it is what it is

mist_infra/scripts/backup_k8s_storage_to_s3.sh

@@ -1,42 +1,96 @@
 #!/usr/bin/env bash
 set -euo pipefail
 
-# ----------------------------------------------------------
+# --------------------------------------------------
-# Weekly full backup of all Kubernetes PV data
+# Config
-# ----------------------------------------------------------
+# --------------------------------------------------
+BACKUP_DATE="$(date +%Y-%m-%d)"
+TMP_DIR="/tmp/backup-${BACKUP_DATE}"
 
-SOURCE_DIR="/k8s_storage"
+K8S_STORAGE="/k8s_storage"
-TMP_DIR="/tmp/mist-backups"
+S3_BASE="s3://mist-backups/${BACKUP_DATE}"
-BUCKET="mist-backups"
 
-DATE="$(date -u +%Y-%m-%d)"
+POSTGRES_NAMESPACE="default"
-TIMESTAMP="$(date -u +%Y-%m-%d_%H-%M-%S)"
+POSTGRES_POD_LABEL="app=postgres"
-HOST="$(hostname)"
+POSTGRES_USER="postgres"
+POSTGRES_DB="stripe_invoice"
 
-ARCHIVE_NAME="k8s_storage_${HOST}_${TIMESTAMP}.tar.gz"
+mkdir -p "$TMP_DIR/postgres"
-ARCHIVE_PATH="${TMP_DIR}/${ARCHIVE_NAME}"
 
-echo "=== Mist weekly PV backup ==="
+echo "=== Backup date: $BACKUP_DATE ==="
-echo "Source:   ${SOURCE_DIR}"
-echo "Archive:  ${ARCHIVE_PATH}"
-echo "Bucket:   s3://${BUCKET}/${DATE}/"
 
-mkdir -p "${TMP_DIR}"
+# --------------------------------------------------
+# 1. Discover schemas (for documentation)
+# --------------------------------------------------
+echo "=== Discovering Postgres schemas ==="
+
+POSTGRES_POD="$(kubectl get pods -n "$POSTGRES_NAMESPACE" \
+  -l "$POSTGRES_POD_LABEL" \
+  -o jsonpath='{.items[0].metadata.name}')"
+
+SCHEMAS="$(kubectl exec -n "$POSTGRES_NAMESPACE" "$POSTGRES_POD" -- \
+  psql -U "$POSTGRES_USER" -d "$POSTGRES_DB" -Atc \
+  "SELECT schema_name FROM information_schema.schemata
+   WHERE schema_name NOT IN ('pg_catalog', 'information_schema')
+   ORDER BY schema_name;")"
+
+# Write README
+{
+  echo "Database: ${POSTGRES_DB}"
+  echo "Schemas:"
+  for s in $SCHEMAS; do
+    echo "- $s"
+  done
+  echo
+  echo "Backup date: ${BACKUP_DATE}"
+  echo "Host: mist"
+} > "$TMP_DIR/postgres/README.txt"
+
+echo "✓ Schemas documented"
+
+# --------------------------------------------------
+# 2. Postgres logical backup (FULL DB)
+# --------------------------------------------------
+echo "=== Dumping Postgres database ==="
+
+kubectl exec -n "$POSTGRES_NAMESPACE" "$POSTGRES_POD" -- \
+  pg_dump -U "$POSTGRES_USER" "$POSTGRES_DB" \
+  | gzip > "$TMP_DIR/postgres/stripe_invoice.sql.gz"
+
+echo "✓ Postgres dump complete"
+
+# --------------------------------------------------
+# 3. Filesystem backup (best-effort)
+# --------------------------------------------------
+echo "=== Archiving /k8s_storage ==="
 
-echo "→ Creating tarball"
 tar \
-  --numeric-owner \
+  --ignore-failed-read \
-  --xattrs \
+  --warning=no-file-changed \
-  --acls \
+  -czf "$TMP_DIR/k8s_storage.tar.gz" \
-  -czf "${ARCHIVE_PATH}" \
+  -C "$K8S_STORAGE" .
-  -C "${SOURCE_DIR}" .
 
-echo "→ Uploading to S3"
+echo "✓ Filesystem archive complete"
-aws s3 cp \
-  "${ARCHIVE_PATH}" \
-  "s3://${BUCKET}/${DATE}/${ARCHIVE_NAME}"
 
-echo "→ Cleaning up local temp"
+# --------------------------------------------------
-rm -f "${ARCHIVE_PATH}"
+# 4. Upload to S3
+# --------------------------------------------------
+echo "=== Uploading to S3 ==="
 
-echo "✅ Backup complete"
+aws s3 cp "$TMP_DIR/k8s_storage.tar.gz" \
+  "${S3_BASE}/k8s_storage.tar.gz"
+
+aws s3 cp "$TMP_DIR/postgres/stripe_invoice.sql.gz" \
+  "${S3_BASE}/postgres/stripe_invoice.sql.gz"
+
+aws s3 cp "$TMP_DIR/postgres/README.txt" \
+  "${S3_BASE}/postgres/README.txt"
+
+echo "✓ Upload complete"
+
+# --------------------------------------------------
+# 5. Cleanup
+# --------------------------------------------------
+rm -rf "$TMP_DIR"
+
+echo "=== Backup finished successfully ==="