` helper
- chore: Extract `/etc/postfix/vhost` logic to new `postfix.sh` helper
- chore: Add inline docs for Postfix configs
> These are possibly more verbose than needed and can be reduced at a later stage.
> They are helpful during this refactor process while investigating that everything is handled correctly.
`accounts.sh`:
- Add note regarding potential bug for bare domain setups with `/etc/postfix/vhost` and `mydestination` sharing same domain value.
`relay.sh`:
- Remove the tabs for a single space delimiter, revised associated comment.
- Add PR reference for original `_populate_relayhost_map` implementation which has some useful details.
Co-authored-by: Georg Lauterbach <44545919+georglauterbach@users.noreply.github.com>
Co-authored-by: Casper <casperklein@users.noreply.github.com>
* chore: Normalize container setup
Easier to grok what is different between configurations.
- Container name usage replaced with variable
- Volumes defined earlier and redeclared when relevant (only real difference is `VOLUME_LETSENCRYPT`)
- Contextual comment about the `acme.json` copy.
- Quoting `SSL_TYPE`, `SSL_DOMAIN` and `-h` values for syntax highlighting.
- Moved `-t` and `${NAME}` to separate line.
- Consistent indentation.
* chore: DRY test logic
Extracts out repeated test logic into methods
* chore: Scope configs to individual test cases (1/3)
- Preparation step for shifting out the container configs to their own scoped test cases. Split into multiple commits to ease reviewing by diffs for this change.
- Re-arrange the hostname and domain configs to match the expected order of the new test cases.
- Shuffle the hostname and domainname grouped tests into tests per container config scope.
- Collapse the `acme.json` test cases into single test case.
* chore: Scope configs to individual test cases (2/3)
- Shifts the hostname and domainname container configs into their respective scoped test cases.
- Moving the `acme.json` container config produces a less favorable diff, so is deferred to a follow-up commit.
- Test cases updated to refer to their `${CONTAINER_NAME}` var instead of the hard-coded string name.
* chore: Scope configs to individual test cases (3/3)
Final commit to shift out the container configs.
- Common vars are exported in `setup_file()` for the test cases to use without needing to repeat the declaration in each test case.
- `teardown_file()` shifts container removal at end of scoped test case.
* chore: Adapt to `common_container_setup` template
- `CONTAINER_NAME` becomes `TEST_NAME` (`common.bash` helper via `init_with_defaults`).
- `docker run ...` and related configuration is now outsourced to the `common.bash` helper, only extra args that the default template does not cover are defined in the test case.
- `TARGET_DOMAIN`establishes the domain folder name for `/etc/letsencrypt/live`.
- `_should*` methods no longer manage a `CONTAINER_NAME` arg, instead using the `TEST_NAME` global that should be valid as test is run as a sequence of test cases.
- `PRIVATE_CONFIG` and the `private_config_path ...` are now using the global `TEST_TMP_CONFIG` initialized at the start of each test case, slightly different as not locally defined/scoped like `PRIVATE_CONFIG` would be within the test case, hence the explicit choice of a different name for context.
* chore: Minor tweaks
- Test case comment descriptions.
- DRY: `docker rm -f` lines moved to `teardown()`
- Use `wait_for_service` helper instead of checking the `changedetector` script itself is running.
- There is a startup delay before the `changedetector` begins monitoring, wait until it ready event is logged.
- Added a helper to query logs for a service (useful later).
- `/bin/sh` commands reduced to `sh`.
- Change the config check to match and compare output, not number of lines returned. Provides better failure output by bats to debug against.
* chore: Add more test functions for `acme.json`
This just extracts out existing logic from the test case to functions to make the test case itself more readable/terse.
* chore: Housekeeping
No changes, just moving logic around and grouping into inline functions, with some added comments.
* chore: Switch to `example.test` certs
This also required copying the source files to match the expected letsencrypt file structure expected in the test/container usage.
* chore: Delete `test/config/letsencrypt/`
No longer necessary, using the `example.test/` certs instead.
These letsencrypt certs weren't for the domains they were used for, and of course long expired.
* chore: Housekeeping
Add more maintainer comments, rename some functions.
* tests: Expand `acme.json` extraction coverage
Finally able to add more test coverage! :)
- Two new methods to validate expected success/failure of extraction for a given FQDN.
