Reliability / Recovery Readiness

Owner → Hermes → Specialist Agents
DCP READ-ONLY: CONNECTED DEVELOPMENT / UI PREVIEW DCP BATCH 5/6 SOAK IN PROGRESS TELEMETRY: UNKNOWN COMMAND EXECUTION LOCKED

Reliability / Recovery Readiness

Evidence source: REAL MOCK STATIC UNKNOWN· REAL = live read-only evidence · MOCK = demo, never verified · STATIC = documented/committed (not a live probe) · UNKNOWN = no evidence. Only PASS counts as verified.
Read-only, evidence-based readiness. A dimension is verified only when a PASS is backed by attributable evidence (real probe or documented runbook). Mock/static values are labelled and never counted as live verification.
No backup or restore PASS is claimed. The backup stack is installed but the timer is disabled, the first real backup has not run, and no restore drill has been executed. Those dimensions intentionally show UNKNOWN.
Services tracked
9
Dimensions verified
7/54
Services fully ready
0

Dimensions: Health · Backup · Restore Test · Monitoring · Rollback · Secrets.

DCP

3/6 verified50% verified
Batch 5/6 production complete; soak pending final PASS.
DimensionStatusSourceEvidenceLast verificationRecommended next action
HealthPASSREALLive read-only DCP /health probe.live probe—
BackupUNKNOWNUNKNOWNBackup stack installed but the timer is disabled/inactive and the first real backup has not run (STATIC fact). Not verified.neverRun the first real backup after the soak/backup gates pass; until then backup evidence stays UNKNOWN.
Restore TestUNKNOWNUNKNOWNNo restore drill has been executed against this service yet. Restore stays UNKNOWN until a real drill artifact exists.neverRun a disposable restore drill and record the PASS/FAIL artifact; never claim restore PASS from a design alone.
MonitoringWARNINGSTATICCompose healthcheck + /health documented, but no verified alert route to the owner is proven.neverRoute a verified alert to the owner (Uptime Kuma / n8n -> Telegram) and record the first real alert.
RollbackPASSSTATICRollback runbook documented: revert DCP_IMAGE_TAG to the prior SHA and `docker compose up -d` (docs/PRODUCTION-DEPLOYMENT.md).documentedConfirm the rollback procedure against the running image/version and record a drill result.
SecretsPASSSTATICSecrets live in /etc/dev-control-plane/production.env (root:root 600, outside the repo); only names appear in the repo.documentedConfirm secret storage/isolation with a read-only review; never render secret contents.
Missing evidence: Backup (UNKNOWN), Restore Test (UNKNOWN), Monitoring (WARNING)

Only PASS counts as verified. UNKNOWN, WARNING, FAIL and NOT_APPLICABLE never count as a pass. STATIC = documented/committed evidence, not a live probe.

PostgreSQL

3/6 verified50% verified
Private on the homelab-backend network; no published ports.
DimensionStatusSourceEvidenceLast verificationRecommended next action
HealthPASSREALLive read-only DCP /health probe.live probe—
BackupUNKNOWNUNKNOWNBackup stack installed but the timer is disabled/inactive and the first real backup has not run (STATIC fact). Not verified.neverRun the first real backup after the soak/backup gates pass; until then backup evidence stays UNKNOWN.
Restore TestUNKNOWNUNKNOWNNo restore drill has been executed against this service yet. Restore stays UNKNOWN until a real drill artifact exists.neverRun a disposable restore drill and record the PASS/FAIL artifact; never claim restore PASS from a design alone.
MonitoringWARNINGSTATICNo dedicated PostgreSQL monitoring/metrics is verified; only the DCP /health database check is documented.neverRoute a verified alert to the owner (Uptime Kuma / n8n -> Telegram) and record the first real alert.
RollbackPASSSTATICRollback documented: restore from a verified backup and/or revert the DCP image; schema changes are additive-only.documentedConfirm the rollback procedure against the running image/version and record a drill result.
SecretsPASSSTATICKept private (no published ports); DSN supplied from the external secret file, never committed.documentedConfirm secret storage/isolation with a read-only review; never render secret contents.
Missing evidence: Backup (UNKNOWN), Restore Test (UNKNOWN), Monitoring (WARNING)

Only PASS counts as verified. UNKNOWN, WARNING, FAIL and NOT_APPLICABLE never count as a pass. STATIC = documented/committed evidence, not a live probe.

