aurora-postgresql.sh

#!/bin/bash

# 設定
REGION="ap-northeast-1"  # 必要に応じて変更してください

echo "=== Aurora PostgreSQL Cluster Discovery ==="
echo "Region: $REGION"
echo ""

# Aurora PostgreSQL クラスターを取得
echo "Discovering Aurora PostgreSQL clusters..."
AURORA_CLUSTERS=$(aws rds describe-db-clusters \
  --region $REGION \
  --query 'DBClusters[?Engine==`aurora-postgresql`].{ClusterIdentifier:DBClusterIdentifier,Status:Status,Engine:Engine}' \
  --output table)

if [ $? -eq 0 ]; then
    echo "Available Aurora PostgreSQL Clusters:"
    echo "$AURORA_CLUSTERS"
    echo ""

    # クラスター一覧を配列で取得
    CLUSTER_ARRAY=($(aws rds describe-db-clusters \
      --region $REGION \
      --query 'DBClusters[?Engine==`aurora-postgresql`].DBClusterIdentifier' \
      --output text))

    if [ ${#CLUSTER_ARRAY[@]} -gt 0 ]; then
        # 複数クラスターがある場合は選択画面を表示
        if [ ${#CLUSTER_ARRAY[@]} -gt 1 ]; then
            echo "Multiple clusters found. Please select one:"
            for i in "${!CLUSTER_ARRAY[@]}"; do
                echo "$((i+1))) ${CLUSTER_ARRAY[$i]}"
            done
            echo ""
            echo "Enter selection (1-${#CLUSTER_ARRAY[@]}) or press Enter for #1:"
            read selection

            if [ -z "$selection" ]; then
                selection=1
            fi

            # 選択の検証
            if [[ "$selection" =~ ^[0-9]+$ ]] && [ "$selection" -ge 1 ] && [ "$selection" -le ${#CLUSTER_ARRAY[@]} ]; then
                DB_CLUSTER_IDENTIFIER=${CLUSTER_ARRAY[$((selection-1))]}
            else
                echo "ERROR: Invalid selection"
                exit 1
            fi
        else
            # クラスターが1つの場合は自動選択
            DB_CLUSTER_IDENTIFIER=${CLUSTER_ARRAY[0]}
        fi

        echo "Selected cluster: $DB_CLUSTER_IDENTIFIER"

        # クラスターのライターインスタンスを取得
        echo "Finding writer instance for cluster: $DB_CLUSTER_IDENTIFIER"

        DB_WRITER_INSTANCE=$(aws rds describe-db-clusters \
          --db-cluster-identifier $DB_CLUSTER_IDENTIFIER \
          --region $REGION \
          --query 'DBClusters[0].DBClusterMembers[?IsClusterWriter==`true`].DBInstanceIdentifier' \
          --output text)

        if [ -n "$DB_WRITER_INSTANCE" ] && [ "$DB_WRITER_INSTANCE" != "None" ]; then
            echo "Writer instance: $DB_WRITER_INSTANCE"

            # クラスター詳細情報を取得
            echo ""
            echo "=== Cluster Details ==="
            aws rds describe-db-clusters \
              --db-cluster-identifier $DB_CLUSTER_IDENTIFIER \
              --region $REGION \
              --query 'DBClusters[0].{
                ClusterIdentifier:DBClusterIdentifier,
                Engine:Engine,
                EngineVersion:EngineVersion,
                Status:Status,
                MultiAZ:MultiAZ,
                Members:DBClusterMembers[].{
                  Instance:DBInstanceIdentifier,
                  IsWriter:IsClusterWriter,
                  Status:DBInstanceStatus
                }
              }' \
              --output table
        else
            echo "ERROR: No writer instance found for cluster $DB_CLUSTER_IDENTIFIER"
            exit 1
        fi
    else
        echo "ERROR: No Aurora PostgreSQL clusters found in region $REGION"
        exit 1
    fi
else
    echo "ERROR: Failed to retrieve cluster information"
    exit 1
fi

echo ""
echo "=== Starting CDC Throughput Analysis ==="

# 時刻設定
START_TIME=$(date -u -v-7d +%Y-%m-%dT%H:%M:%S)
END_TIME=$(date -u +%Y-%m-%dT%H:%M:%S)

echo "Cluster: $DB_CLUSTER_IDENTIFIER"
echo "Writer Instance: $DB_WRITER_INSTANCE"
echo "Period: $START_TIME to $END_TIME"
echo ""