- Added an RSA test prior to the wildcard to test a renewal simulation (just with different cert type).
- Added extra method to make sure we're detecting multiple successful change events, not just a previous logged success (false positive).
* tests: Refactor the negotiate_tls functionality
Covers all ports (except POP) and correctly tests against expected verification status with new `example.test` certs.
The `FQDN` var will be put to use in a follow-up commit.
* tests: Verify the certs contain the expected FQDNs
* chore: Extract TLS test methods into a separate helper script
Can be useful for other TLS tests to utilize.
* chore: Housekeeping
* chore: Fix test typo
There was a mismatch between the output and expected output between these two files "find key for" and "find key & cert for". Changed to "find key and/or cert for" to make the warning more clear that it's issued for either or both failure conditions.
Co-authored-by: Georg Lauterbach <44545919+georglauterbach@users.noreply.github.com>
Better logical flow, handling and inline documentation.
Despite the verbosity, it's better to make this visible here for maintenance and debugging purposes than trying to dig through issue/PR or commit history for it.
* fix: Panic when HOSTNAME is misconfigured
* chore: Add more comment docs for maintainers
* tests(fix): Use `--domainname` not ENV `DOMAINNAME`
Co-authored-by: Casper <casperklein@users.noreply.github.com>
Co-authored-by: Georg Lauterbach <44545919+georglauterbach@users.noreply.github.com>
These are improvements for better supporting the requirements of other tests.
- Opted for passing an array reference instead of an ENV file. This seems to be a better approach and supports more than just ENV changes.
- Likewise, shifted to a `create` + `start` approach, instead of `docker run` for added flexibility.
- Using `TEST_TMP_CONFIG` instead of `PRIVATE_CONFIG` to make the difference in usage with config volume in tests more clear.
- Changed the config volume from read-only volume mount to be read-write instead, which seems required for other tests.
- Added notes about logged failures from a read-only config volume during container startup.
- Added `TEST_CA_CERT` as a default CA cert path for the test files volume. This can be used by default by openssl methods.
Split into scoped commits with messages if further details are needed, view those via the associated PR :)
**Commit Summary:**
**`check-for-changes.sh`**
- Prevent `SSL_DOMAIN` silently skipping when value has wildcard prefix `*.` (_at least this was known as a bugfix when originally committed in linked PR_).
- Improved inlined docs for maintainers.
- Additional logging for debugging.
**`helper-functions.sh:_extract_certs_from_acme`**:
- Fail if the input arg (_`$CERT_DOMAIN`, aka the FQDN_) provided for extraction is empty.
- Use `$CERT_DOMAIN` in place of `$HOSTNAME` and `$1` for a consistent value (_previously could mismatch, eg with `SSL_DOMAIN` defined_).
- The conditional is now only for handling extraction failure (_key or cert value is missing from extraction_).
- Log an actual warning or success (debug) based on outcome.
- Don't use `SSL_DOMAIN` with wildcard value for the `mkdir` letsencrypt directory name (_wildcard prefix `*.` is first stripped instead_).
**`acme_extract`** (_new python utility for `acme.json` handling_):
- Extracted out into a python script that can be treated as a utility in the `$PATH` like other helper scripts. It can now be used and optionally tested directly instead of via `helper-functions.sh`.
-Made compatible with Python 3, as Python 2 is EOL and no longer in newer versions of Debian.
These files will replace the existing `test/config/letsencrypt` content which has some random provisioned FQDN for letsencrypt that doesn't match the FQDN tested, `acme.json` files with FQDNs that don't match those certs FQDNs and changes to certs that won't expire until 2031. `test/config/letsencrypt` will be removed with the associated test update PR.
The changes amount to:
- Re-configuring the FQDN values that some certs were created for (_needed for flexibility in testing_).
- Adding an `*.example.test` wildcard (_both RSA and ECDSA_).
- Adding `acme.json` encoded versions (_traefik extraction support will use these instead_).
- Updated / new internal docs for maintainers of this content.
For more detailed information on those changes, please see the associated commit messages via the PR.
Dovecot quota support would log auth failures when Postfix validated incoming mail to accept/reject and the `check_policy_service` for `quota-status` was queried with a recipient that was an account alias.
When Dovecot is not aware of the user account, it will not be able to check a quota and inform Postfix that everything is fine, Postfix will accept the mail and send it to Dovecot, where if the quota is exceeded will result in a bounce back to the sender. This is considered "backscatter" and can be abused by spammers forging the sender address which can get your server blacklisted.
