feat(backup-system): migración completa a GitOps

- CronJob rclone-mega-backup entra en git (solo existía imperativo) - Scripts backup.sh/verify.sh/rclone-mega.sh como ConfigMap backup-scripts (antes hostPath /data/backups/scripts — cambios ahora pasan por git+ArgoCD) - rclone.conf vía InfisicalStaticSecret (homelab/prod//backup-system → RCLONE_CONF), sustituye al hostPath de ~/.config/rclone/rclone.conf - Imágenes pinneadas: rclone 1.74.3, bitnami/kubectl por digest (v1.36.2) - ClusterRole/CRB de backup-sa entran en git - Application ArgoCD backup-system (auto-sync + prune + selfHeal) Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-07-09 09:33:35 +00:00
parent ae0b5f9492
commit d1d2fac287
8 changed files with 286 additions and 12 deletions
@@ -0,0 +1,23 @@
+apiVersion: argoproj.io/v1alpha1
+kind: Application
+metadata:
+  name: backup-system
+  namespace: argocd
+  annotations:
+    notifications.argoproj.io/subscribe.on-sync-failed.telegram: "5138407666"
+    notifications.argoproj.io/subscribe.on-health-degraded.telegram: "5138407666"
+spec:
+  project: default
+  source:
+    repoURL: https://git.chemavx.xyz/chemavx/k8s-manifests
+    targetRevision: HEAD
+    path: backup-system
+  destination:
+    server: https://kubernetes.default.svc
+    namespace: backup-system
+  syncPolicy:
+    automated:
+      prune: true
+      selfHeal: true
+    syncOptions:
+      - CreateNamespace=true
@@ -0,0 +1,15 @@
+apiVersion: rbac.authorization.k8s.io/v1
+kind: ClusterRole
+metadata:
+  name: backup-role
+rules:
+  - apiGroups: [""]
+    resources:
+      - pods
+      - pods/exec
+      - persistentvolumeclaims
+      - persistentvolumes
+    verbs:
+      - get
+      - list
+      - create
@@ -0,0 +1,12 @@
+apiVersion: rbac.authorization.k8s.io/v1
+kind: ClusterRoleBinding
+metadata:
+  name: backup-role-binding
+roleRef:
+  apiGroup: rbac.authorization.k8s.io
+  kind: ClusterRole
+  name: backup-role
+subjects:
+  - kind: ServiceAccount
+    name: backup-sa
+    namespace: backup-system
@@ -0,0 +1,143 @@
+apiVersion: v1
+data:
+  backup.sh: |
+    #!/bin/bash
+    set -e
+
+    BACKUP_DIR="/data/backups/backups"
+    DATE=$(date +%Y-%m-%d_%H-%M-%S)
+    KUBECONFIG=/etc/rancher/k3s/k3s.yaml
+
+    mkdir -p "$BACKUP_DIR"
+
+    log() { echo "[$(date '+%Y-%m-%d %H:%M:%S')] $*"; }
+
+    # === n8n ===
+    log "Backing up n8n..."
+    tar -czf "$BACKUP_DIR/n8n_${DATE}.tar.gz" -C /data/n8n . 2>/dev/null && log "n8n OK"
+
+    # === openclaw ===
+    log "Backing up openclaw..."
+    tar -czf "$BACKUP_DIR/openclaw_${DATE}.tar.gz" -C /data/openclaw . 2>/dev/null && log "openclaw OK"
+
+    # === homarr ===
+    log "Backing up homarr..."
+    HOMARR_DB_PV=$(kubectl get pvc -n homarr homarr-db-pvc -o jsonpath='{.spec.volumeName}' 2>/dev/null)
+    HOMARR_REDIS_PV=$(kubectl get pvc -n homarr homarr-redis-pvc -o jsonpath='{.spec.volumeName}' 2>/dev/null)
+    mkdir -p /tmp/backup-homarr/db /tmp/backup-homarr/redis
+    cp /var/lib/rancher/k3s/storage/${HOMARR_DB_PV}_homarr_homarr-db-pvc/db.sqlite /tmp/backup-homarr/db/ 2>/dev/null || true
+    cp /var/lib/rancher/k3s/storage/${HOMARR_REDIS_PV}_homarr_homarr-redis-pvc/dump.rdb /tmp/backup-homarr/redis/ 2>/dev/null || true
+    tar -czf "$BACKUP_DIR/homarr_${DATE}.tar.gz" -C /tmp/backup-homarr . 2>/dev/null && log "homarr OK"
+    rm -rf /tmp/backup-homarr
+
+    # === authentik PostgreSQL ===
+    log "Backing up authentik postgres..."
