OverviewHistoryStatsSecurity
npx skills add ...
Documentation
SKILL.md
npx skills add manutej/luxor-claude-marketplace --skill postgresql-database-engineering
Comprehensive PostgreSQL database engineering skill covering indexing strategies, query optimization, performance tuning, partitioning, replication, backup and recovery, high availability, and production database management. Master advanced PostgreSQL features including MVCC, VACUUM operations, connection pooling, monitoring, and scalability patterns.
npx skills add manutej/luxor-claude-marketplace --skill postgresql-database-engineering
A comprehensive skill for professional PostgreSQL database engineering, covering everything from query optimization and indexing strategies to high availability, replication, and production database management. This skill enables you to design, optimize, and maintain high-performance PostgreSQL databases at scale.
Use this skill when:
PostgreSQL uses a process-based architecture with several key components:
PostgreSQL's foundational concurrency mechanism:
Key Implications:
PostgreSQL supports four isolation levels:
Choosing Isolation:
PostgreSQL offers multiple index types for different use cases:
PostgreSQL's query planner determines execution strategies:
Planner Components:
Key Statistics:
n_distinct: Number of distinct values (for selectivity)correlation: Physical row ordering correlationmost_common_vals: MCV list for skewed distributionshistogram_bounds: Value distribution histogramUnderstanding EXPLAIN:
Table partitioning for managing large datasets:
Partition Pruning:
Partition-Wise Operations:
PostgreSQL replication options:
Synchronous vs Asynchronous:
Managing database connections efficiently:
Pooling Modes:
Critical maintenance operations:
Key configuration parameters:
Decision Matrix:
| Query Pattern | Index Type | Reason |
|---|---|---|
WHERE id = 5 | B-tree | Equality lookup |
WHERE created_at > '2024-01-01' | B-tree | Range query |
ORDER BY name | B-tree | Sorting support |
WHERE tags @> ARRAY['sql'] | GIN | Array containment |
WHERE data->>'status' = 'active' | GIN (jsonb_path_ops) | JSONB query |
WHERE to_tsvector(content) @@ query | GIN | Full-text search |
WHERE location <-> point(0,0) | GiST | Nearest neighbor |
WHERE timestamp BETWEEN ... (large table) | BRIN | Sequential time-series |
WHERE ip_address << '192.168.0.0/16' | GiST or SP-GiST | IP range query |
Multi-column indexes for complex queries:
Column Ordering Rules:
Example:
Index subset of rows:
Benefits:
Use Cases:
WHERE deleted_at IS NULLWHERE created_at > NOW() - INTERVAL '90 days'WHERE status IN ('pending', 'processing')Index computed values:
Examples:
Include non-key columns for index-only scans:
Benefit: Query satisfied entirely from index, no table lookup
Monitoring Index Usage:
Detecting Bloat:
Understanding query execution:
Key Metrics:
Problem: One query per row in a loop Solution: JOIN or batch queries
Problem: Fetches unnecessary columns Solution: Select only needed columns
Problem: Index not used due to type mismatch Solution: Ensure query types match column types
Problem: WHERE UPPER(email) = 'USER@EXAMPLE.COM'
Solution: Use expression index or compare correctly
Problem: WHERE status = 'A' OR status = 'B'
Solution: Use IN: WHERE status IN ('A', 'B')
Join Types:
Nested Loop
Hash Join
Merge Join
Join Order Matters:
SET join_collapse_limitTechniques:
Materialized Views:
Levels:
Time-series example:
Automated partition management:
Dropping old partitions:
Primary server configuration (postgresql.conf):
Create replication user:
pg_hba.conf on primary:
Standby server setup:
Standby configuration (created by -R flag):
On primary:
On standby:
Promoting standby to primary:
Controlled switchover:
On publisher (source):
On subscriber (destination):
pg_basebackup:
Continuous archiving (WAL archiving):
pg_dump:
pg_restore:
Setup:
Recovery:
3-2-1 Rule:
Backup Schedule:
Testing Backups:
Database Health:
Query Performance:
System Resources:
Connection stats:
Cache hit ratio:
Table bloat:
Long-running queries:
Lock monitoring:
Installation:
Configuration (postgresql.conf):
Top queries by total time:
Top queries by average time:
Normalization:
Data Types:
Constraints:
Zero-Downtime Migrations:
Add new column
Backfill data (in batches)
Add NOT NULL constraint
Index Creation:
CREATE INDEX CONCURRENTLY in productionpg_stat_progress_create_indexLarge Table Modifications:
pg_repack for table rewritesAuthentication:
