Repository navigation
Conversation
… as a piece The sabotage arm "the end of a worker line too long to relay is sent as a line" survived 9 of 60 runs of a_worker_line_too_long_to_relay_is_dropped_whole. The runs were in the fedora:44 CI image on a loaded host. With the test pinned to one CPU, the arm survived 12 of 30. The relay's buffer stops doubling at 2 MiB, the length of the test's longest line. A read that fills it with that line stops just before the line's NUL. With `s->skip = 1` gone, the rest of that line reaches the controller as an empty line. The check counted only lines that start with "x". In every run the arm survived, the five shorter lines had their NUL in the read that passed LINE_LIMIT, and were dropped whole. The 2 MiB line's end arrived empty. The check now counts an empty line as a piece. A 2 MiB line never fits the buffer with its NUL, so its start is always dropped. With the arm, its end always reaches the controller, empty or not. The check caught the arm in 100 of 100 runs, and the test passed 100 of 100 without it.
…ritten its pid (FreeMixer#113) The two "left in its own session" tests of test-probe failed 10 of 60 runs (30 of each). The runs were in the fedora:44 CI image on a loaded host. They failed with "the stress plugin did not leave its process" or with the ValueError of FreeMixer#113. The fault forked and returned at once. In 100 runs by the tests' own run() and run_lv2(), each run took 10 to 90 ms. In 36, the pidfile was missing (18) or empty (18). The runner can kill the process left behind before that process writes its pid. In all 100 runs, the probe's document says that the process left behind was killed. No process with that pidfile in its environment was alive then or a second later. The runner does its job. The fault returned too soon. Writing the pid by rename, as FreeMixer#113 proposes, would leave the file missing instead. The test cannot tell a missing file from a fault that never forked. In both fixtures, the fault now waits on a pipe. The process left behind closes the pipe once it is in its own session and its pid is written. With a 200 ms sleep added after its setsid(), both tests fail 10 of 10 runs each before this change. After it, they pass 10 of 10 each. Without the sleep, they passed 120 of 120 runs (60 of each).
Author
|
The CI failure was |
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
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.
Closes #113.
Two outcomes of the
testjob depended on timing in its fedora:44 container on a loaded host.a_worker_line_too_long_to_relay_is_dropped_whole, run alone 60 times, missed the sabotage arm "the end of a worker line too long to relay is sent as a line" 9 times. The relay's buffer stops doubling at 2 MiB, the length of the test's longest line. When one read fills it with that line, the read stops just before the line's NUL. With the arm, the line's end arrives as an empty line, which the check did not count. The check now counts it. It catches the arm in 100 of 100 runs and passes 100 of 100 without it.setsid(), both tests fail every run before the change and pass every run after.The steps of ci.yml's
private-refsandtestjobs and of fuzz.yml pass 3 of 3 runs there.