[BUG] Physical directory host fixture expires within its shortened command budget #63

New issue

Closed

opened 2026-09-19 11:03:55 +02:00 by erikinkinen · 0 comments

erikinkinen commented

2026-09-19 11:03:55 +02:00

Owner

Summary

The real-runner host fixture intermittently fails during a valid physical-directory sequence under a three-second command budget, shorter than the maintained native probe budget.

Reproduction

Run python3 -B -Werror tests/native_directory_fixture_test.py as part of the complete helper suite. One retained full-suite run fails test_real_runner_physical_directory_and_cwd_sequence; an unchanged isolated diagnostic subsequently passes all five tests. The fixture and runner bytes match Integration 8b1c037aed2503e1f2a4b17c8a8be0666d62a305. Failure frequency and the host delay cause are not established.

Expected and actual behavior

Expected: the fixture verifies QMP/COM1 dispatch, exact consent, output and cwd transitions within the normal probe budget. Actual: the runner exits 1 with timed out while settling physical input during command 3 (ls -la). The first choice is observed, but the command deadline expires before the sequence completes.

Environment and identity

Linux x86-64 Python host fixture with private fake-QEMU QMP/COM1 sockets, tiny authored PF2 font, three-second command budget, 60-second wrapper bound, 15-second progress-stall bound and 65-second outer subprocess bound. This is not a real guest failure. No introducing revision is established.

Evidence and prior checks

The original 169-command helper result retains 168 passes and this failure. The unchanged isolated diagnostic passes five tests in 26.435 seconds. A source correction selects the probe’s existing ten-second command budget and adds controlled-clock tests for command/overall expiry and insufficient consent observation time. Actual VM limits and consent requirements stay unchanged; no scheduling or performance cause is claimed.

Published host-fixture correction — 19 September 2026: Signed Integration c479813423fb71c606fa30e5cf7d4dfb16f557d0 selects the existing ten-second native probe command budget and adds controlled-clock tests for insufficient consent observation time and exact command/overall deadline expiry. The four affected suites pass 9/5/5/7 tests without warnings; all product and actual VM deadlines are unchanged. Original CI 1671/1672 is running, so this report remains open. The earlier failed run and unchanged isolated pass are retained; the host delay cause remains unestablished.

Original CI observation — 19 September 2026: CI 1671 finishes with 471 passes/16 failures and CI 1672 with 465 passes/22 failures across all 487 scenarios at c479813423fb71c606fa30e5cf7d4dfb16f557d0. Every failed scenario is an ext3/ext4 positive filesystem case reporting error: qemu timed out after 120s, with scenario status 1. The existing filesystem deadline report retains the expanded matrix; quota issue 18 applies to the quota failure in 1672. All six complete logs are hashed (41,592,478 bytes), with zero warning candidates. Rust, Markdown and the full helper step pass. The zero-realm startup and exact-selector race scenarios explicitly pass in both runs. No common cause, source regression boundary or performance diagnosis is inferred; all actual deadlines and failure statuses remain unchanged, and neither workflow was rerun. The corrected host-fixture consumers reach the complete catalog without a helper failure. This supplies the pending original CI observation for the narrow fixture correction; the separate filesystem VM failures remain open.

Resolved — 19 September 2026: the signed correction in c479813423fb71c606fa30e5cf7d4dfb16f557d0 has its four focused 9/5/5/7-test suites and both original CI helper steps passing. A subsequent full local run of all 169 helper commands also passes without warnings at the unchanged helper source. The test now selects the existing native probe command budget, while controlled expiry and consent-observation checks retain their failure requirements. The original failed attempt and unchanged isolated pass remain evidence. This closes the host-fixture defect only; filesystem VM failures in issue 20 and the distinct physical-directory VM report remain open.

Tracking

Integration PR 12. Validate the correction and all fixture consumers, retain the initial failure, and observe original CI. This host failure does not establish a common cause with physical-directory VM issue 35.