# メトリクス取得関数
get_metrics() {
    local identifier=$1
    local dimension_name=$2
    local metric_name=$3
    local stat_type=$4
    local output_file=$5

    echo "Retrieving $metric_name ($stat_type) for $identifier..."
    aws cloudwatch get-metric-statistics \
      --namespace AWS/RDS \
      --metric-name $metric_name \
      --dimensions Name=$dimension_name,Value=$identifier \
      --start-time "$START_TIME" \
      --end-time "$END_TIME" \
      --period 3600 \
      --statistics $stat_type \
      --region $REGION \
      --output json > "$output_file"

    if [ $? -ne 0 ]; then
        echo "  Warning: Failed to retrieve $metric_name for $identifier"
        return 1
    fi
    return 0
}

# メトリクス取得実行
echo "=== Retrieving Metrics ==="

# クラスターレベルのメトリクス
get_metrics "$DB_CLUSTER_IDENTIFIER" "DBClusterIdentifier" "WriteThroughput" "Maximum" "cluster_write_max.json"
get_metrics "$DB_CLUSTER_IDENTIFIER" "DBClusterIdentifier" "WriteThroughput" "Average" "cluster_write_avg.json"
get_metrics "$DB_CLUSTER_IDENTIFIER" "DBClusterIdentifier" "WriteIOPS" "Maximum" "cluster_iops_max.json"
get_metrics "$DB_CLUSTER_IDENTIFIER" "DBClusterIdentifier" "WriteIOPS" "Average" "cluster_iops_avg.json"

# ライターインスタンスのメトリクス
get_metrics "$DB_WRITER_INSTANCE" "DBInstanceIdentifier" "WriteThroughput" "Maximum" "writer_write_max.json"
get_metrics "$DB_WRITER_INSTANCE" "DBInstanceIdentifier" "WriteThroughput" "Average" "writer_write_avg.json"
get_metrics "$DB_WRITER_INSTANCE" "DBInstanceIdentifier" "WriteIOPS" "Maximum" "writer_iops_max.json"
get_metrics "$DB_WRITER_INSTANCE" "DBInstanceIdentifier" "WriteIOPS" "Average" "writer_iops_avg.json"

# Aurora PostgreSQL 固有メトリクス（WALディスク使用量）
get_metrics "$DB_WRITER_INSTANCE" "DBInstanceIdentifier" "TransactionLogsDiskUsage" "Maximum" "writer_wal_disk.json"

echo ""

# メトリクス解析関数
extract_value() {
    local file=$1
    local stat_type=$2

    if [ ! -f "$file" ]; then
        echo "0"
        return
    fi

    # ファイルサイズチェック
    if [ ! -s "$file" ]; then
        echo "0"
        return
    fi

    if command -v jq &> /dev/null; then
        if [ "$stat_type" = "Maximum" ]; then
            local value=$(cat "$file" | jq -r '.Datapoints[].Maximum' 2>/dev/null | grep -v null | sort -n | tail -1)
        else
            local value=$(cat "$file" | jq -r '.Datapoints[].Average' 2>/dev/null | grep -v null | awk '{sum+=$1; count++} END {if(count>0) print sum/count; else print 0}')
        fi
    else
        if [ "$stat_type" = "Maximum" ]; then
            local value=$(grep -o "\"Maximum\": [0-9.]*" "$file" | cut -d' ' -f2 | sort -n | tail -1)
        else
            local value=$(grep -o "\"Average\": [0-9.]*" "$file" | cut -d' ' -f2 | awk '{sum+=$1; count++} END {if(count>0) print sum/count; else print 0}')
        fi
    fi

    echo "${value:-0}"
}

# データ抽出
echo "=== Metric Analysis ==="
echo ""

# クラスターメトリクス
CLUSTER_WRITE_MAX=$(extract_value "cluster_write_max.json" "Maximum")
CLUSTER_WRITE_AVG=$(extract_value "cluster_write_avg.json" "Average")
CLUSTER_IOPS_MAX=$(extract_value "cluster_iops_max.json" "Maximum")
CLUSTER_IOPS_AVG=$(extract_value "cluster_iops_avg.json" "Average")

# ライターメトリクス
WRITER_WRITE_MAX=$(extract_value "writer_write_max.json" "Maximum")
WRITER_WRITE_AVG=$(extract_value "writer_write_avg.json" "Average")
WRITER_IOPS_MAX=$(extract_value "writer_iops_max.json" "Maximum")
WRITER_IOPS_AVG=$(extract_value "writer_iops_avg.json" "Average")
WRITER_WAL_DISK=$(extract_value "writer_wal_disk.json" "Maximum")