Redis

0/6 verified0% verified
DimensionStatusSourceEvidenceLast verificationRecommended next action
HealthUNKNOWNUNKNOWNNo evidence collected. Not assumed healthy.neverPoint a safe read-only probe at the service (or expose it through DCP /health) before claiming health.
BackupUNKNOWNUNKNOWNBackup stack installed but the timer is disabled/inactive and the first real backup has not run (STATIC fact). Not verified.neverRun the first real backup after the soak/backup gates pass; until then backup evidence stays UNKNOWN.
Restore TestUNKNOWNUNKNOWNNo restore drill has been executed against this service yet. Restore stays UNKNOWN until a real drill artifact exists.neverRun a disposable restore drill and record the PASS/FAIL artifact; never claim restore PASS from a design alone.
MonitoringUNKNOWNUNKNOWNNo evidence collected. Not assumed healthy.neverRoute a verified alert to the owner (Uptime Kuma / n8n -> Telegram) and record the first real alert.
RollbackUNKNOWNUNKNOWNNo evidence collected. Not assumed healthy.neverConfirm the rollback procedure against the running image/version and record a drill result.
SecretsUNKNOWNUNKNOWNNo evidence collected. Not assumed healthy.neverConfirm secret storage/isolation with a read-only review; never render secret contents.
Missing evidence: Health (UNKNOWN), Backup (UNKNOWN), Restore Test (UNKNOWN), Monitoring (UNKNOWN), Rollback (UNKNOWN), Secrets (UNKNOWN)

Only PASS counts as verified. UNKNOWN, WARNING, FAIL and NOT_APPLICABLE never count as a pass. STATIC = documented/committed evidence, not a live probe.

n8n

1/6 verified17% verified
Notification consumer only; no host privilege, no Docker socket.
DimensionStatusSourceEvidenceLast verificationRecommended next action
HealthUNKNOWNUNKNOWNNo evidence collected. Not assumed healthy.neverPoint a safe read-only probe at the service (or expose it through DCP /health) before claiming health.
BackupUNKNOWNUNKNOWNBackup stack installed but the timer is disabled/inactive and the first real backup has not run (STATIC fact). Not verified.neverRun the first real backup after the soak/backup gates pass; until then backup evidence stays UNKNOWN.
Restore TestUNKNOWNUNKNOWNNo restore drill has been executed against this service yet. Restore stays UNKNOWN until a real drill artifact exists.neverRun a disposable restore drill and record the PASS/FAIL artifact; never claim restore PASS from a design alone.
MonitoringWARNINGSTATICn8n is the documented alert path (signed webhook, HMAC from env), but no end-to-end alert has been verified.neverRoute a verified alert to the owner (Uptime Kuma / n8n -> Telegram) and record the first real alert.
RollbackUNKNOWNUNKNOWNNo evidence collected. Not assumed healthy.neverConfirm the rollback procedure against the running image/version and record a drill result.
SecretsPASSSTATICWebhook signing secret only from DCP_AUTOMATION_HMAC_SECRET (env); config/commit never holds it (fail-closed if absent).documentedConfirm secret storage/isolation with a read-only review; never render secret contents.
Missing evidence: Health (UNKNOWN), Backup (UNKNOWN), Restore Test (UNKNOWN), Monitoring (WARNING), Rollback (UNKNOWN)

Only PASS counts as verified. UNKNOWN, WARNING, FAIL and NOT_APPLICABLE never count as a pass. STATIC = documented/committed evidence, not a live probe.