## Summary The real-runner host fixture intermittently fails during a valid physical-directory sequence under a three-second command budget, shorter than the maintained native probe budget. ## Reproduction Run `python3 -B -Werror tests/native_directory_fixture_test.py` as part of the complete helper suite. One retained full-suite run fails `test_real_runner_physical_directory_and_cwd_sequence`; an unchanged isolated diagnostic subsequently passes all five tests. The fixture and runner bytes match Integration `8b1c037aed2503e1f2a4b17c8a8be0666d62a305`. Failure frequency and the host delay cause are not established. ## Expected and actual behavior Expected: the fixture verifies QMP/COM1 dispatch, exact consent, output and cwd transitions within the normal probe budget. Actual: the runner exits 1 with `timed out while settling physical input` during command 3 (`ls -la`). The first choice is observed, but the command deadline expires before the sequence completes. ## Environment and identity Linux x86-64 Python host fixture with private fake-QEMU QMP/COM1 sockets, tiny authored PF2 font, three-second command budget, 60-second wrapper bound, 15-second progress-stall bound and 65-second outer subprocess bound. This is not a real guest failure. No introducing revision is established. ## Evidence and prior checks The original 169-command helper result retains 168 passes and this failure. The unchanged isolated diagnostic passes five tests in 26.435 seconds. A source correction selects the probe’s existing ten-second command budget and adds controlled-clock tests for command/overall expiry and insufficient consent observation time. Actual VM limits and consent requirements stay unchanged; no scheduling or performance cause is claimed. Published host-fixture correction — 19 September 2026: Signed Integration `c479813423fb71c606fa30e5cf7d4dfb16f557d0` selects the existing ten-second native probe command budget and adds controlled-clock tests for insufficient consent observation time and exact command/overall deadline expiry. The four affected suites pass 9/5/5/7 tests without warnings; all product and actual VM deadlines are unchanged. Original CI 1671/1672 is running, so this report remains open. The earlier failed run and unchanged isolated pass are retained; the host delay cause remains unestablished. Original CI observation — 19 September 2026: [CI 1671](https://git.erikinkinen.fi/erix/integration/actions/runs/1671) finishes with 471 passes/16 failures and [CI 1672](https://git.erikinkinen.fi/erix/integration/actions/runs/1672) with 465 passes/22 failures across all 487 scenarios at `c479813423fb71c606fa30e5cf7d4dfb16f557d0`. Every failed scenario is an ext3/ext4 positive filesystem case reporting `error: qemu timed out after 120s`, with scenario status 1. The existing [filesystem deadline report](https://git.erikinkinen.fi/erix/integration/issues/20) retains the expanded matrix; [quota issue 18](https://git.erikinkinen.fi/erix/integration/issues/18) applies to the quota failure in 1672. All six complete logs are hashed (41,592,478 bytes), with zero warning candidates. Rust, Markdown and the full helper step pass. The zero-realm startup and exact-selector race scenarios explicitly pass in both runs. No common cause, source regression boundary or performance diagnosis is inferred; all actual deadlines and failure statuses remain unchanged, and neither workflow was rerun. The corrected host-fixture consumers reach the complete catalog without a helper failure. This supplies the pending original CI observation for the narrow fixture correction; the separate filesystem VM failures remain open. Resolved — 19 September 2026: the signed correction in `c479813423fb71c606fa30e5cf7d4dfb16f557d0` has its four focused 9/5/5/7-test suites and both original CI helper steps passing. A subsequent full local run of all 169 helper commands also passes without warnings at the unchanged helper source. The test now selects the existing native probe command budget, while controlled expiry and consent-observation checks retain their failure requirements. The original failed attempt and unchanged isolated pass remain evidence. This closes the host-fixture defect only; filesystem VM failures in [issue 20](https://git.erikinkinen.fi/erix/integration/issues/20) and the distinct physical-directory VM report remain open. ## Tracking [Integration PR 12](https://git.erikinkinen.fi/erix/integration/pulls/12). Validate the correction and all fixture consumers, retain the initial failure, and observe original CI. This host failure does not establish a common cause with [physical-directory VM issue 35](https://git.erikinkinen.fi/erix/integration/issues/35).

erikinkinen added the

bug

phase-6

labels

2026-09-19 11:03:55 +02:00

erikinkinen referenced this issue from a commit

2026-09-19 11:05:16 +02:00

test: Cover startup with realm admission disabled

erikinkinen referenced this issue

2026-09-19 11:07:50 +02:00

[FEATURE] Freeze the complete source, build, ABI and self-hosting inventory #1

erikinkinen referenced this issue