docs: record the measured result for acceptance step 7 - #140
Merged
Conversation
Step 7 — "a crashed agent does not pin the machine forever" — was the last unrun item in the v0.1 checklist, open since the first release. It is now executed. The assertion was released 926 s after the last event, against a 900 s workingTimeout: a 2.9% overshoot from the 30 s refresh tick plus App Nap. That is far tighter than the 917 s and 1270 s already recorded, because those measured a session expiring while others were still live. This measured the assertion being released, which needs the working count to reach zero and is what the step asks for. Measured by scripts/verify/t8-crashed-agent-hold-release.sh rather than by hand. Issue #8 argued the step could not be automated because the app exposed no programmatic state; issue #9 shipped a read route afterwards, and polling it mints no session, so a shell can observe what an agent cannot. The run also showed that watching the working count alone is ambiguous: it falls both when a session is retired and when a live one blocks on its human, and both happened 241 s apart. The recorded note says so. Signed-off-by: Carlos Eduardo Arango Gutierrez <eduardoa@nvidia.com>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Acceptance step 7 — "a crashed agent does not pin the machine forever" — was the last unrun
item in the v0.1 checklist, open since the first release. It is now executed, and this records
the measured result rather than a predicted one.
The assertion was released 926 s after the last event, against a 900 s
workingTimeout.That is a 2.9% overshoot from the 30 s refresh tick plus App Nap — far tighter than the 917 s
and 1270 s already recorded in the file, because those measured a session EXPIRING while others
were still live. This measured the assertion actually being RELEASED, which requires the
working count to reach zero and is what the step asks for.
Measured by
scripts/verify/t8-crashed-agent-hold-release.sh, not by hand. The issue argued thestep could not be automated because the app exposed no programmatic state; #9 shipped a read
route afterwards, and polling it mints no session, so a shell can observe what an agent cannot.
The run also proved that watching the working count alone is ambiguous: it falls both when a
session is retired and when a live one blocks on its human, and both happened here 241 s apart.
The note records that, along with the two conditions a re-run needs — sandbox disabled, because
curl --unix-socketsilently falls back to TCP inside one, and a genuinely quiet machine,because any live session holds it alone.
Docs only. 1233 tests in 12 suites, 0 failures, 1 pre-existing known issue.
Refs #8.