feat(helm): add separate MongoDB backup CronJob for linker_output and webpages_text

build/ci/production-values.yaml

@@ -10,6 +10,7 @@ restore:
 backup:
   mongo:
     enabled: true
+    linkerOutputEnabled: true
     prefix: ""
     public: true
     bucket: sefaria-mongo-backup

helm-chart/sefaria/templates/configmap/create-linker-output-dumps.yaml

@@ -0,0 +1,69 @@
+{{- if .Values.backup.mongo.linkerOutputEnabled }}
+apiVersion: v1
+kind: ConfigMap
+metadata:
+  name: create-linker-output-dumps-{{ .Values.deployEnv }}
+  labels:
+    deployEnv: "{{ .Values.deployEnv }}"
+    {{- include "sefaria.labels" . | nindent 4 }}
+data:
+  create-linker-output-dumps.sh: |-
+    #!/usr/bin/env bash
+    set -eo pipefail
+
+    {{- if .Values.restore.dynamicName }}
+    DATABASE="${SEFARIA_DB}-{{ .Values.deployEnv }}"
+    {{- else }}
+    DATABASE="${SEFARIA_DB}"
+    {{- end }}
+
+    if [[ -z "${MONGO_HOST:-}" ]]; then
+      echo "Mongo Host not specified"
+      exit 1
+    fi
+
+    URI="mongodb://"
+
+    if [[ -n "${SEFARIA_DB_USER:-}" ]]; then
+      URI="${URI}${SEFARIA_DB_USER}"
+      if [[ -n "${SEFARIA_DB_PASSWORD:-}" ]]; then
+          URI="${URI}:${SEFARIA_DB_PASSWORD}"
+      fi
+      URI="${URI}@"
+    fi
+
+    URI="${URI}${MONGO_HOST}/?ssl=false"
+
+    if [[ -n "${SEFARIA_DB_USER:-}" ]]; then
+      URI="${URI}&authSource=admin"
+    fi
+
+    if [[ -n "${MONGO_REPLICASET_NAME:-}" ]]; then
+      URI="${URI}&replicaSet=${MONGO_REPLICASET_NAME}"
+    fi
+
+    DATADIR="/mongodumps/shared_volume"
+
+    echo "Dumping linker_output collection"
+    until mongodump --uri="$URI" -v -d "$DATABASE" \
+      --collection=linker_output \
+      -o "${DATADIR}/dump"
+    do
+      echo "mongodump failed, retrying in 5s..."
+      rm -rf "${DATADIR}/dump"
+      sleep 5
+    done
+
+    echo "Dumping webpages_text collection"
+    until mongodump --uri="$URI" -v -d "$DATABASE" \
+      --collection=webpages_text \
+      -o "${DATADIR}/dump"
+    do
+      echo "mongodump failed, retrying in 5s..."
+      sleep 5
+    done
+
+    echo "${DATABASE}" > "${DATADIR}/db_name.txt"
+
+    echo "dump complete - upload should start"
+{{- end }}

helm-chart/sefaria/templates/configmap/upload-linker-output-dumps.yaml

@@ -0,0 +1,49 @@
+{{- if .Values.backup.mongo.linkerOutputEnabled }}
+apiVersion: v1
+kind: ConfigMap
+metadata:
+  name: upload-linker-output-dumps-{{ .Values.deployEnv }}
+  labels:
+    deployEnv: "{{ .Values.deployEnv }}"
+    {{- include "sefaria.labels" . | nindent 4 }}
+data:
+  upload-linker-output-dumps.sh: |-
+    #!/usr/bin/env bash
+    set -euo pipefail
+    set -x
+
+    DATADIR="/mongodumps/shared_volume"
+    today="$(date +'%d.%m.%y')"
+    # Keep 5 snapshots — delete the file from 5 days ago
+    five_days_ago="$(date --date='5 days ago' +'%d.%m.%y')"
+
+    if [ ! -f "${DATADIR}/db_name.txt" ]; then
+        echo "ERROR: db_name.txt missing"
+        exit 1
+    fi
+    DB=$(cat "${DATADIR}/db_name.txt")
+    if [ -z "${DB}" ]; then
+        echo "ERROR: db_name.txt is empty"
+        exit 1
+    fi
+
+    # Remove the snapshot from 5 days ago (rotation — keep 5 days)
+    gsutil rm "gs://${BUCKET}/${PREFIX}linker_output_${five_days_ago}.tar.gz" || true
+    gsutil rm "gs://${BUCKET}/${PREFIX}webpages_text_${five_days_ago}.tar.gz" || true
+
+    # Upload linker_output as its own tarball
+    echo "Uploading linker_output dump"
+    tar czf - -C "${DATADIR}" \
+        "./dump/${DB}/linker_output.bson" \
+        "./dump/${DB}/linker_output.metadata.json" \
+      | gsutil cp - "gs://${BUCKET}/${PREFIX}linker_output_${today}.tar.gz"
+
+    # Upload webpages_text as its own tarball
+    echo "Uploading webpages_text dump"
+    tar czf - -C "${DATADIR}" \
+        "./dump/${DB}/webpages_text.bson" \
+        "./dump/${DB}/webpages_text.metadata.json" \
+      | gsutil cp - "gs://${BUCKET}/${PREFIX}webpages_text_${today}.tar.gz"
+
+    echo "Linker output dump upload complete"
+{{- end }}

