Initial commit
This commit is contained in:
1199
skills/wp-performance-review/references/anti-patterns.md
Normal file
1199
skills/wp-performance-review/references/anti-patterns.md
Normal file
File diff suppressed because it is too large
Load Diff
447
skills/wp-performance-review/references/caching-guide.md
Normal file
447
skills/wp-performance-review/references/caching-guide.md
Normal file
@@ -0,0 +1,447 @@
|
||||
# WordPress Caching Strategy Guide
|
||||
|
||||
Comprehensive guide to caching strategies for high-performance WordPress applications.
|
||||
|
||||
## Caching Review Checklist
|
||||
|
||||
When reviewing caching implementation:
|
||||
|
||||
**Page Cache Compatibility**
|
||||
- [ ] No `session_start()` on frontend
|
||||
- [ ] No `setcookie()` on public pages (unless necessary)
|
||||
- [ ] No `$_SESSION` usage on cacheable pages
|
||||
- [ ] POST used only for writes, GET for reads
|
||||
|
||||
**Object Cache Usage**
|
||||
- [ ] Expensive queries wrapped with `wp_cache_get`/`wp_cache_set`
|
||||
- [ ] Cache keys include relevant variables (locale, user role if needed)
|
||||
- [ ] TTL set appropriately (not too long, not too short)
|
||||
- [ ] `wp_cache_get_multiple` for batch lookups
|
||||
- [ ] Cache invalidation on relevant `save_post` hooks
|
||||
|
||||
**Race Condition Prevention**
|
||||
- [ ] `wp_cache_add` used for locking when needed
|
||||
- [ ] Stale-while-revalidate for high-traffic cached items
|
||||
- [ ] Cache pre-warming via cron for critical data
|
||||
|
||||
## Caching Layers Overview
|
||||
|
||||
```
|
||||
User Request
|
||||
↓
|
||||
┌─────────────────┐
|
||||
│ CDN / Edge │ ← Full page cache (HTML)
|
||||
│ Cache │ Static assets (JS, CSS, images)
|
||||
└────────┬────────┘
|
||||
↓
|
||||
┌─────────────────┐
|
||||
│ Page Cache │ ← Full page cache (server-level)
|
||||
│ (Varnish, etc) │ Bypassed by: cookies, POST, query vars
|
||||
└────────┬────────┘
|
||||
↓
|
||||
┌─────────────────┐
|
||||
│ Object Cache │ ← Database query results
|
||||
│ (Redis/Memcached)│ Transients, options, computed data
|
||||
└────────┬────────┘
|
||||
↓
|
||||
┌─────────────────┐
|
||||
│ Database │ ← MySQL query cache (if enabled)
|
||||
│ (MySQL) │ InnoDB buffer pool
|
||||
└─────────────────┘
|
||||
```
|
||||
|
||||
## Page Cache
|
||||
|
||||
### Cache Headers
|
||||
|
||||
```php
|
||||
// Set cache-friendly headers
|
||||
function set_cache_headers() {
|
||||
if (!is_user_logged_in() && !is_admin()) {
|
||||
header('Cache-Control: public, max-age=300, s-maxage=3600');
|
||||
header('Vary: Accept-Encoding');
|
||||
}
|
||||
}
|
||||
add_action('send_headers', 'set_cache_headers');
|
||||
|
||||
// Prevent caching for dynamic pages
|
||||
function prevent_page_cache() {
|
||||
if (is_user_specific_page()) {
|
||||
header('Cache-Control: private, no-cache, no-store');
|
||||
nocache_headers();
|
||||
}
|
||||
}
|
||||
```
|
||||
|
||||
### Cache Bypass Triggers (Avoid These)
|
||||
|
||||
```php
|
||||
// ❌ Starting PHP sessions bypasses cache
