generated from erix/meta
[BUG] Release physical native-editor CI probe fails after the VM catalog passes #13
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#13
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
Integration CI run 1553 failed the physical native-editor probe on the release image after its complete 486-scenario VM catalog passed. The earlier development physical and COM1 editor probes passed. The failure prevents completion of the remaining post-image acceptance checks.
Reproduction
Observed once in the prepared CI workspace at Integration
fb851108e7373add0da4bc9af3d0aa0bd5b4059f. The workflow builds the runtime release appliance and invokes:These commands require the CI-prepared pinned component graph and build environment. No new standalone rerun was performed during this audit. A separately triggered later main run 1554 passed the same release-physical probe at 07:15:40 UTC. Its Integration commit has the same source tree, but image digests and inner probe evidence are unavailable through the published job log; that later pass is not a demonstrated fix.
Expected and actual behavior
Expected: all ten editor commands, history navigation, insertion/deletion, wrapped-line edits and the final caret satisfy the existing physical-input and framebuffer assertions.
Actual: the release-physical invocation began at 06:50:05 UTC on 2026-09-12 and failed about 44.2 seconds later with exit status 1:
The wrapper takes this path when its inner VM command fails, before its final event-list assertions. The published log does not expose that command's specific failure. Do not attribute the failure to injection, renderer/caret verification or a timeout without the retained inner evidence.
Environment and identity
fb851108e7373add0da4bc9af3d0aa0bd5b4059f; resolved kernel4330175e27718490f06b93098e95e3c1553cb765, rootdac3c1847af9915164d4a819f5f15e1827884e8b1, exsh70521817b0c8d907e765b0d8ef8ed1eaa89f67ce, ttyd62a1ffd8b966f9ffd59c017d9dc83521f9491426.1:10.0.11+ds-0+deb13u1and OVMF2025.02-8+deb13u1. Exact resolved accelerator, firmware hashes and Rust toolchain are not available in the failing probe's published log.Evidence and prior checks
Run 1553, API run 8889, integration job 19124 attempt 1 is terminal failure. Its complete retained job log shows 486/486 catalog scenarios passed, followed by 22 successful post-image workflow steps and the failure at step 23. Eight later workflow steps did not execute, including four independent terminal-mode probes within the final step.
The raw
vm.log, physical-input events, serial stream, screenshots and image identity were collected into the runner's CI directory. The Actions artifacts API lists no uploaded artifact for this run; the workflow constructs a tarball without an upload step. Recover or expose these bounded raw files before root-cause diagnosis. Preserve the original failure and all deadlines; no marker relaxation or unchanged retry-to-green is acceptance.All existing Integration issues were checked; no prior bug issue duplicates this observed failure. The zero-hash fetch and invalid helper-option errors earlier in the log are intentional passing negative fixtures and are unrelated.
Tracking
Related acceptance work: integration#6, integration#5, integration#11.
Regression closure requires exact source/image-bound release physical and COM1 editor observations, preserved raw failure evidence, and completion of the remaining post-image checks. A catalog-only pass does not close this defect.
erikinkinen referenced this issue2026-09-12 09:46:11 +02:00
erikinkinen referenced this issue2026-09-15 05:31:23 +02:00