helm-chart/sefaria/templates/cronjob/mongo-backup-linker-output.yaml

@@ -0,0 +1,100 @@
+{{- if eq .Values.backup.mongo.linkerOutputEnabled true }}
+---
+apiVersion: batch/v1
+kind: CronJob
+metadata:
+  name: {{ .Values.deployEnv }}-mongobackup-linker-output
+  labels:
+    {{- include "sefaria.labels" . | nindent 4 }}
+spec:
+  concurrencyPolicy: Replace
+  schedule: "0 2 * * *"
+  jobTemplate:
+    spec:
+      backoffLimit: 1
+      template:
+        spec:
+          serviceAccount: {{ .Values.backup.mongo.serviceAccount }}
+          affinity:
+            podAffinity:
+              preferredDuringSchedulingIgnoredDuringExecution:
+              - weight: 1
+                podAffinityTerm:
+                  labelSelector:
+                    matchExpressions:
+                    - key: app
+                      operator: In
+                      values:
+                      - mongo
+                  topologyKey: kubernetes.io/hostname
+          tolerations:
+            - key: schedule-on-database-vm
+              operator: "Equal"
+              value: "true"
+              effect: "NoSchedule"
+            - key: mongo-replicaset
+              operator: "Equal"
+              value: "reserved"
+              effect: "NoSchedule"
+          initContainers:
+          - name: mongo-linker-output-dumper
+            image: mongo:{{ .Values.backup.mongo.version }}
+            envFrom:
+            - secretRef:
+                name: {{ .Values.secrets.localSettings.ref }}
+                optional: true
+            - configMapRef:
+                name: local-settings-{{ .Values.deployEnv }}
+            - secretRef:
+                name: local-settings-secrets-{{ .Values.deployEnv }}
+                optional: true
+            volumeMounts:
+            - name: shared-volume
+              mountPath: /mongodumps/shared_volume
+            - name: create-linker-output-dumps-script
+              mountPath: /scripts/create-linker-output-dumps.sh
+              subPath: create-linker-output-dumps.sh
+              readOnly: true
+            command: ["bash"]
+            args: ["-c", "/scripts/create-linker-output-dumps.sh"]
+            resources:
+              requests:
+                ephemeral-storage: 12Gi
+              limits:
+                memory: "500Mi"
+          containers:
+          - name: linker-output-dump-uploader
+            image: google/cloud-sdk
+            volumeMounts:
+            - name: shared-volume
+              mountPath: /mongodumps/shared_volume
+            - name: upload-linker-output-dumps-script
+              mountPath: /scripts/upload-linker-output-dumps.sh
+              subPath: upload-linker-output-dumps.sh
+              readOnly: true
+            env:
+            - name: PREFIX
+              value: {{ .Values.backup.mongo.prefix }}
+            - name: BUCKET
+              value: {{ .Values.backup.mongo.bucket }}
+            command: ["bash"]
+            args: ["-c", "/scripts/upload-linker-output-dumps.sh"]
+            resources:
+              limits:
+                memory: "500Mi"
+          restartPolicy: OnFailure
+          volumes:
+          - name: create-linker-output-dumps-script
+            configMap:
+              name: create-linker-output-dumps-{{ .Values.deployEnv }}
+              defaultMode: 0755
+          - name: upload-linker-output-dumps-script
+            configMap:
+              name: upload-linker-output-dumps-{{ .Values.deployEnv }}
+              defaultMode: 0755
+          - name: shared-volume
+            emptyDir:
+              sizeLimit: 12Gi
+  successfulJobsHistoryLimit: 1
+  failedJobsHistoryLimit: 2
+{{- end }}

helm-chart/sefaria/values.yaml

@@ -49,6 +49,8 @@ restore:
 backup:
   mongo:
     enabled: false
+    linkerOutputEnabled: false
+    historyEnabled: false
     prefix: ""
     public: true
     bucket: sefaria-mongo-backup