|
||||
session_start(); // Avoid on frontend
|
||||
|
||||
// ❌ Unique query parameters create cache misses
|
||||
// https://example.com/?utm_source=twitter&utm_campaign=123
|
||||
// Solution: Strip marketing params at CDN level
|
||||
|
||||
// ❌ POST requests always bypass cache
|
||||
// Use GET for read operations
|
||||
|
||||
// ❌ Setting cookies prevents caching
|
||||
setcookie('my_cookie', 'value'); // Use sparingly
|
||||
```
|
||||
|
||||
### TTL (Time To Live) Strategy
|
||||
|
||||
| Content Type | Recommended TTL |
|
||||
|--------------|-----------------|
|
||||
| Homepage | 5-15 minutes |
|
||||
| Archive pages | 15-60 minutes |
|
||||
| Single posts | 1-24 hours |
|
||||
| Static pages | 24+ hours |
|
||||
| Media files | 1 year (versioned) |
|
||||
|
||||
## Object Cache
|
||||
|
||||
### Basic Usage
|
||||
|
||||
```php
|
||||
// Store data
|
||||
wp_cache_set('my_key', $data, 'my_group', 3600); // 1 hour expiry
|
||||
|
||||
// Retrieve data
|
||||
$data = wp_cache_get('my_key', 'my_group');
|
||||
if (false === $data) {
|
||||
$data = expensive_computation();
|
||||
wp_cache_set('my_key', $data, 'my_group', 3600);
|
||||
}
|
||||
|
||||
// Delete data
|
||||
wp_cache_delete('my_key', 'my_group');
|
||||
|
||||
// Add only if doesn't exist (atomic)
|
||||
$added = wp_cache_add('my_key', $data, 'my_group', 3600);
|
||||
```
|
||||
|
||||
### Batch Operations (Efficient)
|
||||
|
||||
```php
|
||||
// ❌ BAD: Multiple round-trips
|
||||
foreach ($ids as $id) {
|
||||
$data[$id] = wp_cache_get("item_$id", 'items');
|
||||
}
|
||||
|
||||
// ✅ GOOD: Single round-trip
|
||||
$keys = array_map(fn($id) => "item_$id", $ids);
|
||||
$data = wp_cache_get_multiple($keys, 'items');
|
||||
```
|
||||
|
||||
### Cache Groups
|
||||
|
||||
```php
|
||||
// Use groups to organize and bulk-delete
|
||||
wp_cache_set('post_123', $data, 'my_plugin_posts');
|
||||
wp_cache_set('post_456', $data, 'my_plugin_posts');
|
||||
|
||||
// Clear entire group (if supported by backend)
|
||||
wp_cache_flush_group('my_plugin_posts'); // Redis supports this
|
||||
```
|
||||
|
||||
### Cache Key Versioning
|
||||
|
||||
```php
|
||||
// Version cache keys for easy invalidation
|
||||
function get_cache_key($base) {
|
||||
$version = wp_cache_get('cache_version', 'my_plugin') ?: 1;
|
||||
return "{$base}_v{$version}";
|
||||
}
|
||||
|
||||
// Invalidate all by incrementing version
|
||||
function invalidate_all_cache() {
|
||||
wp_cache_incr('cache_version', 1, 'my_plugin');
|
||||
}
|
||||
```
|
||||
|
||||
## Race Conditions
|
||||
|
||||
### Problem: Concurrent Cache Regeneration
|
||||
|
||||
When cache expires, multiple requests may simultaneously regenerate it.
|
||||
|
||||
```
|
||||
Request A ─┬─ Cache miss ─→ Start regeneration ───────→ Set cache
|
||||
│
|
||||
Request B ─┴─ Cache miss ─→ Start regeneration ───────→ Set cache (duplicate!)
|
||||
│
|
||||
Request C ─┴─ Cache miss ─→ Start regeneration ───────→ Set cache (duplicate!)