# 値の検証とフォールバック
validate_value() {
    local value=$1
    if [ -z "$value" ] || [ "$value" = "null" ] || [ "$value" = "" ]; then
        echo "0"
    else
        echo "$value"
    fi
}

CLUSTER_WRITE_MAX=$(validate_value "$CLUSTER_WRITE_MAX")
CLUSTER_WRITE_AVG=$(validate_value "$CLUSTER_WRITE_AVG")
WRITER_WRITE_MAX=$(validate_value "$WRITER_WRITE_MAX")
WRITER_WRITE_AVG=$(validate_value "$WRITER_WRITE_AVG")

# 結果表示
echo "Cluster-Level Metrics:"
echo "  WriteThroughput - Maximum: $CLUSTER_WRITE_MAX bytes/sec"
echo "  WriteThroughput - Average: $CLUSTER_WRITE_AVG bytes/sec"
echo "  WriteIOPS - Maximum: $CLUSTER_IOPS_MAX ops/sec"
echo "  WriteIOPS - Average: $CLUSTER_IOPS_AVG ops/sec"
echo ""

echo "Writer Instance Metrics:"
echo "  WriteThroughput - Maximum: $WRITER_WRITE_MAX bytes/sec"
echo "  WriteThroughput - Average: $WRITER_WRITE_AVG bytes/sec"
echo "  WriteIOPS - Maximum: $WRITER_IOPS_MAX ops/sec"
echo "  WriteIOPS - Average: $WRITER_IOPS_AVG ops/sec"
echo "  TransactionLogsDiskUsage - Maximum: $WRITER_WAL_DISK bytes"
echo ""

# CDC 計算関数
calculate_cdc() {
    local throughput=$1
    local label=$2

    if [ -n "$throughput" ] && [ "$throughput" != "0" ] && [ "$throughput" != "" ]; then
        # Aurora PostgreSQL CDC 係数: 0.8 (WAL overhead consideración)
        local cdc_bytes=$(echo "$throughput" | awk '{printf "%.0f", $1 * 0.8}')
        local cdc_mbps=$(echo "$cdc_bytes" | awk '{printf "%.3f", $1 * 8 / 1000000}')
        printf "  %-20s: %s bytes/sec (%.3f Mbps)\n" "$label" "$cdc_bytes" "$cdc_mbps"
    else
        printf "  %-20s: No data available\n" "$label"
    fi
}

# CDC 推定値
echo "=== CDC Throughput Estimates ==="
echo ""
echo "Based on Cluster WriteThroughput:"
calculate_cdc "$CLUSTER_WRITE_MAX" "Peak CDC"
calculate_cdc "$CLUSTER_WRITE_AVG" "Average CDC"

echo ""
echo "Based on Writer Instance WriteThroughput:"
calculate_cdc "$WRITER_WRITE_MAX" "Peak CDC"
calculate_cdc "$WRITER_WRITE_AVG" "Average CDC"

echo ""
echo "=== Data Point Validation ==="
for file in cluster_write_max.json writer_write_max.json; do
    if [ -f "$file" ]; then
        if command -v jq &> /dev/null; then
            datapoints=$(cat "$file" | jq '.Datapoints | length' 2>/dev/null)
        else
            datapoints=$(grep -c '"Timestamp"' "$file" 2>/dev/null)
        fi
        echo "$file: ${datapoints:-0} data points"
    else
        echo "$file: File not found"
    fi
done

echo ""
echo "=== PostgreSQL-Specific Recommendations ==="
echo "1. Use cluster-level metrics for overall CDC capacity planning"
echo "2. Peak values should guide maximum Confluent Cloud throughput settings"
echo "3. Average values should guide standard operational capacity"
echo "4. Consider 20-50% safety margin for production deployment"
echo "5. Monitor TransactionLogsDiskUsage (WAL) for replication slot lag"
echo "6. Ensure logical replication parameters are properly configured:"
echo "   - rds.logical_replication = 1"
echo "   - rds.logical_wal_cache (appropriate size for workload)"
echo "7. PostgreSQL CDC coefficient (0.8) accounts for WAL overhead"

# クリーンアップ
echo ""
echo "Cleaning up temporary files..."
rm -f cluster_*.json writer_*.json

echo "Analysis complete."