The solution is to either disable quota support `ENABLE_QUOTAS=0`, or as a workaround, add dummy accounts to Dovecot userdb for aliases in `postfix-virtual.cf` (not `postfix-aliases.cf`), these dummy accounts will map to the real user account mailbox (real users are defined in `postfix-accounts.cf`).
The workaround is naive, in that we only check for basic 1-to-1 alias mapping to real accounts. This will still be an issue for aliases that map to another alias or multiple addresses (real or alias). Unfortunately Postfix will not expand aliases until accepting mail where this would be too late.
A better solution is to proxy the `check_policy_service` from Dovecot `quota-status` that Postfix queries in `main.cf:smtpd_recipient_restrictions`, however this requires a fair amount more of additional work and still requires an implementation to recursively query aliases for nested or multiple address mappings, which can then be forwarded to the `quota-status` service configured by Dovecot in `/etc/dovecot/conf.d/90-quota.conf`.
LDAP users are unaffected as quota support is not supported/implemented with `docker-mailserver` at this time, it is always considered disabled when using LDAP.
---
Additionally Dovecot configuration for `passdb` has been fixed to use the correct password hash scheme of `SHA512-CRYPT`.
Co-authored-by: Casper <casperklein@users.noreply.github.com>
Co-authored-by: Brennan Kinney <5098581+polarathene@users.noreply.github.com>
* fix: Spam bounced test copy/paste typo
* tests(docs): Expand inline documentation
Should assist maintainers like myself that are not yet familiar with this functionality, saving some time :)
* Refactor bounced test + Introduce initial container template
DRY'd up the test and extracted a common init pattern for other tests to adopt in future.
The test does not need to run distinct containers at once, so a common name is fine, although the `init_with_defaults()` method could be given an arg to add a suffix: `init_with_defaults "_${BATS_TEST_NUMBER}"` which could be called in `setup()` for tests that can benefit from being run in parallel.
Often it seems the containers only need the bare minimum config such as accounts provided to actually make the container happy to perform a test, so sharing a `:ro` config mount is fine, or in future this could be better addressed.
---
The test would fail if the test cases requiring smtp access ran before postfix was ready (_only a few seconds after setup scripts announce being done_). Added the wait condition for smtp, took a while to track that failure down.
* chore(refactor): DRY up the `_setup_ssl` method
- `/etc/postfix/ssl` was a bit misleading in usage here. As a maintainer (of my own contribution!) I was confused why only `/etc/postfix/ssl` was referenced and not `/etc/dovecot/ssl`.
- The postfix specific path is unnecessary, dovecot was referencing it via it's config, the same can be done from postfix to a generic DMS specific config location instead.
- This location is defined and created early as `/etc/dms/tls` (with var `DMS_TLS_PATH`). All usage of `/etc/postfix/ssl` has been replaced, making it easier to grok. Several `mkdir` commands related to this have been dropped as a result.
- Likewise, a related `TMP_DMS_TLS_PATH` var provides a reference to the config volume path `/tmp/docker-mailserver` which is used for conditions on presently hard-coded paths.
- Other values that benefit from being DRY have been lifted up into vars. Definitely easier to follow now and makes some further opportunities clearer to tackle in a future refactor.
- `chmod` has been updated where appropriate. Public key/cert is acceptable to have as readable by non-root users (644). The custom type with single fullchain file was not root accessible only, but should as it contains a private key.
- That said, the security benefit can be a bit moot due to source files that were copied remain present, the user would be responsible to ensure similar permissions on their source files.
- I've not touched LetsEncrypt section as I don't have time to investigate into that yet (not familiar with that portion).
---
* chore: Remove mkcert logic and dovecot cert
- No longer serving a purpose.
- Our own TLS startup script handles a variety of cert scenarios, while the dropped code was always generating a self-signed cert and persisting an unused cert regardless with `ONE_DIR=1`.
- To avoid similar issues that DH params had with doveadm validating filepath values in the SSL config, the default dummy values match postfix pointing to "snakeoil" cert. That serves the same purpose as mkcert was covering in the image.
- Bonus, no more hassle with differing mkcert target paths for users replacing our supplied Dovecot with the latest community edition.
