Files
felhom.eu/documentation/audits/release-space-fixes-2026-11-10/README.md
T
admin bcc73b428d
gates / gates (push) Successful in 2m1s
Space fixes released: agent 0.173.0 vouched, controller 0.352.0 global floor, hub 0.180.0 live; evidence, rows, status, report
The close-out of the attended release run of 2026-11-10 (afternoon).
- Evidence: documentation/audits/release-space-fixes-2026-11-10/ (agent, controller, hub) and the hub row after the
  fallback-kernel test.
- Rows: dated lines on R-1368, R-1371, R-1301, R-1353. 236 open before and after; opened 0, closed 0.
- Capability map: dated notes under C13 and G5; no status changed.
- disk-full runbook and 16 §14: the self-update folder is PROVEN as a product path on three test boxes.
- STATUS: new "Updated" paragraph; the old one moved to STATUS-HISTORY.
- REPORT-release-space-fixes-2026-11-10.md: wrong claims first, one deviation (the hub deploy ran before CI for the
  manifest-bump commit was read), what stays not proven.

Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
2026-11-10 17:17:04 +01:00

4.4 KiB

Evidence — the release of the three Space fixes (2026-11-10, attended release run)

What this is. The files that prove what the release run did on the boxes and on the hub. The report is REPORT-release-space-fixes-2026-11-10.md at the repository root. The fallback-kernel test of the same run has its own folder: audits/kernel-cleanup-spike-2026-11-10/evidence/fallback-boot/.

Method. No browser. CC read the boxes over SSH, read the hub's log from its pod, and called the hub's own addresses with CC's token (the same addresses the hub's pages call). Times are UTC, except lines copied from a host's journal or the hub's log, which print CET (UTC+1). The N100's host clock is about 32 seconds behind (R-1301); its lines read early by that much.

Secrets. The files were scanned before the commit. They hold no token, no password and no key. The signed job envelopes are not here.

agent/ — the agent release (R-1368 part 2)

File What it shows
01-folder-before.txt The self-update folder on the three test boxes before: 49 / 48 / 33 entries
03-sign-upload.txt One signed agent_update per box, uploaded to the hub
04-delivery.txt Each box took the job, committed, and removed the old files (removed=49, 48, 33)
05-after-demo-hp.txt, 05-after-demo-felhom.txt, 05-after-tester1.txt The folder after: 1 entry each; no pending marker; the unit active
06-rollback-sign.txt The signed job to the older version, for the rollback proof on Tester 1
07-rollback-proof.txt One crash 9 seconds into the dwell; the wrapper put the released version back in the same second
08-after-rollback-tester1.txt After the rollback: the released version runs; 2 files in the folder (the failed one stays, by design)
12-vouch-readback.txt The operator's vouch, read from the hub's log
13-n100-clock.txt The N100's host clock against DooPlex and demo-hp; chrony failed at boot (R-1301)

controller/ — the controller release (R-1371 part 2)

File What it shows
00-fill-before.txt The fill watch on the boxes before the release: 3 filesystems checked, two of them the same one
03-crossing-on-0.350.0.txt Red on a box. Scratch guest 9202, the version before, one test file: TWO warnings
04-rc1-on-9202.txt Green on a box. 9202 on the release candidate, the same fill: ONE warning; cleared silently after the file was removed by its exact name
05-boxwalk-before.txt The three test boxes before the per-box floors
07-per-box-floors.txt The three per-box floors, each 303 … flash=floor_set
08-boxwalk-after.txt The three boxes on the release: healthy, containers 10 / 26 / 9, none unhealthy
09-fs-identity-after.txt The watch's target list on each box after: 2 filesystems, one target each
10-https-only.txt scripts/https_only_check.py, exit 0
11-global-floor.txt The global floor, in its own save; the hub's read-back
12-watch.txt 30 minutes after the global floor: every box on the release, no new alarm; the mailbox as a second channel
13-per-box-floors-cleared.txt The three per-box floors cleared; the boxes stay on the release by the global floor

hub/ — the hub release (R-1371 part 3, D68)

File What it shows
02-d68-tests.txt The TestD68_* tests, green, on the released commit
03-ci-heading.txt CI for the commit the image was built from, read by commit hash
07-deploy.txt Before and after the deploy: only the hub's Deployment was out of sync; Synced/Healthy; the image id
08-ci-bump.txt CI for the manifest-bump commit. Read 3 minutes after the deploy, not before (a deviation; see the report)
09-first-list-read.txt The pages that show a household's alarm list refuse a token (403, by design). So the open point of 16 §14.4 is not measured on the running hub
10-watch.txt 30 minutes after the deploy: Synced/Healthy, no restart, no error line, 6 reports from each box, no alarm sent

Not proven by this folder

  • That a household which saved its own alarm list before no longer gets the host-disk alarm. The proof is the tests and the deployed commit. No test alarm was made, by the brief.
  • A real fill on a household's box. A fill at the critical line (95 %) on any box.
  • A removal by hand in the agent's self-update folder; the .bak-* copies of the agent (R-1368 part 4).