Authorization:
Network Security:
Audit Logging:
Daily:
Weekly:
Monthly:
Quarterly:
Configuration:
Forcing parallel execution:
When parallelism helps:
Stored procedures:
Functions with proper error handling:
Access external data sources:
Indexing JSONB:
Efficient JSONB queries:
Basic setup:
Search queries:
Problem: Slow Queries
ANALYZE table_nameProblem: High CPU Usage
Problem: Connection Exhaustion
pg_stat_activityProblem: Autovacuum Not Keeping Up
Problem: Replication Lag
Problem: Transaction ID Wraparound
Find missing indexes on foreign keys:
Identify blocking queries:
Skill Version: 1.0.0 Last Updated: October 2025 Skill Category: Database Engineering, Performance Optimization, Data Architecture Compatible With: PostgreSQL 12+, 13, 14, 15, 16 Prerequisites: SQL knowledge, basic database concepts, Linux command line
max_connections: Maximum client connections
connection_limit: Per-database/user limits-- Query: WHERE status = 'active' AND created_at > '2024-01-01' ORDER BY created_at
-- Optimal index: (status, created_at)
CREATE INDEX idx_users_status_created ON users(status, created_at);-- Case-insensitive search
CREATE INDEX idx_users_email_lower ON users(LOWER(email));
-- Date truncation
CREATE INDEX idx_events_date ON events(DATE(created_at));
-- JSONB field
CREATE INDEX idx_data_status ON documents((data->>'status'));CREATE INDEX idx_users_email_include
ON users(email)
INCLUDE (first_name, last_name, created_at);-- Unused indexes
SELECT schemaname, tablename, indexname, idx_scan
FROM pg_stat_user_indexes
WHERE idx_scan = 0
ORDER BY pg_relation_size(indexrelid) DESC;-- Index bloat estimation
SELECT schemaname, tablename, indexname,
pg_size_pretty(pg_relation_size(indexrelid)) as index_size,
idx_scan, idx_tup_read, idx_tup_fetch
FROM pg_stat_user_indexes
ORDER BY pg_relation_size(indexrelid) DESC;-- Basic EXPLAIN
EXPLAIN SELECT * FROM users WHERE email = 'user@example.com';
-- EXPLAIN ANALYZE (actually runs query)
EXPLAIN ANALYZE SELECT * FROM users WHERE created_at > '2024-01-01';
-- Detailed output
EXPLAIN (ANALYZE, BUFFERS, VERBOSE)
SELECT u.*, o.total
FROM users u
JOIN orders o ON u.id = o.user_id
WHERE u.created_at > '2024-01-01';-- Create partitioned table
CREATE TABLE events (
id BIGSERIAL,
event_type TEXT NOT NULL,
user_id INTEGER NOT NULL,
data JSONB,
created_at TIMESTAMP NOT NULL
) PARTITION BY RANGE (created_at);
-- Create partitions
CREATE TABLE events_2024_01 PARTITION OF events
FOR VALUES FROM ('2024-01-01') TO ('2024-02-01');
CREATE TABLE events_2024_02 PARTITION OF events
FOR VALUES FROM ('2024-02-01') TO ('2024-03-01');
-- Default partition for data outside ranges
CREATE TABLE events_default PARTITION OF events DEFAULT;
-- Indexes on partitions
CREATE INDEX idx_events_2024_01_user ON events_2024_01(user_id);
CREATE INDEX idx_events_2024_02_user ON events_2024_02(user_id);-- Function to create monthly partitions
CREATE OR REPLACE FUNCTION create_monthly_partition(
base_table TEXT,
partition_date DATE
) RETURNS VOID AS $$
DECLARE
partition_name TEXT;
start_date DATE;
end_date DATE;
BEGIN
partition_name := base_table || '_' || TO_CHAR(partition_date, 'YYYY_MM');
start_date := DATE_TRUNC('month', partition_date);
end_date := start_date + INTERVAL '1 month';
EXECUTE format(
'CREATE TABLE IF NOT EXISTS %I PARTITION OF %I
FOR VALUES FROM (%L) TO (%L)',
partition_name, base_table, start_date, end_date
);
-- Create indexes
EXECUTE format(
'CREATE INDEX IF NOT EXISTS %I ON %I(user_id)',
'idx_' || partition_name || '_user', partition_name
);
END;
$$ LANGUAGE plpgsql;-- Detach partition (fast, non-blocking)
ALTER TABLE events DETACH PARTITION events_2023_01;
-- Drop detached partition
DROP TABLE events_2023_01;
-- Or archive before dropping
CREATE TABLE archive.events_2023_01 AS SELECT * FROM events_2023_01;
DROP TABLE events_2023_01;# Replication settings
wal_level = replica
max_wal_senders = 10
max_replication_slots = 10
hot_standby = on
synchronous_commit = on # or off for async
synchronous_standby_names = 'standby1,standby2' # for sync replicationCREATE USER replicator WITH REPLICATION ENCRYPTED PASSWORD 'secure_password';# Allow replication connections
host replication replicator standby_ip/32 md5# Stop standby PostgreSQL
systemctl stop postgresql
# Remove old data directory
rm -rf /var/lib/postgresql/14/main
# Base backup from primary
pg_basebackup -h primary_host -D /var/lib/postgresql/14/main \
-U replicator -P -v -R -X stream -C -S standby1
