felhom.eu/REPORT.md

felhom.eu — task reports

Overwrite this file with a summary of the most recent task only (uniform with the other repos; not cumulative). The cumulative hub history lives in hub/CHANGELOG.md; the scripts history lives in scripts/CHANGELOG.md.

Hub v0.46.0 — observability pass: per-box log pulls, bundle custody, TTL + secret gate — 2026-07-11

felhom.eu's third of the cross-repo observability task (agent v0.83.0 + controller v0.116.1 shipped in their repos). Commits: hub e35b1ae, manifest bump + conventions runbook 7c038df. Deployed: felhom-hub:0.46.0, ArgoCD Synced/Healthy, rollout complete (17:03 CEST).

What shipped

• Store (hub/internal/store/logbundle.go + schema): log_bundle_requests (pending intent per scope+component — controller scope = customer_id / report ACK; agent scope = host_id / heartbeat envelope) + log_bundles (gzip, newest-3, 72 h TTL purged on the existing 60 s sweep). SaveLogBundle runs the token-pattern secret gate BEFORE storing: a hit stores a blocked: possible secret flag row and NO payload (fail-closed, WARN hub-side); [REDACTED] shapes + public checksums pass by design.
• Channels (additive both directions): report ACK controller_log_requested + ingest of controller_log_tail; heartbeat envelope log_tail_requested + ingest of log_tail. Consume-once on arrival. A pre-0.83 agent never fulfills → the request stays visibly pending (S6-tested, harmless).
• UI (host detail, English per the hub operator surface): Diagnostics section — "Request controller logs" / "Request agent logs" (CSRF forms), state rows (pending with the honest latency hint: controller ≤ ~15 min report interval, agent ≈ heartbeat cadence; available with View/Download; blocked), 72 h custody note. The hosts "read-only" invariant is amended by test: these two forms are the ONLY actions.
• Conventions codified: documentation/runbooks/logging-conventions.md (levels, logs-are- English, keys-never-values, capture layers, pull posture) + a one-liner in each repo CLAUDE.md.

Red-proofs (demonstrated, restored)

Scenario	Reverted shape	Failure seen
S4 secret gate	gate disabled	secret-shaped token not blocked
S2/consume-once	clear-on-arrival removed	request survived fulfillment (store + API tests)

S5 TTL: injectable clock — kept at +71 h, purged at +73 h. Keep-newest-3 + byte-cap belts tested.

Live evidence / honest latencies

• Hub 0.46.0 live (ArgoCD Synced/Healthy, image verified, clean startup log).
• Box halves live-proven from 9201: controller ring + agent ring both readable at logging.level=info through the new Debug tabs (see felhom-controller/REPORT.md timestamps).
• Operator follow-up (HUMAN — hub UI is password-gated, CC cannot log in): on hub.felhom.eu/hosts/<demo host> click both request buttons; expect the agent bundle within one heartbeat (~15 min) and the controller bundle on the next report (≤ ~15 min); download both and confirm the controller bundle carries the NAS-add phase lines + the operator log pull served INFO line. The full round-trip is already proven by internal/api/logbundle_ack_test.go.
• Blocked-secret branch: test-level only by design (no real secret planted live).

Open / observations

• Hub bearer key rotation still DUE (pre-existing; manifests/hub.yaml + screenshot-exposed).
• log_bundles UI states requested → pending are merged (a request IS pending from the box's view); expired renders as absence after the purge — states documented in the template hint.