---
* Error handling for SSL_TYPE
- Added a panic utility to exit early when SSL_TYPE conditions are misconfigured.
- Some info text had order of key/cert occurrence swapped to be consistent with key then cert.
- Some existing comments moved and rephrased.
- Additional comments added.
- `-f` test for cert files instead of `-e` (true also for directories/devices/symlinks).
- _notify messages lifted out of conditionals so that they always output when the case is hit.
- ~~Empty SSL_TYPE collapsed into catch all panic, while it's contents is now mapped to a new 'disabled' value.~~
---
* Use sedfile + improve sed expressions + update case style
- Uses sedfile when appropriate (file change intentional, not optional match/check).
- sed expressions modified to be DRY and reduce escaping via `-r` flag (acceptable if actual text content contains no `?`,`+`,`()` or `{}` characters, [otherwise they must be escaped](https://www.gnu.org/software/sed/manual/html_node/Extended-regexps.html)).
- sed captures anything matched between the parenthesis`()` and inserts it via `\1` as part of the replacement.
- case statements adopt the `(` prefix, adopting recent shell style for consistency.
---
* Refactor SSL_TYPE=disabled
- Postfix is also disabled now.
- Included heavy inline documentation reference for maintainers.
- Dropped an obsolete postfix config option 'use_tls' on the relayhost function, it was replaced by 'security_level'.
---
* I'm a friggin' sed wizard now
- The `modern` TLS_LEVEL is the default values for the configs they modify. As such, `sedfile` outputs an "Error" which isn't an actual concern, back to regular `sed`.
- I realized that multiple edits for the same file can all be done at once via `-e` (assuming other sed options are the same for each operation), and that `g` suffix is global scope for single line match, not whole file (default as sed iterates through individual lines).
- Some postfix replacements have `smtp` and `smtpd` lines, collapsed into a single `smtpd?` instead now that I know sed better.
---
* tests(fix): Tests that require SSL/TLS to pass
- SSL_TYPE=snakeoil added as temporary workaround.
- nmap tests are being dropped. These were added about 4-5 years ago, I have since made these redundant with the `testssl.sh` tests.
- Additionally the `--link` option is deprecated and IIRC these grades were a bit misleading when I initially used nmap in my own TLS cipher suite update PRs in the past.
- The removed SSL test is already handled in mail_ssl_manual.bats
ldap test:
- Replace `--link` alias option with `--network` and alias assignment.
- Parameterized some values and added the `SSL_TYPE` to resolve the starttls test failure.
privacy test:
- Also needed `SSL_TYPE` to pass the starttls test.
`tests.bats` had another starttls test for imap:
- Workaround for now is to give the main test container `SSL_TYPE=snakeoil`.
---
* Remove the expired lets-encrypt cert
This expired in March 2021. It was originally required when first added back in 2016 as LetsEncrypt was fairly new and not as broadly accepted into OS trust stores.
No longer the case today.
---
* chore: Housekeeping
Not required for this PR branch, little bit of tidying up while working on these two test files.
- privacy test copied over content when extracted from `tests.bats` that isn't relevant.
- ldap test was not as easy to identify the source of DOVECOT_TLS. Added comment to make the prefix connection to `configomat.sh` and `.ext` files more easier to find.
- Additionally converted the two localhost FQDN to vars.
---
* Default SSL_TYPE becomes `''` (aka equivalent to desired `disabled` case)
- This is to prevent other tests from failing by hitting the panic catchall case.
- More ideal would be adjusting tests to default to `disabled`, rather than treating `disabled` as an empty / unset SSL_TYPE value.
---
* Add inline documentation for `dms_panic`
- This could later be better formatted and placed into contributor docs.
Panic with kill (shutdown) not exit (errex):
- `kill 1` from `_shutdown` will send SIGTERM signal to PID 1 (init process).
- `exit 1` within the `start-mailserver.sh` init scripts context, will just exit the initialization script leaving the container running when it shouldn't.
The two previous `_shutdown` methods can benefit from using `dms_panic` wrapper instead to standardize on panic messages.
This feature was originally introduced by the PR: https://github.com/docker-mailserver/docker-mailserver/pull/1463
- Assign default DH params to use via Dockerfile build instead of copy and update at runtime.
- Parameterized service names and paths.
- Refactor postfix and dovecot dh methods to wrap shared dh logic
- I don't see any value in checking the alternative service for dh params file to copy over, so that's now dropped too.
