Add restore handbook and Unraid job guide
This commit is contained in:
@@ -8,6 +8,7 @@ Verwandte Dokumente:
|
|||||||
|
|
||||||
- `docs/ROLLBACK.md` - Rueckweg bei Fehlern im laufenden GitOps-Betrieb
|
- `docs/ROLLBACK.md` - Rueckweg bei Fehlern im laufenden GitOps-Betrieb
|
||||||
- `docs/RESTORE_MATRIX.md` - Restore-Quellen und Verifikationsregeln pro Dienst
|
- `docs/RESTORE_MATRIX.md` - Restore-Quellen und Verifikationsregeln pro Dienst
|
||||||
|
- `docs/RESTORE_HANDBOOK.md` - praktische Restore-Betriebsanleitung
|
||||||
- `ops/borg-ui/BACKUP_SCOPE.md` - Zielbild des Borg-Scopes
|
- `ops/borg-ui/BACKUP_SCOPE.md` - Zielbild des Borg-Scopes
|
||||||
|
|
||||||
---
|
---
|
||||||
|
|||||||
@@ -0,0 +1,200 @@
|
|||||||
|
# Restore Handbook - KalliLab CORE
|
||||||
|
|
||||||
|
Stand: 2026-05-07
|
||||||
|
|
||||||
|
Dieses Handbuch ist die praktische Betriebsanleitung fuer Restore-Checks und Restore-Lab in KalliLab CORE.
|
||||||
|
|
||||||
|
Es ergaenzt:
|
||||||
|
|
||||||
|
- `docs/RESTORE_MATRIX.md`
|
||||||
|
- `docs/DISASTER_RECOVERY.md`
|
||||||
|
- `ops/restore-tests/*`
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 1. Ziel
|
||||||
|
|
||||||
|
Dieses Handbuch beantwortet vier Fragen:
|
||||||
|
|
||||||
|
1. Was ist die Restore-Quelle?
|
||||||
|
2. Wo wird getestet?
|
||||||
|
3. Wie pruefen wir Erfolg?
|
||||||
|
4. Wie machen wir das regelmaessig mit wenig Handarbeit?
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 2. Grundmuster
|
||||||
|
|
||||||
|
Alle validierten Restore-Tests folgen demselben Muster:
|
||||||
|
|
||||||
|
- Quelle bleibt das produktive Borg-Repo bei Hetzner
|
||||||
|
- Borg-Zugriff laeuft ueber den vorhandenen `borg-ui`-Container
|
||||||
|
- Passphrase kommt aus `/mnt/user/appdata/secrets/borg_repo_passphrase.txt`
|
||||||
|
- Testdaten landen unter `/mnt/user/backups/restore-lab/<dienst>`
|
||||||
|
- Reports landen unter `/mnt/user/backups/restore-reports`
|
||||||
|
- Testinstanzen laufen lokal ohne Traefik und ohne produktive Domain
|
||||||
|
- nach Erfolg werden Testcontainer und Testdaten wieder entfernt
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 3. Bereits praktisch verifiziert
|
||||||
|
|
||||||
|
### Vaultwarden
|
||||||
|
|
||||||
|
- Report: `/mnt/user/backups/restore-reports/vaultwarden-2026-05-07.md`
|
||||||
|
- Nachweis:
|
||||||
|
- Borg-Restore erfolgreich
|
||||||
|
- Testcontainer startete
|
||||||
|
- Login-Seite war erreichbar
|
||||||
|
|
||||||
|
### Gitea
|
||||||
|
|
||||||
|
- Report: `/mnt/user/backups/restore-reports/gitea-2026-05-07.md`
|
||||||
|
- Nachweis:
|
||||||
|
- Borg-Restore erfolgreich
|
||||||
|
- Web-UI antwortete
|
||||||
|
- SSH-Port reagierte
|
||||||
|
|
||||||
|
### Paperless
|
||||||
|
|
||||||
|
- Report: `/mnt/user/backups/restore-reports/paperless-2026-05-07.md`
|
||||||
|
- Nachweis:
|
||||||
|
- Borg-Datei-Restore erfolgreich
|
||||||
|
- Paperless-Dump aus Borg importiert
|
||||||
|
- Login-Seite war erreichbar
|
||||||
|
- Test-DB enthielt `25` Dokumente
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 4. Verzeichnisstruktur
|
||||||
|
|
||||||
|
### Produktiv
|
||||||
|
|
||||||
|
- `/mnt/user/appdata`
|
||||||
|
- `/mnt/user/services`
|
||||||
|
- `/mnt/user/documents`
|
||||||
|
- `/mnt/user/backups/borg/dumps/latest`
|
||||||
|
|
||||||
|
### Restore-Lab
|
||||||
|
|
||||||
|
- `/mnt/user/backups/restore-lab/vaultwarden`
|
||||||
|
- `/mnt/user/backups/restore-lab/gitea`
|
||||||
|
- `/mnt/user/backups/restore-lab/paperless`
|
||||||
|
|
||||||
|
### Reports
|
||||||
|
|
||||||
|
- `/mnt/user/backups/restore-reports`
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 5. Restore-Frequenz
