fluxer/deploy/helm/infra/templates/geoip-sync-cronjob.yaml

# SPDX-License-Identifier: AGPL-3.0-or-later
#
# Daily sync of the MaxMind GeoLite2 City and ASN MMDB files into S3.
# Runtime services read these out of the CDN bucket. Adopted into helm
# from a previously hand-applied kubectl manifest.

apiVersion: batch/v1
kind: CronJob
metadata:
  name: geoip-sync
  namespace: {{ .Values.global.namespace }}
  labels:
    app.kubernetes.io/name: geoip-sync
    {{- include "fluxer.labels" . | nindent 4 }}
spec:
  schedule: {{ .Values.geoipSync.schedule | quote }}
  concurrencyPolicy: Forbid
  successfulJobsHistoryLimit: 1
  failedJobsHistoryLimit: 3
  jobTemplate:
    spec:
      activeDeadlineSeconds: {{ .Values.geoipSync.activeDeadlineSeconds }}
      backoffLimit: {{ .Values.geoipSync.backoffLimit }}
      template:
        metadata:
          labels:
            app.kubernetes.io/name: geoip-sync
            app.kubernetes.io/part-of: fluxer
        spec:
          restartPolicy: OnFailure
          terminationGracePeriodSeconds: 60
          imagePullSecrets:
            - name: {{ .Values.global.imagePullSecret }}
          containers:
            - name: sync
              image: {{ .Values.geoipSync.image }}
              imagePullPolicy: IfNotPresent
              securityContext:
                allowPrivilegeEscalation: false
                readOnlyRootFilesystem: false
                runAsNonRoot: true
                runAsUser: 1000
                runAsGroup: 1000
                capabilities:
                  drop: ["ALL"]
                seccompProfile:
                  type: RuntimeDefault
              env:
                - name: GEOIP_BUCKET
                  value: {{ .Values.geoipSync.bucket | quote }}
                - name: GEOIP_CITY_UPSTREAM_URL
                  value: {{ .Values.geoipSync.cityUpstreamUrl | quote }}
                - name: GEOIP_ASN_UPSTREAM_URL
                  value: {{ .Values.geoipSync.asnUpstreamUrl | quote }}
              envFrom:
                - secretRef:
                    name: {{ .Values.geoipSync.envSecret }}
              resources:
                requests:
                  cpu: 100m
                  memory: 256Mi
                limits:
                  memory: 512Mi
              command:
                - sh
                - -c
                - |
                  set -eu

                  : "${GEOIP_BUCKET:?missing GEOIP_BUCKET}"
                  : "${GEOIP_CITY_UPSTREAM_URL:?missing GEOIP_CITY_UPSTREAM_URL}"
                  : "${GEOIP_ASN_UPSTREAM_URL:?missing GEOIP_ASN_UPSTREAM_URL}"
                  : "${FLUXER_S3_ACCESS_KEY_ID:?missing FLUXER_S3_ACCESS_KEY_ID}"
                  : "${FLUXER_S3_SECRET_ACCESS_KEY:?missing FLUXER_S3_SECRET_ACCESS_KEY}"
                  : "${FLUXER_S3_ENDPOINT:?missing FLUXER_S3_ENDPOINT}"
                  : "${FLUXER_S3_REGION:?missing FLUXER_S3_REGION}"

                  export AWS_ACCESS_KEY_ID="$FLUXER_S3_ACCESS_KEY_ID"
                  export AWS_SECRET_ACCESS_KEY="$FLUXER_S3_SECRET_ACCESS_KEY"
                  export AWS_DEFAULT_REGION="$FLUXER_S3_REGION"

                  WORKDIR=$(mktemp -d)
                  trap 'rm -rf "$WORKDIR"' EXIT

                  # MMDB files end with the ASCII string "MaxMind.com" after
                  # their metadata marker. Verifying this tail before upload
                  # catches the case where an upstream returns an HTML error
                  # page or a zero-byte body.
                  fetch_and_verify() {
                    local url="$1"
                    local dest="$2"
                    echo "-> fetching $url"
                    curl --fail --location --silent --show-error \
                         --user-agent 'fluxer-geoip-sync/1.0' \
                         --max-time 120 \
                         --output "$dest" \
                         "$url"
                    local size
                    size=$(wc -c < "$dest")
                    if [ "$size" -lt 1024 ]; then
                      echo "refusing to upload ${dest}: file is ${size} bytes, too small" >&2
                      return 1
                    fi
                    if ! tail -c 2048 "$dest" | grep -q "MaxMind.com"; then
                      echo "refusing to upload ${dest}: MaxMind.com marker not found in trailer" >&2
                      return 1
                    fi
                    echo "   ok (${size} bytes)"
                  }

                  fetch_and_verify "$GEOIP_CITY_UPSTREAM_URL" "$WORKDIR/GeoLite2-City.mmdb"
                  fetch_and_verify "$GEOIP_ASN_UPSTREAM_URL"  "$WORKDIR/GeoLite2-ASN.mmdb"

                  # Atomic-ish replacement: upload to a versioned side-key
                  # first, then copy to the canonical key. If the final copy
                  # fails the previous canonical file is untouched.
                  STAMP=$(date -u +%Y%m%dT%H%M%SZ)

                  aws --endpoint-url "$FLUXER_S3_ENDPOINT" s3 cp \
                      "$WORKDIR/GeoLite2-City.mmdb" \
                      "s3://${GEOIP_BUCKET}/archive/GeoLite2-City-${STAMP}.mmdb"
                  aws --endpoint-url "$FLUXER_S3_ENDPOINT" s3 cp \
                      "$WORKDIR/GeoLite2-ASN.mmdb" \
                      "s3://${GEOIP_BUCKET}/archive/GeoLite2-ASN-${STAMP}.mmdb"

                  aws --endpoint-url "$FLUXER_S3_ENDPOINT" s3 cp \
                      "$WORKDIR/GeoLite2-City.mmdb" \
                      "s3://${GEOIP_BUCKET}/GeoLite2-City.mmdb"
                  aws --endpoint-url "$FLUXER_S3_ENDPOINT" s3 cp \
                      "$WORKDIR/GeoLite2-ASN.mmdb" \
                      "s3://${GEOIP_BUCKET}/GeoLite2-ASN.mmdb"

                  echo "geoip-sync complete: city=${STAMP} asn=${STAMP}"