+    kubectl exec -n authentik postgresql-0 -- pg_dump -U authentik authentik 2>/dev/null | \
+      gzip > "$BACKUP_DIR/authentik-pg_${DATE}.sql.gz" && log "authentik-pg OK"
+
+    # === polymarket PostgreSQL === (retirado 2026-07-06, decomisión del bot)
+    # postgres escalado a 0; dump definitivo verificado en
+    # $BACKUP_DIR/polymarket-decommission/ (+ último nocturno .last-nightly).
+
+    # === grafana ===
+    log "Backing up grafana..."
+    GRAFANA_PV=$(kubectl get pvc -n monitoring kube-prometheus-stack-grafana -o jsonpath='{.spec.volumeName}' 2>/dev/null)
+    mkdir -p /tmp/backup-grafana
+    cp /var/lib/rancher/k3s/storage/${GRAFANA_PV}_monitoring_kube-prometheus-stack-grafana/grafana.db /tmp/backup-grafana/ 2>/dev/null || true
+    tar -czf "$BACKUP_DIR/grafana_${DATE}.tar.gz" -C /tmp/backup-grafana . 2>/dev/null && log "grafana OK"
+    rm -rf /tmp/backup-grafana
+
+    # === k3s SQLite ===
+    log "Backing up k3s SQLite..."
+    tar -czf "$BACKUP_DIR/k3s-db_${DATE}.tar.gz" \
+      /data/k3s-db/state.db \
+      /data/k3s-db/state.db-shm \
+      /data/k3s-db/state.db-wal 2>/dev/null \
+      && log "k3s-db OK" \
+      || log "k3s-db WARNING: backup parcial (WAL/SHM pueden estar ausentes si DB está idle)"
+
+    # === Cleanup old backups (keep last 7 per service) ===
+    log "Cleaning old backups..."
+    for prefix in n8n openclaw homarr authentik-pg grafana k3s-db; do
+      find "$BACKUP_DIR" \( -name "${prefix}_*.tar.gz" -o -name "${prefix}_*.sql.gz" \) | \
+        sort | head -n -7 | \
+        xargs -r rm -f
+    done
+
+    log "Backup complete!"
+  rclone-mega.sh: |
+    #!/bin/bash
+    set -e
+
+    log() { echo "[$(date '+%Y-%m-%d %H:%M:%S')] $*"; }
+
+    log "Iniciando sync a Mega..."
+    log "Origen: /data/backups/backups"
+    log "Destino: mega:k3s-backups/"
+
+    rclone sync /data/backups/backups mega:k3s-backups/ \
+      --config /rclone/rclone.conf \
+      --log-level INFO \
+      --stats 30s
+
+    log "Sync completado."
+    log "Archivos en Mega:"
+    rclone ls mega:k3s-backups/ --config /rclone/rclone.conf
+  verify.sh: |
+    #!/bin/sh
+    set -e
+
+    log() { echo "[$(date '+%Y-%m-%d %H:%M:%S')] $*"; }
+
+    # polymarket-pg retirado el 2026-07-06 (decomisión del bot)
+    EXPECTED_SERVICES="n8n openclaw homarr authentik-pg grafana k3s-db"
+    ERRORS=0
+
+    # Ventana de frescura: el sync a Mega es semanal (domingos 03:00), así que
+    # un backup es "reciente" si su fecha está en los últimos 8 días (0..7 atrás).
+    # BusyBox date no soporta -d 'yesterday'; se calcula con epoch (-d @segundos).
+    NOW=$(date +%s)
+    ACCEPTED_DATES=""
+    for i in $(seq 0 7); do
+      ACCEPTED_DATES="$ACCEPTED_DATES $(date -d @$(( NOW - i * 86400 )) +%Y-%m-%d)"
+    done
+
+    log "=== Verificación de backups en Mega ==="
+    log "Listando archivos en mega:k3s-backups/..."
+
+    FILES=$(rclone ls mega:k3s-backups/ --config /rclone/rclone.conf 2>&1)
+    log "$FILES"
+
+    log "---"
+    log "Fechas aceptadas (ventana de 8 días):$ACCEPTED_DATES"
+
+    for svc in $EXPECTED_SERVICES; do
+      FOUND=0
+      for d in $ACCEPTED_DATES; do
+        if echo "$FILES" | grep -q "${svc}_${d}"; then
+          FOUND=1
+          break
+        fi
+      done
+      if [ "$FOUND" -eq 1 ]; then
+        log "[OK] $svc — backup reciente encontrado"
+      else
+        log "[WARN] $svc — sin backup en los últimos 8 días"
+        ERRORS=$((ERRORS + 1))
+      fi
+    done
+
+    log "---"
+    TOTAL=$(echo "$FILES" | grep -c "tar.gz\|sql.gz" || true)
+    log "Total archivos en Mega: $TOTAL"
+
+    if [ "$ERRORS" -gt 0 ]; then
+      log "RESULTADO: $ERRORS servicio(s) sin backup reciente. Revisar."
+      exit 1
+    fi
+
+    log "RESULTADO: Todos los backups verificados correctamente."
