This new script is a clean way of handling the installation of packages.
I think the huge `RUN` command in `Dockerfile` was hard to read and
maintain.
Using a script is a non-issue, as the image is rebuilt whenever the
script is touched.
Co-authored-by: Brennan Kinney <5098581+polarathene@users.noreply.github.com>
The build arguments `VCS_REF` and `VCS_VER` were renamed and given
proper values according to their names.
1. `VCS_REVISION` holds the current SHA sum of the (git) HEAD pointer
2. `VCS_VERSION` now holds the contents of the `VERSION` file, i.e. a
semver version tag (one can now inspect the image and find a proper
version tag in the `org.opencontainers.image.version` label)
The build arguments were given defaults in order to allow the
`generic_build` and `generic_test` workflows to omit them (as they are
not need there anyways). When publishing images, this is fina as the
cache will rebuild almost all of the image except the last few layers
which are `LABEL`s anyways.
Mew re-usable workflows are introduced to handle building, testing and publishing the container
image in a uniform and easy way. Now, the `scheduled_builds`, `default_on_push`
and a part of the `test_merge_requests` workflow can use the same code
for building, testing and publishing the container images. This is DRY.
Co-authored-by: Brennan Kinney <5098581+polarathene@users.noreply.github.com>
This PR prepares for other PRs that use the newly introduced helper
functions. The `_log` function itself was adjusted a bit to be shorter
and more concise.
* ci: Cache builds by splitting into two jobs
For the cache to work properly, we need to derive a cache key from the build context (files that affect the Dockerfile build) instead of the cache key changing by commit SHA.
We also need to avoid a test suite failure from preventing the caching of a build, thus splitting into separate jobs.
This first attempt used `upload-artifact` and `download-artifact` to transfer the built image, but it has quite a bit of overhead and prevented multi-platform build (without complicating the workflow further).
* ci: Transfer to dependent job via cache only
While `download-artifact` + `docker load` is a little faster than rebuilding the image from cached layers, `upload-artifact` takes about 2 minutes to upload the AMD64 (330MB) tar image export (likely due to compression during upload?).
The `actions/cache` approach however does not incur that hit and is very quick (<10 secs) to complete it's post upload work. The dependent job still gets a cache-hit, and the build job is able to properly support multi-platform builds.
Added additional notes about timing and size of including ARM builds.
* ci: Move Dockerfile ARG to end of build
When the ARG changes due to commit SHA, it invalidates all cache due to the LABEL layers at the start. Then any RUN layers implicitly invalidate, even when the ARG is not used.
Introduced basic multi-stage build, and relocated the container config / metadata to the end of the build. This avoids invalidating expensive caching layers (size and build time) needlessly.
* 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>
When waiting on an account to be added to `postfix-accounts.cf`, Dovecot account creation during the startup process had already run.
Startup continued without properly creating the mail account for Dovecot. Methods like `setup email list` (with `ENABLE_QUOTAS=1`) would fail. `changedetector` service was required to be triggered to re-create Dovecot users.
- Wrapped the logic for wait + shutdown into a function call.
- Moved `_create_accounts()` to bottom of the setup function.
* 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.
* fix: Reload `amavisd-new` when vhost config is updated
Amavis was not aware of new domains in `/etc/postfix/vhost` as it did not refresh it's sources upon change detection.
- Inline docs for `check-for-changes.sh` have been shuffled around and revised a bit.
- Change processing extracted from the main change detection loop method to their own methods:
- `_get_changed_files()` - Clarifies what is going on (and how) without having to look it up. To reduce noise in the main logic loop, extracted to a separate method.
- `_postfix_dovecot_changes()` - The bulk of change processing was moved to this method. I've added conditionals to only run relevant logic.
- `_ssl_changes()` - Just shifted, no logic changed. `REGEX_NEVER_MATCH` and `ACME_CERT_DIR` vars scope set to `local`.