|
||||
```
|
||||
|
||||
### Solution: Locking Pattern
|
||||
|
||||
```php
|
||||
function get_expensive_data() {
|
||||
$cache_key = 'expensive_data';
|
||||
$lock_key = 'expensive_data_lock';
|
||||
|
||||
// Try to get cached data
|
||||
$data = wp_cache_get($cache_key);
|
||||
if (false !== $data) {
|
||||
return $data;
|
||||
}
|
||||
|
||||
// Try to acquire lock (atomic operation)
|
||||
$lock_acquired = wp_cache_add($lock_key, true, '', 30); // 30 second lock
|
||||
|
||||
if ($lock_acquired) {
|
||||
// We got the lock - regenerate cache
|
||||
$data = expensive_computation();
|
||||
wp_cache_set($cache_key, $data, '', 3600);
|
||||
wp_cache_delete($lock_key); // Release lock
|
||||
return $data;
|
||||
}
|
||||
|
||||
// Another process is regenerating - wait and retry
|
||||
usleep(100000); // 100ms
|
||||
return get_expensive_data(); // Retry (add max retries in production)
|
||||
}
|
||||
```
|
||||
|
||||
### Solution: Stale-While-Revalidate
|
||||
|
||||
```php
|
||||
function get_data_with_stale() {
|
||||
$cache_key = 'my_data';
|
||||
$stale_key = 'my_data_stale';
|
||||
|
||||
$data = wp_cache_get($cache_key);
|
||||
if (false !== $data) {
|
||||
return $data;
|
||||
}
|
||||
|
||||
// Try to get stale data while regenerating
|
||||
$stale_data = wp_cache_get($stale_key);
|
||||
|
||||
// Regenerate in background (non-blocking)
|
||||
if (false === $stale_data) {
|
||||
// No stale data - must wait
|
||||
$data = regenerate_data();
|
||||
wp_cache_set($cache_key, $data, '', 300);
|
||||
wp_cache_set($stale_key, $data, '', 3600); // Keep stale longer
|
||||
return $data;
|
||||
}
|
||||
|
||||
// Schedule background regeneration
|
||||
wp_schedule_single_event(time(), 'regenerate_my_data');
|
||||
|
||||
// Return stale data immediately
|
||||
return $stale_data;
|
||||
}
|
||||
```
|
||||
|
||||
## Cache Stampede Prevention
|
||||
|
||||
### Problem
|
||||
|
||||
Cache expires → Thousands of requests hit database simultaneously.
|
||||
|
||||
### Solution 1: Jitter (Randomized Expiry)
|
||||
|
||||
```php
|
||||
function set_cache_with_jitter($key, $data, $base_ttl) {
|
||||
// Add ±10% randomization to TTL
|
||||
$jitter = rand(-($base_ttl * 0.1), $base_ttl * 0.1);
|
||||
$ttl = $base_ttl + $jitter;
|
||||
wp_cache_set($key, $data, '', $ttl);
|
||||
}
|
||||
```
|
||||
|
||||
### Solution 2: Pre-warming via Cron
|
||||
|
||||
```php
|
||||
// Regenerate cache before it expires
|
||||
add_action('pre_warm_popular_caches', function() {
|
||||
$popular_queries = ['homepage_posts', 'featured_products', 'menu_items'];
|
||||
|
||||
foreach ($popular_queries as $query) {
|
||||
// Force regeneration
|
||||
wp_cache_delete($query);
|
||||
get_cached_query($query); // Regenerates and caches
|
||||
}
|
||||
});
|
||||
|
||||
// Schedule to run before typical expiry
|
||||
if (!wp_next_scheduled('pre_warm_popular_caches')) {
|
||||
wp_schedule_event(time(), 'hourly', 'pre_warm_popular_caches');
|
||||
}
|
||||
```
|
||||
|
||||
### Solution 3: Early Expiry Check
|
||||
|
||||
```php
|
||||
function get_with_early_expiry($key, $ttl, $regenerate_callback) {
|
||||
$data = wp_cache_get($key);
|
||||
|
||||
if (false !== $data) {
|
||||
// Check if we should pre-regenerate (last 10% of TTL)
|
||||
$meta = wp_cache_get("{$key}_meta");
|
||||
if ($meta && (time() - $meta['created']) > ($ttl * 0.9)) {
|
||||
// Trigger background regeneration
|
||||
wp_schedule_single_event(time(), 'regenerate_cache', [$key]);
|
||||
}
|
||||
return $data;
|
||||
}
|
||||
|
||||
// Cache miss - regenerate
|
||||
$data = call_user_func($regenerate_callback);
|
||||
wp_cache_set($key, $data, '', $ttl);
|
||||
wp_cache_set("{$key}_meta", ['created' => time()], '', $ttl);
|
||||
|
||||