- Another conditional check is dropped and the default fallback message for existing DH params file is no longer relevant.
- Improved the remaining `_notify` messages. Collapsing the warning into a single logged message also seemed relevant.
- There is no apparent need for special handling with `ONE_DIR=1`. Dropped it.
- Refactor DH params tests
- Combine custom and default DH param tests into single test file
- docs: Add instructions to use custom DH params
There is no official documented support for custom DH parameters. As no guarantee is provided, this is considered an internal change, not a breaking one.
* changed the locking function to better support multiple servers running at once and sharing the same config
* helper function testing now runs inside of container
Co-authored-by: Brennan Kinney <5098581+polarathene@users.noreply.github.com>
Decoupling setup process from `setup.sh` script by introducing a setup script _inside_ the container that coordinates the setup process.
**This is not a breaking change**. This way, we do not have to keep track of versions of `setup.sh`.
This change brings the additional benefit for Kubernetes users to be able to make use of `setup` now, without the need for `setup.sh`.
---
* move setup process into container; setup.sh versioning not needed anymore
* add tilde functionality to docs
Co-authored-by: Brennan Kinney <5098581+polarathene@users.noreply.github.com>
Co-authored-by: Casper <casperklein@users.noreply.github.com>
Centralize the collection of the HOSTNAME and DOMAINAME so that it's predictable and uniform across the various scripts (using the helper). Ensure it supports the various configurations users can have (both subdomain and without subdomain, override and no override).
---
* using _obtain_hostname_and_domainname helper + covers when not a subdomain
doc: OVERRIDE_HOSTNAME takes priority
* added tests for non-subdomain hostname + further improvements
* moved SRS DOMAINANME tests into hostname test file + Allowing DOMAINNAME ENV to override what would be automatically set
---
Co-authored-by: Georg Lauterbach <44545919+georglauterbach@users.noreply.github.com>
* docker_container first, then fall back to docker_image
+ test changes to support
+ test change to wait for smtp port to fix flakey tests since https://github.com/docker-mailserver/docker-mailserver/pull/2104
* quick fix
* Update setup.sh
Co-authored-by: Georg Lauterbach <44545919+georglauterbach@users.noreply.github.com>
Co-authored-by: Casper <casperklein@users.noreply.github.com>
* fix: make sure the SASLAUTHD_LDAP_HOST/PROTO logic makes sense and use LDAP_SERVER_HOST as a fallback (#1983)
* chore(docs): document changes to LDAP/SASLAUTHD as of #1983
* fix!: apply default value modifications suggested in #1983https://github.com/docker-mailserver/docker-mailserver/issues/1983#issuecomment-844848224
* chore(test): Test SASLAUTHD_LDAP_SERVER with protocol and ..._SSL=0, as well as with default bind credentials
Note that there are currently no regression tests for this as there's only one setup_file, so that would require big changes to the testing methodology.
* refactor!: completely remove SASLAUTHD_LDAP_SSL and SASLAUTHD_LDAP_PROTO
Co-authored-by: Georg Lauterbach <44545919+aendeavor@users.noreply.github.com>
Co-authored-by: Frederic Werner <20406381+wernerfred@users.noreply.github.com>
* Use dovecot's LDAP uris option instead of hosts (fixes#1510)
* Clean up variables & environment documentation for #1901
Co-authored-by: Frederic Werner <20406381+wernerfred@users.noreply.github.com>
Co-authored-by: Georg Lauterbach <44545919+aendeavor@users.noreply.github.com>
* splitting start-mailserver.sh
* refactoring part 2
* refactored setup-stack.sh
* stzarted adjusting target/bin/*.sh to use new usage format
* corrected lowercase-uppercase test error
* better handling of .bashrc variable export
* linting tests and fix for default assignements
* last stylistic changes and rebase
* feat: Change Postfix smtpd_tls key and cert files to chain_files
Since Postfix 3.4, `smtpd_tls_cert_file` and `smtpd_tls_key_file` have been deprecated in favor of `smtpd_tls_chain_files` which supports a list of values where a single or sequence of file paths provide a private key followed by it's certificate chain.
* feat: Dual certificate support
`smtpd_tls_chain_files` allows for multiple key+cert bundles so that you can provide different key types, such as ECDSA and RSA.
