# Velero Helm values — the AUTHORITATIVE values live inline in helmrelease.yaml (this file is a
# doc pointer, kept for parity with the other components' layout).
#
# LIVE wiring (see helmrelease.yaml):
#   - provider aws + velero-plugin-for-aws (init container) → S3 ${S3_BACKUP_ENDPOINT}, path-style,
#     at s3://${VELERO_BACKUP_BUCKET}/${VELERO_BACKUP_PREFIX} in region ${S3_BACKUP_REGION}. Four
#     cluster-vars knobs: dev = the on-Proxmox MinIO (http://192.168.1.9:9000, bucket `velero`, no
#     prefix, us-east-1); prod = Cloudflare R2 (bucket `k3s-prod-backup`, prefix `velero`, `auto`),
#     sharing one bucket with CNPG/barman under a disjoint prefix.
#   - config.checksumAlgorithm: "" — MANDATORY for R2, which rejects the AWS SDK's
#     x-amz-checksum-* headers with HTTP 400 InvalidArgument. Harmless against MinIO.
#   - velero-plugin-for-aws PINNED TO v1.14.3-rc.2 — also MANDATORY for R2. v1.14.2 sends an empty
#     `x-amz-tagging` header on every PutObject, which R2 answers with HTTP 501 NotImplemented, so
#     every Backup lands its kopia volume data and then FAILS writing velero-backup.json. Upstream
#     PR #309 makes the field conditional; rc.2 is the earliest build carrying it. Move to v1.14.3
#     when it ships GA. Measured on prod 2026-09-12.
#   - resourcePolicy on the FSB schedule -> ConfigMap velero-volume-policies (volume-policies.yaml):
#     skips volumes whose source is NFS, so the 8.6 TiB external bulk tier is not uploaded nightly
#     while the same namespaces' real Longhorn PVCs still are. Proven on prod 2026-09-12.
#   - deployNodeAgent:true + uploaderType:kopia → File System Backup of PVC data (no CSI data-mover).
#   - snapshotsEnabled:false → no VolumeSnapshotLocation.
#   - credentials.existingSecret: velero-cloud-credentials (SOPS `cloud` INI key, ns velero).
#   - two daily Schedules (${VELERO_BACKUP_SCHEDULE} / ttl ${VELERO_RETENTION}): the whole cluster
#     via FSB, with monitoring+unifi backed up RESOURCES-ONLY (their volume data is reconstructable).
# DR drill (tested restore) remains a Phase-4 acceptance gate (R-STATE-03) — the PM runs it post-deploy.