Authelia

0/6 verified0% verified
Interactive auth belongs at the proxy layer before remote exposure.
DimensionStatusSourceEvidenceLast verificationRecommended next action
HealthUNKNOWNUNKNOWNNo evidence collected. Not assumed healthy.neverPoint a safe read-only probe at the service (or expose it through DCP /health) before claiming health.
BackupUNKNOWNUNKNOWNBackup stack installed but the timer is disabled/inactive and the first real backup has not run (STATIC fact). Not verified.neverRun the first real backup after the soak/backup gates pass; until then backup evidence stays UNKNOWN.
Restore TestUNKNOWNUNKNOWNNo restore drill has been executed against this service yet. Restore stays UNKNOWN until a real drill artifact exists.neverRun a disposable restore drill and record the PASS/FAIL artifact; never claim restore PASS from a design alone.
MonitoringUNKNOWNUNKNOWNNo evidence collected. Not assumed healthy.neverRoute a verified alert to the owner (Uptime Kuma / n8n -> Telegram) and record the first real alert.
RollbackUNKNOWNUNKNOWNNo evidence collected. Not assumed healthy.neverConfirm the rollback procedure against the running image/version and record a drill result.
SecretsUNKNOWNUNKNOWNAuth/secrets evidence not verified; secret contents never rendered.neverConfirm secret storage/isolation with a read-only review; never render secret contents.
Missing evidence: Health (UNKNOWN), Backup (UNKNOWN), Restore Test (UNKNOWN), Monitoring (UNKNOWN), Rollback (UNKNOWN), Secrets (UNKNOWN)

Only PASS counts as verified. UNKNOWN, WARNING, FAIL and NOT_APPLICABLE never count as a pass. STATIC = documented/committed evidence, not a live probe.

Uptime Kuma

0/6 verified0% verified
Monitoring tool itself; coverage and alerting are not yet verified.
DimensionStatusSourceEvidenceLast verificationRecommended next action
HealthUNKNOWNUNKNOWNNo evidence collected. Not assumed healthy.neverPoint a safe read-only probe at the service (or expose it through DCP /health) before claiming health.
BackupUNKNOWNUNKNOWNBackup stack installed but the timer is disabled/inactive and the first real backup has not run (STATIC fact). Not verified.neverRun the first real backup after the soak/backup gates pass; until then backup evidence stays UNKNOWN.
Restore TestUNKNOWNUNKNOWNNo restore drill has been executed against this service yet. Restore stays UNKNOWN until a real drill artifact exists.neverRun a disposable restore drill and record the PASS/FAIL artifact; never claim restore PASS from a design alone.
MonitoringWARNINGSTATICUptime Kuma is present as a monitoring tool, but the monitor set and alert delivery to the owner are unverified.neverRoute a verified alert to the owner (Uptime Kuma / n8n -> Telegram) and record the first real alert.
RollbackUNKNOWNUNKNOWNNo evidence collected. Not assumed healthy.neverConfirm the rollback procedure against the running image/version and record a drill result.
SecretsUNKNOWNUNKNOWNNo evidence collected. Not assumed healthy.neverConfirm secret storage/isolation with a read-only review; never render secret contents.
Missing evidence: Health (UNKNOWN), Backup (UNKNOWN), Restore Test (UNKNOWN), Monitoring (WARNING), Rollback (UNKNOWN), Secrets (UNKNOWN)

Only PASS counts as verified. UNKNOWN, WARNING, FAIL and NOT_APPLICABLE never count as a pass. STATIC = documented/committed evidence, not a live probe.

Netdata

0/6 verified0% verified
DimensionStatusSourceEvidenceLast verificationRecommended next action
HealthUNKNOWNUNKNOWNNo evidence collected. Not assumed healthy.neverPoint a safe read-only probe at the service (or expose it through DCP /health) before claiming health.
BackupUNKNOWNUNKNOWNBackup stack installed but the timer is disabled/inactive and the first real backup has not run (STATIC fact). Not verified.neverRun the first real backup after the soak/backup gates pass; until then backup evidence stays UNKNOWN.
Restore TestUNKNOWNUNKNOWNNo restore drill has been executed against this service yet. Restore stays UNKNOWN until a real drill artifact exists.neverRun a disposable restore drill and record the PASS/FAIL artifact; never claim restore PASS from a design alone.
MonitoringWARNINGSTATICNetdata is present for metrics, but scraped coverage and alert routing are unverified.neverRoute a verified alert to the owner (Uptime Kuma / n8n -> Telegram) and record the first real alert.
RollbackUNKNOWNUNKNOWNNo evidence collected. Not assumed healthy.neverConfirm the rollback procedure against the running image/version and record a drill result.
SecretsUNKNOWNUNKNOWNNo evidence collected. Not assumed healthy.neverConfirm secret storage/isolation with a read-only review; never render secret contents.
Missing evidence: Health (UNKNOWN), Backup (UNKNOWN), Restore Test (UNKNOWN), Monitoring (WARNING), Rollback (UNKNOWN), Secrets (UNKNOWN)