|
||||||
|
|
||||||
|
- jeden Montag, 06:30:
|
||||||
|
- Frische-Check fuer Dumps und Reports
|
||||||
|
- 1. Samstag im Monat, 07:00:
|
||||||
|
- Vaultwarden
|
||||||
|
- 3. Samstag im Monat, 07:00:
|
||||||
|
- Gitea
|
||||||
|
- jeder 2. Monat, 2. Samstag, 08:00:
|
||||||
|
- Paperless
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 6. Betriebsmodi
|
||||||
|
|
||||||
|
### V1
|
||||||
|
|
||||||
|
- manuell validierte Restore-Pfade
|
||||||
|
- Host-Job-Definitionen liegen im Repo
|
||||||
|
- Scheduler kann bereits Plan- und Frische-Checks fahren
|
||||||
|
- volle Automatik je Dienst wird danach gezielt nachgezogen
|
||||||
|
|
||||||
|
### V2
|
||||||
|
|
||||||
|
- echte Vollautomatik fuer die drei validierten Dienste
|
||||||
|
- `ntfy` bei Erfolg/Fehler
|
||||||
|
- Hermes liest spaeter Reports und baut Uebersichten
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 7. User Script Jobs auf Unraid
|
||||||
|
|
||||||
|
Die Vorlagen stehen in:
|
||||||
|
|
||||||
|
- `ops/restore-tests/unraid-user-scripts.md`
|
||||||
|
|
||||||
|
Host-Repo-Pfad:
|
||||||
|
|
||||||
|
```text
|
||||||
|
/mnt/user/services/homelab
|
||||||
|
```
|
||||||
|
|
||||||
|
V1-Jobs:
|
||||||
|
|
||||||
|
1. `restore-freshness-weekly`
|
||||||
|
2. `restore-vaultwarden-monthly`
|
||||||
|
3. `restore-gitea-monthly`
|
||||||
|
4. `restore-paperless-bimonthly`
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 8. Erfolgskriterien
|
||||||
|
|
||||||
|
Ein Restore-Test gilt nur dann als erfolgreich, wenn:
|
||||||
|
|
||||||
|
- Restore-Quelle lesbar war
|
||||||
|
- Daten im Restore-Lab ankamen
|
||||||
|
- Testcontainer startete
|
||||||
|
- Smoke-Test erfolgreich war
|
||||||
|
- Report geschrieben wurde
|
||||||
|
|
||||||
|
Nur `Container laeuft` reicht nicht.
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 9. Sicherheitsregeln
|
||||||
|
|
||||||
|
- keine produktiven Pfade beschreiben
|
||||||
|
- keine produktiven Container fuer Restore-Tests verwenden
|
||||||
|
- keine produktiven Domains fuer Testinstanzen verwenden
|
||||||
|
- keine Secrets im Repo
|
||||||
|
- keine Restore-Automatik fuer neue Dienste ohne bewusste Freigabe
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 10. Schnellstart
|
||||||
|
|
||||||
|
### Frische-Check
|
||||||
|
|
||||||
|
Auf dem Unraid-Host:
|
||||||
|
|
||||||
|
```bash
|
||||||
|
pwsh -File /mnt/user/services/homelab/ops/restore-tests/run-restore-checks.ps1 -Mode freshness
|
||||||
|
```
|
||||||
|
|
||||||
|
### Vaultwarden Planlauf
|
||||||
|
|
||||||
|
```bash
|
||||||
|
pwsh -File /mnt/user/services/homelab/ops/restore-tests/run-restore-checks.ps1 -Mode vaultwarden -WhatIf
|
||||||
|
```
|
||||||
|
|
||||||
|
### Gitea Planlauf
|
||||||
|
|
||||||
|
```bash
|
||||||
|
pwsh -File /mnt/user/services/homelab/ops/restore-tests/run-restore-checks.ps1 -Mode gitea -WhatIf
|
||||||
|
```
|
||||||
|
|
||||||
|
### Paperless Planlauf
|
||||||
|
|
||||||
|
```bash
|
||||||
|
pwsh -File /mnt/user/services/homelab/ops/restore-tests/run-restore-checks.ps1 -Mode paperless -WhatIf
|
||||||
|
```
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
## 11. Naechste Ausbaustufen
|
||||||
|
|
||||||
|
1. Vollautomatik fuer Vaultwarden, Gitea und Paperless
|
||||||
|
2. `ntfy`-Meldungen fuer Erfolg/Fehler
|
||||||
|
3. Hermes-Zusammenfassung ueber vorhandene Reports
|
||||||
|
4. naechster Referenz-Restore fuer `mail-archiver` oder `mealie`
|
||||||
@@ -0,0 +1,106 @@
|
|||||||
|
# Unraid User Scripts fuer Restore-Checks
|
||||||
|
|
||||||
|
## Ziel
|
||||||
|
|
||||||
|
Diese Vorlagen binden die validierten Restore-Checks in Unraid User Scripts ein.