# Start standby
systemctl start postgresql# standby.signal file created automatically
# postgresql.auto.conf contains:
primary_conninfo = 'host=primary_host port=5432 user=replicator password=secure_password'
primary_slot_name = 'standby1'-- Check replication status
SELECT client_addr, state, sync_state, replay_lag
FROM pg_stat_replication;
-- Check replication slots
SELECT slot_name, active, restart_lsn, confirmed_flush_lsn
FROM pg_replication_slots;-- Check replication lag
SELECT now() - pg_last_xact_replay_timestamp() AS replication_lag;
-- Check recovery status
SELECT pg_is_in_recovery();# Trigger failover
pg_ctl promote -D /var/lib/postgresql/14/main
# Or using SQL
SELECT pg_promote();# 1. Stop writes on primary
# 2. Wait for standby to catch up
# 3. Promote standby
# 4. Reconfigure old primary as new standby-- Create publication
CREATE PUBLICATION my_publication FOR TABLE users, orders;
-- Or all tables
CREATE PUBLICATION all_tables FOR ALL TABLES;-- Create subscription
CREATE SUBSCRIPTION my_subscription
CONNECTION 'host=publisher_host dbname=mydb user=replicator password=pass'
PUBLICATION my_publication;
-- Monitor subscription
SELECT * FROM pg_stat_subscription;# Full physical backup
pg_basebackup -h localhost -U postgres -D /backup/base \
-F tar -z -P -v
# With WAL files for point-in-time recovery
pg_basebackup -h localhost -U postgres -D /backup/base \
-X stream -F tar -z -P# postgresql.conf
wal_level = replica
archive_mode = on
archive_command = 'cp %p /archive/wal/%f'# Single database
pg_dump -h localhost -U postgres -F c -b -v -f mydb.dump mydb
# All databases
pg_dumpall -h localhost -U postgres -f all_databases.sql
# Specific tables
pg_dump -h localhost -U postgres -t users -t orders -F c -f tables.dump mydb
# Schema only
pg_dump -h localhost -U postgres --schema-only -F c -f schema.dump mydb# Restore database
pg_restore -h localhost -U postgres -d mydb -v mydb.dump
# Parallel restore
pg_restore -h localhost -U postgres -d mydb -j 4 -v mydb.dump
# Restore specific tables
pg_restore -h localhost -U postgres -d mydb -t users -v mydb.dump# 1. Restore base backup
tar -xzf base.tar.gz -C /var/lib/postgresql/14/main
# 2. Create recovery.signal file
touch /var/lib/postgresql/14/main/recovery.signal
# 3. Configure recovery target (postgresql.conf or postgresql.auto.conf)
restore_command = 'cp /archive/wal/%f %p'
recovery_target_time = '2024-01-15 14:30:00'
# Or: recovery_target_name = 'before_disaster'
# Or: recovery_target_lsn = '0/3000000'
# 4. Start PostgreSQL
systemctl start postgresqlSELECT count(*) as total_connections,
count(*) FILTER (WHERE state = 'active') as active,
count(*) FILTER (WHERE state = 'idle') as idle,
count(*) FILTER (WHERE state = 'idle in transaction') as idle_in_transaction
FROM pg_stat_activity;SELECT sum(heap_blks_read) as heap_read,
sum(heap_blks_hit) as heap_hit,
sum(heap_blks_hit) / (sum(heap_blks_hit) + sum(heap_blks_read)) AS ratio
FROM pg_statio_user_tables;SELECT schemaname, tablename,
pg_size_pretty(pg_total_relation_size(schemaname||'.'||tablename)) AS size,
n_dead_tup,
n_live_tup,
round(n_dead_tup * 100.0 / NULLIF(n_live_tup + n_dead_tup, 0), 2) AS dead_ratio
FROM pg_stat_user_tables
WHERE n_dead_tup > 1000
ORDER BY n_dead_tup DESC;SELECT pid, now() - query_start AS duration, state, query
FROM pg_stat_activity
WHERE state != 'idle'
AND query NOT LIKE '%pg_stat_activity%'
ORDER BY duration DESC;SELECT blocked_locks.pid AS blocked_pid,
blocked_activity.usename AS blocked_user,
blocking_locks.pid AS blocking_pid,
blocking_activity.usename AS blocking_user,
blocked_activity.query AS blocked_statement,
blocking_activity.query AS blocking_statement
FROM pg_catalog.pg_locks blocked_locks
JOIN pg_catalog.pg_stat_activity blocked_activity ON blocked_activity.pid = blocked_locks.pid
JOIN pg_catalog.pg_locks blocking_locks ON blocking_locks.locktype = blocked_locks.locktype
JOIN pg_catalog.pg_stat_activity blocking_activity ON blocking_activity.pid = blocking_locks.pid
WHERE NOT blocked_locks.granted
AND blocking_locks.granted;CREATE EXTENSION pg_stat_statements;shared_preload_libraries = 'pg_stat_statements'
pg_stat_statements.track = all
pg_stat_statements.max = 10000SELECT query,
calls,
total_exec_time,
mean_exec_time,
max_exec_time,
rows
FROM pg_stat_statements
ORDER BY total_exec_time DESC
LIMIT 20;SELECT query,
calls,
mean_exec_time,
total_exec_time
FROM pg_stat_statements
WHERE calls > 100
ORDER BY mean_exec_time DESC
LIMIT 20;