Only PASS counts as verified. UNKNOWN, WARNING, FAIL and NOT_APPLICABLE never count as a pass. STATIC = documented/committed evidence, not a live probe.

Homepage

0/6 verified0% verified
DimensionStatusSourceEvidenceLast verificationRecommended next action
HealthUNKNOWNUNKNOWNNo evidence collected. Not assumed healthy.neverPoint a safe read-only probe at the service (or expose it through DCP /health) before claiming health.
BackupUNKNOWNUNKNOWNBackup stack installed but the timer is disabled/inactive and the first real backup has not run (STATIC fact). Not verified.neverRun the first real backup after the soak/backup gates pass; until then backup evidence stays UNKNOWN.
Restore TestUNKNOWNUNKNOWNNo restore drill has been executed against this service yet. Restore stays UNKNOWN until a real drill artifact exists.neverRun a disposable restore drill and record the PASS/FAIL artifact; never claim restore PASS from a design alone.
MonitoringUNKNOWNUNKNOWNNo evidence collected. Not assumed healthy.neverRoute a verified alert to the owner (Uptime Kuma / n8n -> Telegram) and record the first real alert.
RollbackUNKNOWNUNKNOWNNo evidence collected. Not assumed healthy.neverConfirm the rollback procedure against the running image/version and record a drill result.
SecretsUNKNOWNUNKNOWNDashboard may hold service keys; storage/isolation not verified.neverConfirm secret storage/isolation with a read-only review; never render secret contents.
Missing evidence: Health (UNKNOWN), Backup (UNKNOWN), Restore Test (UNKNOWN), Monitoring (UNKNOWN), Rollback (UNKNOWN), Secrets (UNKNOWN)

Only PASS counts as verified. UNKNOWN, WARNING, FAIL and NOT_APPLICABLE never count as a pass. STATIC = documented/committed evidence, not a live probe.

9Router

0/6 verified0% verified
DimensionStatusSourceEvidenceLast verificationRecommended next action
HealthUNKNOWNUNKNOWNNo evidence collected. Not assumed healthy.neverPoint a safe read-only probe at the service (or expose it through DCP /health) before claiming health.
BackupUNKNOWNUNKNOWNBackup stack installed but the timer is disabled/inactive and the first real backup has not run (STATIC fact). Not verified.neverRun the first real backup after the soak/backup gates pass; until then backup evidence stays UNKNOWN.
Restore TestUNKNOWNUNKNOWNNo restore drill has been executed against this service yet. Restore stays UNKNOWN until a real drill artifact exists.neverRun a disposable restore drill and record the PASS/FAIL artifact; never claim restore PASS from a design alone.
MonitoringUNKNOWNUNKNOWNNo evidence collected. Not assumed healthy.neverRoute a verified alert to the owner (Uptime Kuma / n8n -> Telegram) and record the first real alert.
RollbackUNKNOWNUNKNOWNNo evidence collected. Not assumed healthy.neverConfirm the rollback procedure against the running image/version and record a drill result.
SecretsUNKNOWNUNKNOWNNo evidence collected. Not assumed healthy.neverConfirm secret storage/isolation with a read-only review; never render secret contents.
Missing evidence: Health (UNKNOWN), Backup (UNKNOWN), Restore Test (UNKNOWN), Monitoring (UNKNOWN), Rollback (UNKNOWN), Secrets (UNKNOWN)

Only PASS counts as verified. UNKNOWN, WARNING, FAIL and NOT_APPLICABLE never count as a pass. STATIC = documented/committed evidence, not a live probe.