|
||||||
|
|
||||||
|
Host-Repo-Pfad:
|
||||||
|
|
||||||
|
```text
|
||||||
|
/mnt/user/services/homelab
|
||||||
|
```
|
||||||
|
|
||||||
|
## Script 1 - `restore-freshness-weekly`
|
||||||
|
|
||||||
|
Zeit:
|
||||||
|
|
||||||
|
- Montag, 06:30
|
||||||
|
|
||||||
|
Inhalt:
|
||||||
|
|
||||||
|
```bash
|
||||||
|
#!/bin/bash
|
||||||
|
pwsh -File /mnt/user/services/homelab/ops/restore-tests/run-restore-checks.ps1 -Mode freshness \
|
||||||
|
> /mnt/user/backups/restore-reports/freshness-$(date +%F).md
|
||||||
|
```
|
||||||
|
|
||||||
|
Erwartung:
|
||||||
|
|
||||||
|
- prueft Dump-Frische
|
||||||
|
- prueft Report-Frische
|
||||||
|
- startet keine Container
|
||||||
|
|
||||||
|
## Script 2 - `restore-vaultwarden-monthly`
|
||||||
|
|
||||||
|
Zeit:
|
||||||
|
|
||||||
|
- 1. Samstag im Monat, 07:00
|
||||||
|
|
||||||
|
V1-Inhalt:
|
||||||
|
|
||||||
|
```bash
|
||||||
|
#!/bin/bash
|
||||||
|
pwsh -File /mnt/user/services/homelab/ops/restore-tests/run-restore-checks.ps1 -Mode vaultwarden -WhatIf \
|
||||||
|
> /mnt/user/backups/restore-reports/vaultwarden-plan-$(date +%F).md
|
||||||
|
```
|
||||||
|
|
||||||
|
## Script 3 - `restore-gitea-monthly`
|
||||||
|
|
||||||
|
Zeit:
|
||||||
|
|
||||||
|
- 3. Samstag im Monat, 07:00
|
||||||
|
|
||||||
|
V1-Inhalt:
|
||||||
|
|
||||||
|
```bash
|
||||||
|
#!/bin/bash
|
||||||
|
pwsh -File /mnt/user/services/homelab/ops/restore-tests/run-restore-checks.ps1 -Mode gitea -WhatIf \
|
||||||
|
> /mnt/user/backups/restore-reports/gitea-plan-$(date +%F).md
|
||||||
|
```
|
||||||
|
|
||||||
|
## Script 4 - `restore-paperless-bimonthly`
|
||||||
|
|
||||||
|
Zeit:
|
||||||
|
|
||||||
|
- jeder 2. Monat, 2. Samstag, 08:00
|
||||||
|
|
||||||
|
V1-Inhalt:
|
||||||
|
|
||||||
|
```bash
|
||||||
|
#!/bin/bash
|
||||||
|
pwsh -File /mnt/user/services/homelab/ops/restore-tests/run-restore-checks.ps1 -Mode paperless -WhatIf \
|
||||||
|
> /mnt/user/backups/restore-reports/paperless-plan-$(date +%F).md
|
||||||
|
```
|
||||||
|
|
||||||
|
## Warum V1 mit `-WhatIf`
|
||||||
|
|
||||||
|
- keine unkontrollierten Restore-Laeufe im Cron
|
||||||
|
- erst Host-Scheduler sauber verdrahten
|
||||||
|
- spaeter gezielt auf echte Vollautomatik umstellen
|
||||||
|
|
||||||
|
## V2 Zielbild
|
||||||
|
|
||||||
|
Spaeter werden die drei Restore-Scripts von Plan-/Scaffold-Modus auf echte Host-Ausfuehrung umgestellt:
|
||||||
|
|
||||||
|
1. Restore aus Borg
|
||||||
|
2. Testcontainer starten
|
||||||
|
3. Smoke-Test
|
||||||
|
4. Report schreiben
|
||||||
|
5. optional `ntfy`
|
||||||
|
6. Bereinigung
|
||||||
|
|
||||||
|
## Optionales `ntfy` Wrapper-Muster
|
||||||
|
|
||||||
|
Wenn `ntfy` spaeter dazukommt, soll der Host-Job nur Erfolg/Fehler referenzieren, nicht den ganzen Report in die Nachricht kippen.
|
||||||
|
|
||||||
|
Beispiel:
|
||||||
|
|
||||||
|
```bash
|
||||||
|
#!/bin/bash
|
||||||
|
REPORT="/mnt/user/backups/restore-reports/freshness-$(date +%F).md"
|
||||||
|
if pwsh -File /mnt/user/services/homelab/ops/restore-tests/run-restore-checks.ps1 -Mode freshness > "$REPORT"; then
|
||||||
|
echo "Restore freshness check ok: $REPORT"
|
||||||
|
else
|
||||||
|
echo "Restore freshness check failed: $REPORT"
|
||||||
|
fi
|
||||||
|
```
|
||||||
Reference in New Issue
Block a user