Make grammar check in CI more user friendly. Previously Sphinx docs
build might have failed because outdated grammar files and this would
have prevented CI from generating grammar file patch.
We've switched to an updated dataset for shotgun jobs. The change in
underlying traffic caused the more sensitive doh-get (and partially dot)
jobs to overload the resolver, making the jobs unstable and unreliable,
due to an increased number of timeouts.
Readjust the load parameters slightly to avoid exceeding ~2 % of
timeouts in the cold cache scenario to stabilize the job results.
Each run of `meson test` overwrites the default log file testlog.txt,
this means we lose the backtraces of previous run. This commit assigns
a different log file for each run.
The doh unit test has been timing out recently and we don't have a fix
for it yet. Mark it as warning rather than a hard failure, since it's a
known issue.
Related #5448
Move tests which use different configuration to dedicated modules to
avoid possible interference with other tests.
Parametrize the test cases to have a dedicated test for each server
configuration.
Turn the check_filter() comments into log messages to help with
debugging.
The `cross-version-config-tests` CI job had to be removed due to radical
differences between meson and autoconf. With the release of 9.21.11
this job now can be reintroduced by comparing two meson built BIND
source trees.
This reverts commit b4a2674d98.
The --fs-capture-search option is no more. The ability to analyse Python
scripts in Coverity turned out to be questionable anyways.
Add a new CI job that checks whether all meson.build files in the
repository are formatted in the exact same way as "muon fmt" would
format them. This enforces formatting consistency across all
meson.build files in the repository and enables updating their contents
using dedicated tools, e.g. "meson rewrite".
System tests' after_script missed the PYTHON environmental variable
setup.
$ find -name 'tsan.*' -exec "$PYTHON" util/parse_tsan.py {} \;
find: '': No such file or directory
We skip those by default as:
a) we don't want to stress the upstream servers in every CI pipeline
b) system tests need to be runnable in a isolated environment by default
Our split between build and test phases in CI triggers odd corner case
in Meson:
- Newer Meson versions (1.7.0+) do not build test targets as part of
"all" target.
- We copy build artifacts from build phase into test container.
- meson test --no-rebuild does not build test artifacts even if they are
missing.
- To build these test binaries Meson has special target
"meson-test-prereq". This target exists only in Meson >= 0.63.
- Ubuntu 22.04 has only Meson 0.61.2 so this target does not exist.
To counter this problem, we introduced BUILD_TEST_PREREQ variable in CI
to explicitly build "meson-test-prereq" target in the "build" phase only
inside images with new-enough Meson versions. This worked, but it forced
us to keep track of Meson versions on various
distros and update the variable accordingly.
This commit inverts the logic so we build the special target by default
(in the build phase) and skip building it only if Meson version is too
old. So once we drop the old image, the variable (or rather it's usage)
will be gone and we don't need to touch it for newer images.
We have also considered installing newer Meson into the test image, but
decided to keep the old version around so we can test minimal Meson
version specified in meson.build file.
In the tumbleweed image, we utilize LibreSSL. Several BIND 9 libraries
are linked against LibreSSL's libcrypto.so.55, and when Kerberos is
enabled, we link against libk5crypto.so.3, which in turn links against
OpenSSL's libcrypto.so.3. This might theoretically lead to a symbol
conflict.
In case the changelog file doesn't have an empty line at the end of the
file, the job may fail with the following error:
WARNING: Bullet list ends without a blank line; unexpected unindent.
This typically happens in MRs targeting the -S edition, as those
changelogs usually don't have an empty newline. This change ensures the
changelog job can pass and verify the title/desc contents even in those
cases.
The meson build switched to generating the file grammars and using meson
to build the manpages/ARM. This is because meson doesn't work well when
writing files outside the build directory.
However, this has been suboptimal when someone only wants to build the
documentation (like RTD). Sphinx can now be used outside meson like it
was with autoconf.
Grammars are now updated by the developer with CI checking if one is
needed or not, like clang-format.
Move the util/generate-stress-test-configs.py script from the BIND 9
source repository to the BIND 9 QA repository. This simplifies the
maintenance of that script by eliminating the need to backport every
change applied to it to multiple branches.
Meson is a modern build system that has seen a rise in adoption and some
version of it is available in almost every platform supported.
Compared to automake, meson has the following advantages:
* Meson provides a significant boost to the build and configuration time
by better exploiting parallelism.
* Meson is subjectively considered to be better in readability.
These merits alone justify experimenting with meson as a way of
improving development time and ergonomics. However, there are some
compromises to ensure the transition goes relatively smooth:
* The system tests currently rely on various files within the source
directory. Changing this requirement is a non-trivial task that can't
be currently justified. Currently the last compiled build directory
writes into the source tree which is in turn used by pytest.
* The minimum version supported has been fixed at 0.61. Increasing this
value will require choosing a baseline of distributions that can
package with meson. On the contrary, there will likely be an attempt
to decrease this value to ensure almost universal support for building
BIND 9 with meson.
