prepare_dr_restore.sh

#!/usr/bin/env bash
# QuestDB DR prep. SOURCE this ( . ~/prepare_dr_restore.sh ) so QDB_REPLICATION_ROLE
# lands in your shell. Runs in a function with `return`, so it never leaks shell
# options (set -u/-e) or drops your SSH session.

# ===== per-host config: the ONLY lines that differ between nodes =====
DR_ROOT=/data2/disaster_qdb_root       # replica2 uses /data/disaster_qdb_root
QDB_ROLE=primary                       # replicas use: replica
WAL_PURGE_S3="${QDB_WAL_PURGE_S3:-}"   # primary: export QDB_WAL_PURGE_S3=s3://<bucket>/<replication-store>/ ; replicas leave unset
# =====================================================================

_dr_prepare() {
  local RESTORE_MARKER="$DR_ROOT/_backup_restore"

  if [ -z "${QDB_BACKUP_OBJECT_STORE:-}" ]; then
    echo "ERROR: QDB_BACKUP_OBJECT_STORE is not set; aborting." >&2; return 1
  fi

  local BUCKET ROOT REGION PREFIX LATEST_KEY REST LATEST_FOLDER
  BUCKET=$(printf '%s' "$QDB_BACKUP_OBJECT_STORE" | sed -n 's/.*bucket=\([^;]*\).*/\1/p')
  ROOT=$(printf   '%s' "$QDB_BACKUP_OBJECT_STORE" | sed -n 's/.*root=\([^;]*\).*/\1/p')
  REGION=$(printf '%s' "$QDB_BACKUP_OBJECT_STORE" | sed -n 's/.*region=\([^;]*\).*/\1/p')
  if [ -z "$BUCKET" ] || [ -z "$ROOT" ]; then
    echo "ERROR: could not parse bucket/root from QDB_BACKUP_OBJECT_STORE; aborting." >&2; return 1
  fi

  PREFIX="$ROOT/backup/"
  LATEST_KEY=$(aws s3api list-objects-v2 \
    --bucket "$BUCKET" --prefix "$PREFIX" ${REGION:+--region "$REGION"} \
    --query 'sort_by(Contents, &LastModified)[-1].Key' --output text) \
    || { echo "ERROR: could not list s3://$BUCKET/$PREFIX; aborting." >&2; return 1; }

  REST=${LATEST_KEY#"$PREFIX"}; LATEST_FOLDER=${REST%%/*}
  if [ -z "$LATEST_FOLDER" ] || [ "$LATEST_FOLDER" = "None" ]; then
    echo "ERROR: no backup found under s3://$BUCKET/$PREFIX; aborting." >&2; return 1
  fi

  # all checks passed — reset DR root, (primary) purge replication store, set role, write marker
  rm -rf "$DR_ROOT"
  mkdir -p "$DR_ROOT" || { echo "ERROR: mkdir $DR_ROOT failed; aborting." >&2; return 1; }
  if [ -n "$WAL_PURGE_S3" ]; then
    aws s3 rm "$WAL_PURGE_S3" --recursive ${REGION:+--region "$REGION"} \
      || { echo "ERROR: failed to purge $WAL_PURGE_S3; aborting." >&2; return 1; }
  fi
  export QDB_REPLICATION_ROLE="$QDB_ROLE"

  touch "$RESTORE_MARKER"
  echo "backup.instance.name=$LATEST_FOLDER" >> "$RESTORE_MARKER"
  echo "restore marker set -> $LATEST_FOLDER  (role=$QDB_ROLE, DR_ROOT=$DR_ROOT)"
}

_dr_prepare