[ScriptedAlchemy]

Root cause

Refs #4152 and #4099 (partial storage-footprint observability).

The existing Linux /proc/self/fd SQLite temp-spill census silently discarded directory-entry, symlink-target and metadata read failures, then returned unavailable_entry_count: 0. If an FD could not be inspected, the returned byte count was only a lower bound but Doctor could label that within-budget count HealthyCompleteCoverage. This is misleading precisely when unlinked etilqs_* files are being investigated.

Fix

• Account for unreadable directory entries, non-NotFound FD-link inspection errors and non-NotFound metadata failures in the existing footprint's unavailable_entry_count; do not add a parallel census or bypass its typed finding authority.
• Continue excluding FDs that close between enumeration and observation: NotFound is not an open-space leak.
• Preserve the byte lower bound. The existing footprint_findings maps an incomplete within-budget census to Partial, while a lower bound already above budget remains Degraded.
• Do not change budget values, file lifetime, pruning, admission, or retention.

Regression coverage

A Unix fixture passes a controlled FD directory through the production census and actual Doctor finding projection. One recognized unlinked SQLite temp file contributes 24 MiB, one unreadable FD entry makes coverage partial, and a vanished descriptor contributes neither space nor false unavailability. With a 1 MiB bound, the same incomplete census still reports Degraded. The existing complete-coverage and over-budget temp-spill tests remain intact.

Validation

Inspected the production census, Doctor caller, and typed bounded_finding mapping; verified this isolated one-file change against the current master base, including the published source blob and unchanged other paths. No Rust compilation, rustfmt, or regression execution was possible locally: this runner has no checkout or Bazel/Rust toolchain. Hosted CI should run //crates/tracedecay-maintenance:unit_test and the repository formatting gate on this PR's isolated head before merge. No live Linux temp-spill reproduction or performance/size improvement is claimed.

Remaining issue scope

This improves the truthfulness of Linux temp-spill census coverage, not SQLite TEMP memory corruption, source attribution of SQL spills, macOS unlinked-file measurement, cleanup, the 153 GB profile-retention incident, or the requested 24-hour bounded-growth acceptance. Both issues remain open. The independent PR repair/merge automation owns CI and merging.

---

[changeset-bot]

⚠️ No Changeset found

Latest commit: 4929d75

Merging this PR will not cause a version bump for any packages. If these changes should not result in a new version, you're good to go. If these changes should result in a version bump, you need to add a changeset.

Click here to learn what changesets are, and how to add one.

Click here if you're a maintainer who wants to add a changeset to this PR

[devin-ai-integration]

✅ Devin Review: No Issues Found

Devin Review analyzed this PR and found no bugs or issues to report.

[ScriptedAlchemy]

Consolidated into #4742 (integrate/open-pull-requests) at the captured head shown in its description, as requested. Continue review and further fixes on #4742. This closure does not assert CI readiness; the replacement remains draft. Source branch retained.