1.2-maint: fix the testsuite#9917
Merged
ThomasWaldmann merged 8 commits intoJul 20, 2026
Merged
Conversation
(cherry picked from commit 133b630)
ThomasWaldmann
changed the title
The pip cache key used only ${{ runner.os }} ("Linux") and had an
over-broad "${{ runner.os }}-" restore-key fallback. GitHub Actions lets a
branch restore caches from the default branch (master), whose cache key is
"${{ runner.os }}-${{ runner.arch }}-pip-..." and which runs on ARM64
runners (ubuntu-24.04-arm). The bare "Linux-" fallback prefix-matches
master's "Linux-ARM64-pip-*" caches, so our x86_64 jobs were restoring an
ARM64 pip cache.
Include ${{ runner.arch }} in the key and drop the bare "${{ runner.os }}-"
restore-key so caches never cross architectures (matches how master keys it).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013bFWgq2KV6oYxmfH6Zqvp3
(cherry picked from commit 93d9571)
The Node 20 action runtimes are deprecated. Bump all actions to the current
major versions (matching what the master branch uses), which run on Node 24:
- actions/checkout v4 -> v7
- actions/setup-python v5 -> v6
- actions/cache v4 -> v6
- actions/upload-artifact v4 -> v7
- codecov/codecov-action v4 -> v7
- github/codeql-action/{init,analyze} v2 -> v4 (v2 is also fully deprecated)
msys2/setup-msys2 stays at v2 (current). Inputs used (token, env_vars,
name, path, languages) are unchanged across these bumps.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013bFWgq2KV6oYxmfH6Zqvp3
(cherry picked from commit 774983f)
…ate fails The fifo_feeder thread blocks on os.open(fifo, O_WRONLY) until a reader opens the FIFO. If 'borg create --read-special' exits without opening the FIFO (e.g. it raises), the feeder stays blocked forever and t.join() hangs the whole test worker. This turned a would-be test failure into an infinite CI hang (observed on Linux py311-fuse2 and other jobs, but not py39-fuse2). Drain the FIFO in the finally block so the feeder always unblocks, turning a hang into a normal (diagnosable) test result. The OSError guard covers FreeBSD 13 raising BlockingIOError on the non-blocking read. Adapted from master commit 8ce3d22 ("Fix test hanging reading FIFO when borg create failed", Peter Gerber) plus the later FreeBSD BlockingIOError guard; that fix never landed on 1.4-maint or 1.2-maint. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013bFWgq2KV6oYxmfH6Zqvp3 (cherry picked from commit d03a620)
ThomasWaldmann
marked this pull request as draft
July 19, 2026 22:17
ThomasWaldmann
force-pushed
the
1.2-fixes-rebuild
branch
from
July 19, 2026 23:10
d937587 to
cf63768
Compare
ThomasWaldmann
changed the title
ProgressIndicatorBase installs a handler on the process-global 'borg.output.progress' logger and only removes it in __del__. In-process (fork=False) command execution leaves this logger at level INFO / propagate=False with lingering handlers, and that state leaks across tests. When a later progress test runs, the "if not self.logger.handlers" guard skips the stderr handler setup, so progress output goes to the logging system instead of the stderr captured by capfd. Under xdist this surfaces as order/distribution-dependent failures of the progress tests (test_progress_percentage_*, test_extract_progress, test_progress_on, test_check_usage, ...). Add an autouse fixture that removes handlers and resets level/propagate on the progress logger after each test. This also makes the intra-test workaround in test_check_usage unnecessary (the progress logger level is set per command run by _setup_implied_logging), so remove it. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013bFWgq2KV6oYxmfH6Zqvp3 (cherry picked from commit 196f9c8) (cherry picked from commit 11a0a9c)
Originally, we only wanted to get rid of selinux xattrs, but macOS also uses some xattr keys that disturb our test results, so we filter them also. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013bFWgq2KV6oYxmfH6Zqvp3 (cherry picked from commit ac51d3e) (cherry picked from commit 677eaaf)
The autouse test-isolation fixtures (clean_env, reset_progress_logger, default_patches) lived only in the repo-root conftest.py. That file is not installed with the package, and pytest does not apply a rootdir conftest's fixtures to tests collected from a different directory tree. On CI the tests run against the sdist-installed package via `pytest --pyargs borg.testsuite` (tox), i.e. from site-packages/borg/testsuite/ - so the fixtures never ran there, and tests leaked state into each other: - BORG_NEW_PASSPHRASE leaked from test_change_passphrase, making later repokey `init` encrypt with the wrong passphrase -> ~330 spurious PassphraseWrong failures in archiver create tests. - XDG_CONFIG_HOME was not isolated, so nonce/key tests read+wrote the real security dir and poisoned each other (nonce/IV assertions). - the borg.output.progress logger handler leaked (progress test failures). These only showed up on py>=3.10 because pytest is unpinned: py3.9 resolves an older pytest that still applied the rootdir fixtures to --pyargs tests. Move conftest.py into src/borg/testsuite/ (matching 1.4-maint/master) so it is installed alongside the tests and its fixtures always apply. Adjust the flake8 / pre-commit targets accordingly (it is now covered by `flake8 src`). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013bFWgq2KV6oYxmfH6Zqvp3
Set BORG_NEW_PASSPHRASE / BORG_PASSPHRASE via the environment_variable context manager so they are restored after the test instead of being left in os.environ. While the clean_env fixture (now effective, see previous commit) already scrubs BORG_* vars between tests, not leaking them in the first place is the correct behaviour and defends against the leak reaching tests that bypass the fixture. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013bFWgq2KV6oYxmfH6Zqvp3
ThomasWaldmann
force-pushed
the
1.2-fixes-rebuild
branch
from
July 20, 2026 13:02
677eaaf to
19927f4
Compare
ThomasWaldmann
marked this pull request as ready for review
July 20, 2026 13:18
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The 1.2-maint testsuite showed all sorts of issues (most were already fixed in 1.4-maint/master, but not in 1.2-maint).
Get it back into a working state.