+kind: ConfigMap
+metadata:
+  name: backup-scripts
+  namespace: backup-system
@@ -19,7 +19,8 @@ spec:
            runAsUser: 0
          containers:
            - name: verify
-              image: rclone/rclone:latest
+              # OJO: imagen BusyBox — verify.sh debe evitar GNU-ismos (date -d 'yesterday')
+              image: docker.io/rclone/rclone:1.74.3
              command: ["/bin/sh", "/scripts/verify.sh"]
              resources:
                limits:
@@ -29,16 +30,18 @@ spec:
                  cpu: 50m
                  memory: 64Mi
              volumeMounts:
-                - mountPath: /rclone/rclone.conf
+                - mountPath: /rclone
                  name: rclone-conf
                - mountPath: /scripts
                  name: scripts
          volumes:
            - name: rclone-conf
-              hostPath:
-                path: /home/chemavx/.config/rclone/rclone.conf
-                type: File
+              secret:
+                secretName: rclone-conf-infisical
+                items:
+                  - key: RCLONE_CONF
+                    path: rclone.conf
            - name: scripts
-              hostPath:
-                path: /data/backups/scripts
-                type: Directory
+              configMap:
+                name: backup-scripts
+                defaultMode: 0755
@@ -20,7 +20,9 @@ spec:
            kubernetes.io/hostname: chemavx-k8
          containers:
          - name: backup
-            image: bitnami/kubectl:latest
+            # bitnami/kubectl v1.36.2 — pinneado por digest: los tags de bitnami
+            # en Docker Hub ya no son fiables (migración Broadcom/bitnamilegacy)
+            image: docker.io/bitnami/kubectl@sha256:558420daf32bbc382e3e9af4537f4073085b336ddd47399a3b70e70087115978
            command: ["/bin/bash", "/scripts/backup.sh"]
            env:
            - name: KUBECONFIG
@@ -62,9 +64,9 @@ spec:
              path: /data/openclaw
              type: DirectoryOrCreate
          - name: scripts
-            hostPath:
-              path: /data/backups/scripts
-              type: Directory
+            configMap:
+              name: backup-scripts
+              defaultMode: 0755
          - name: k3s-storage
            hostPath:
              path: /var/lib/rancher/k3s/storage
@@ -0,0 +1,52 @@
+apiVersion: batch/v1
+kind: CronJob
+metadata:
+  name: rclone-mega-backup
+  namespace: backup-system
+spec:
+  schedule: "0 3 * * 0"
+  concurrencyPolicy: Forbid
+  failedJobsHistoryLimit: 2
+  successfulJobsHistoryLimit: 3
+  jobTemplate:
+    spec:
+      template:
+        spec:
+          nodeSelector:
+            kubernetes.io/hostname: chemavx-k8
+          restartPolicy: OnFailure
+          securityContext:
+            runAsUser: 0
+          containers:
+            - name: rclone
+              image: docker.io/rclone/rclone:1.74.3
+              command: ["/bin/sh", "/scripts/rclone-mega.sh"]
+              resources:
+                requests:
+                  cpu: 100m
+                  memory: 128Mi
+                limits:
+                  cpu: 500m
+                  memory: 512Mi
+              volumeMounts:
+                - mountPath: /data/backups/backups
+                  name: backup-dir
+                - mountPath: /rclone
+                  name: rclone-conf
+                - mountPath: /scripts
+                  name: scripts
+          volumes:
+            - name: backup-dir
+              hostPath:
+                path: /data/backups/backups
+                type: DirectoryOrCreate
+            - name: rclone-conf
+              secret:
+                secretName: rclone-conf-infisical
+                items:
+                  - key: RCLONE_CONF
+                    path: rclone.conf
+            - name: scripts
+              configMap:
+                name: backup-scripts
+                defaultMode: 0755
@@ -0,0 +1,24 @@
+apiVersion: secrets.infisical.com/v1beta1
+kind: InfisicalStaticSecret
+metadata:
+  name: rclone-conf
+  namespace: backup-system
+spec:
+  # Identidad compartida read-only (sin write grant) — el valor RCLONE_CONF
+  # se gestiona desde la UI de Infisical (proyecto homelab / prod / /backup-system).
+  infisicalAuthRef:
+    name: infisical-auth
+    namespace: infisical-operator
+  sources:
+    - projectId: 17e98e9d-70f5-43d1-8382-7da818dfcdd0   # project "homelab"
+      environmentSlug: prod
+      secretPath: /backup-system
+  syncOptions:
+    refreshInterval: 60s
+  targets:
+    - kind: Secret
+      name: rclone-conf-infisical
+      namespace: backup-system
+      creationPolicy: Owner
+      # El Secret lleva la key RCLONE_CONF con el contenido íntegro de rclone.conf;
+      # los CronJobs la montan como fichero /rclone/rclone.conf (items en el volume).