return $data;
|
||||
}
|
||||
```
|
||||
|
||||
## Transients
|
||||
|
||||
### When to Use Transients
|
||||
|
||||
- Data with known expiration
|
||||
- External API responses
|
||||
- Computed data that's expensive to regenerate
|
||||
|
||||
### Transient Best Practices
|
||||
|
||||
```php
|
||||
// ✅ GOOD: Named clearly, reasonable TTL
|
||||
set_transient('weather_data_seattle', $data, HOUR_IN_SECONDS);
|
||||
|
||||
// ❌ BAD: Dynamic keys create transient bloat
|
||||
set_transient("user_{$user_id}_preferences", $data, DAY_IN_SECONDS);
|
||||
// Better: Use user meta or object cache
|
||||
```
|
||||
|
||||
### Transient Storage Warning
|
||||
|
||||
Without persistent object cache, transients are stored in wp_options table:
|
||||
|
||||
```php
|
||||
// Check if object cache is available
|
||||
if (wp_using_ext_object_cache()) {
|
||||
// Transients use object cache (good)
|
||||
set_transient('my_data', $data, HOUR_IN_SECONDS);
|
||||
} else {
|
||||
// Transients go to database (potentially bad)
|
||||
// Consider: file cache, skip caching, or warn admin
|
||||
}
|
||||
```
|
||||
|
||||
## Partial Output Caching
|
||||
|
||||
Cache rendered HTML fragments for repeated use:
|
||||
|
||||
```php
|
||||
function get_cached_sidebar() {
|
||||
$cache_key = 'sidebar_html_' . get_locale();
|
||||
$html = wp_cache_get($cache_key, 'partials');
|
||||
|
||||
if (false === $html) {
|
||||
ob_start();
|
||||
get_sidebar();
|
||||
$html = ob_get_clean();
|
||||
wp_cache_set($cache_key, $html, 'partials', HOUR_IN_SECONDS);
|
||||
}
|
||||
|
||||
return $html;
|
||||
}
|
||||
```
|
||||
|
||||
## In-Memory Caching (Request-Scoped)
|
||||
|
||||
### Static Variables
|
||||
|
||||
```php
|
||||
function get_current_user_data() {
|
||||
static $user_data = null;
|
||||
|
||||
if (null === $user_data) {
|
||||
$user_data = expensive_user_query();
|
||||
}
|
||||
|
||||
return $user_data;
|
||||
}
|
||||
```
|
||||
|
||||
### Global Variable Pattern
|
||||
|
||||
```php
|
||||
// Store request-scoped data
|
||||
global $my_plugin_cache;
|
||||
$my_plugin_cache = [];
|
||||
|
||||
function get_item($id) {
|
||||
global $my_plugin_cache;
|
||||
|
||||
if (!isset($my_plugin_cache[$id])) {
|
||||
$my_plugin_cache[$id] = fetch_item($id);
|
||||
}
|
||||
|
||||
return $my_plugin_cache[$id];
|
||||
}
|
||||
```
|
||||
|
||||
## Cache Invalidation Strategy
|
||||
|
||||
### Event-Based Invalidation
|
||||
|
||||
```php
|
||||
// Clear post-related caches when post is updated
|
||||
add_action('save_post', function($post_id, $post) {
|
||||
// Clear specific post cache
|
||||
wp_cache_delete("post_{$post_id}", 'posts');
|
||||
|
||||
// Clear related caches
|
||||
wp_cache_delete('recent_posts', 'listings');
|
||||
wp_cache_delete('homepage_posts', 'listings');
|
||||
|
||||
// Clear category archive caches
|
||||
$categories = wp_get_post_categories($post_id);
|
||||
foreach ($categories as $cat_id) {
|
||||
wp_cache_delete("category_{$cat_id}_posts", 'archives');
|
||||
}
|
||||
}, 10, 2);
|
||||
```
|
||||
|
||||
### Tag-Based Invalidation (Advanced)
|
||||
|
||||
```php
|
||||
// Store cache tags
|
||||
function set_tagged_cache($key, $data, $tags, $ttl) {
|
||||
wp_cache_set($key, $data, '', $ttl);
|
||||
|
||||
foreach ($tags as $tag) {
|
||||
$tag_keys = wp_cache_get("tag_{$tag}", 'cache_tags') ?: [];
|
||||
$tag_keys[] = $key;
|
||||
wp_cache_set("tag_{$tag}", array_unique($tag_keys), 'cache_tags');
|
||||
}
|
||||
}
|
||||
|
||||
// Invalidate by tag
|
||||
function invalidate_tag($tag) {
|
||||
$keys = wp_cache_get("tag_{$tag}", 'cache_tags') ?: [];
|
||||
foreach ($keys as $key) {
|
||||
wp_cache_delete($key);
|
||||
}
|
||||
wp_cache_delete("tag_{$tag}", 'cache_tags');
