generated from erix/meta
[BUG] Serial mount-policy CI probes fail on development and release images #46
Labels
No labels
bug
ci
docs
duplicate
enhancement
help wanted
invalid
performance
phase-6
question
refactor
security
wontfix
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set
Reference
erix/integration#46
Loading…
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Summary
The serial native-image-policy gate fails in two retained runs, after all 486 catalog scenarios and earlier image/input gates pass. The release probe fails in CI 1591; the development probe fails in CI 1594. The failed input action and cause are unknown.
Reproduction
Observed in retained original CI; no new VM rerun or failure-frequency estimate.
These are the respective commands from separate original runs, not a new sequential reproduction. The wrapper selects the image identity, native-image-policy input profile and fixed RTC base 2030-12-31T12:00:00. It checks mount policy and packaged versus live timestamps.
Expected and actual behavior
stage=serial-input runner_exit=1 vm_exit=0 expected_vm_exit=0 physical_input_exit=unknown serial_input_exit=1 physical_cleanup_requested=0appears in both runs. Expected behavior is completion of the full policy probe. The composed VM-command success does not override the failing serial injector.Environment and identity
9f004461b284493b7b53d386c8727435b02abe0f.fe8c8103719af9c127dd769f4427d2a346e1ffd9.The workflow selects its original pinned component graph and x86_64 QEMU/OVMF images. Complete CI logs and original wrappers were inspected. Image/firmware artifact bytes and inner runner receipts were not downloaded; their exact digests and the guest cause remain unverified. Existing 120-second total, 15-second serial-progress and 10-second input-command limits remain unchanged.
Evidence and prior checks
Additional original CI observation — 15 September 2026
Push CI 1625 at original Integration
f8fba7520df5b6e76662741cf72b3e8b6a1edf21independently passes all 486 catalog scenarios and the preceding native input, directory, time, editor and utility gates. The development native-image-policy probe passes; the release serial policy probe fails at 15:44:50 UTC with the same composed status: serial-input stage, runner 1, VM 0 as expected, serial injector 1. Rust and Markdown pass. Three complete logs total 13,304,525 bytes, with no compiler/lint warning candidates. The workflow collects an archive locally, but the run artifact API returns an empty list; the inner serial receipt and exact image/firmware bytes remain unavailable. This is another observation of the existing policy-probe failure class, not a demonstrated shared cause or a new VM rerun. The paired review run is still running.Run 1591 fails at 05:49:16 UTC on September 13 after the development policy probe passes. Run 1594 fails at 06:21:00 UTC in the development policy probe; its later release policy probe is not reached. A shared cause is not established.
Rust and Markdown jobs pass. Complete logs are retained with no compiler/lint warning candidates. No unchanged retry, cancellation, dispatch or threshold adjustment was performed for this report.
Tracking
Keep the two image profiles distinguishable. Diagnose the failed serial milestone and retain positive and failure/cleanup coverage without relaxing mount-policy assertions or runner deadlines.
erikinkinen referenced this issue2026-09-14 07:36:11 +02:00
erikinkinen referenced this issue from erix/docs2026-09-14 07:57:37 +02:00
erikinkinen referenced this issue2026-09-15 18:57:35 +02:00