To maintain compatibility with the current CERT/KEY ENV vars only a 2nd certificate is supported.
Since Dovecot 2.2.31 a related feature is also available, but it is limited to only providing one alternative certificate via separate cert and key settings.
---
This feature enables support for multiple certificates, eg for serving modern ECDSA certs with RSA as fallback.
* chore: Refactor variable names to meet style guide
Improved some comments too.
* chore: Have function definitions respect style guide
* chore: Minor edits to comments
* chore: Expand on comments for maintenance, alert of insecure config
When `SSL_TYPE` isn't properly setup, we're still offering SSL connections but not warning in logs about the insecurity of such, or why a misconfiguration may have occurred.
This commit more clearly communicates to the user that they should look into the issue before considering deploying to production.
The `TODO` comments communicate to any future maintainer to consider treating these improper configs as disabling TLS instead.
* fix: Use `snakeoil` cert
I mistakenly thought this was placeholder text, which broke some tests. This adds the two files in the correct order (private key followed by cert/chain), to fix that issue.
* fix: Disable alt cert for Dovecot if necessary
Certain scenarios may persist state of previously configured alt cert via ENV vars that are removed from a future run. If the config is not reset to original immutable state, this will correctly disable the config from using alt cert unintentionally.
* fix: Satisfy ShellCheck lint
By switching from string var to array / list expansion, this better stores the extracted result and applies it in a manner that ShellCheck linting approves, removing the need to disable the rule.
* feat: Support dual cert test
Few tweaks to the test script allows re-purposing it for covering dual cert support as well.
* chore: Rearranged cert and key lines
A little reorganization, mostly placing private key ahead of related cert lines.
* chore: Refactor `_set_certificate`
This should make the parameters a little less confusing.
Previously was 3 parameters, but the Postfix parameter (1st) may look like two variables if you don't pay attention to the surrounding quotes; while the Dovecot parameters (2nd + 3rd) would have an opposing order. There was also a variant where the `FULLKEYCHAIN` var was passed in three times.
Now it's two params, with the 2nd param as an optional one. If the 2nd param is provided, then the two params are in the order of private key then certificate, otherwise if only a single parameter it's a single PEM file with the full cert chain and private key bundled.
This avoids implying that Postfix and Dovecot might use different files.
* chore: Document current state of `SSL_TYPE` logic better
Inlined for the benefit of anyone else maintaining this section if I'm unable to address the concerns within my own time.
* docs: ENV vars
`TLS_LEVEL=old` isn't in the codebase anymore, not likely to be relevant to retain.
No point in documenting what is considered invalid / unsupported config value in the first place for `SSL_TYPE`.
`SSL_TYPE=manual` was missing documentation for both related file path ENV vars, they've been added along with their alt fallback variants.
* chore: Update Dovecot LMTP SSL test config
Not sure how relevant this is, the file isn't complete sync with the main dovecot `10-ssl.conf` config, adding the support just in case.
* chore: Rename `FULLKEYCHAIN` to avoid confusion
There doesn't appear to be a standardized name for this type of file bundle, and `keychain` may be misleading (fullkeychain often provides macOS keychain results on search engines).
Opting for a more explicit `KEY_WITH_FULLCHAIN` name instead.
* fix: Invalid var name
`_set_certificate` refactor commit accidentally changed a var name and committed that breaking the dual cert support (thanks tests!).
* test: Refactor `mail_ssl_manual.bats`
Proper test return values instead of `wc -l` based checking.
Tests with dual cert support active, tests that feature (to better detect failure case.
Third test case was unable to verify new self-signed certificate, added new certs signed with self-signed root CA.
Adjusted openssl `CApath` parameter to use `CAfile` instead as `letsencrypt` cert was replaced thus CA cert is missing from the system trust store.
* test: Properly check for files in `mail_ssl_manual.bats`
Fixes lint error.
Also realized I was accidentally asserting a file exists in the test environment, not within the container.
Resolved that and also added an additional test case to ensure the ENV var files are valid when passed in, in the event a change misconfigures them and that the issue is identified earlier.
* chore: Apply PR review feedback
Better format some strings that had mixed quotes when they weren't necessary.
Additionally DRYed up the config path for Postfix and Dovecot within the `_setup_ssl` method.
Co-authored-by: Georg Lauterbach <infrastructure@itbsd.com>