|
||||
}
|
||||
```
|
||||
|
||||
## Memcached vs Redis
|
||||
|
||||
| Feature | Memcached | Redis |
|
||||
|---------|-----------|-------|
|
||||
| Speed | Slightly faster | Fast |
|
||||
| Data types | String only | Strings, lists, sets, hashes |
|
||||
| Persistence | No | Optional |
|
||||
| Memory efficiency | Higher | Lower |
|
||||
| Cache groups | Limited | Full support |
|
||||
| Complexity | Simple | More features |
|
||||
|
||||
**Recommendation**: Use what your host provides. Memcached for simple caching, Redis if you need advanced features.
|
||||
356
skills/wp-performance-review/references/measurement-guide.md
Normal file
356
skills/wp-performance-review/references/measurement-guide.md
Normal file
@@ -0,0 +1,356 @@
|
||||
# High-Traffic Event Preparation & Performance Measurement
|
||||
|
||||
Guide for preparing WordPress sites for traffic surges and measuring performance.
|
||||
|
||||
## When to Use This Reference
|
||||
|
||||
This guide is for:
|
||||
- **Pre-launch audits** - Checklist before high-traffic events
|
||||
- **Performance testing** - How to load test with K6
|
||||
- **Monitoring setup** - Query Monitor, New Relic integration
|
||||
- **Alert configuration** - Threshold recommendations
|
||||
|
||||
**For code review**, use `anti-patterns.md` instead. This guide covers operational preparation.
|
||||
|
||||
## Code-Level Pre-Event Checks
|
||||
|
||||
Before a high-traffic event, scan the codebase for:
|
||||
|
||||
```bash
|
||||
# Things that WILL break under load
|
||||
grep -rn "posts_per_page.*-1" . # Unbounded queries
|
||||
grep -rn "session_start" . # Cache bypass
|
||||
grep -rn "setInterval.*fetch\|setInterval.*ajax" . # Polling
|
||||
|
||||
# Things that DEGRADE under load
|
||||
grep -rn "wp_remote_get\|wp_remote_post" . # Uncached HTTP
|
||||
grep -rn "update_option\|add_option" . # DB writes
|
||||
```
|
||||
|
||||
## Pre-Event Checklist
|
||||
|
||||
Before a high-traffic event (product launch, viral content, marketing campaign), verify:
|
||||
|
||||
### Cache Hit Rate
|
||||
- [ ] **Page cache hit rate > 90%** - Check CDN/edge cache analytics
|
||||
- [ ] **Object cache (memcached) hit rate > 90%** - Check cache stats
|
||||
- [ ] **InnoDB buffer pool hit rate > 95%** - MySQL performance schema
|
||||
- [ ] Identify pages consistently missing cache - optimize or exclude from event
|
||||
|
||||
### Database Health
|
||||
- [ ] **No INSERT/UPDATE on uncached page loads** - Use Query Monitor
|
||||
- Plugins persistently updating options = high DB CPU + binlog growth
|
||||
- [ ] **Review P95 upstream response times** - Identify slowest endpoints
|
||||
- High traffic to slow endpoints ties up PHP workers
|
||||
- [ ] **Check for slow queries** - Queries > 100ms under normal load
|
||||
- [ ] **Verify indexes** - Run EXPLAIN on common queries
|
||||
|
||||
### Error Monitoring
|
||||
- [ ] **Clean PHP logs** - No fatal errors, minimize warnings
|
||||
- Easier to debug issues if logs aren't full of noise
|
||||
- [ ] **Zero 500 errors in 24h** - Check application error rate
|
||||
- [ ] **Review error tracking** - New Relic, Sentry, etc.
|
||||
|
||||
### Load Testing
|
||||
- [ ] **Run load test simulating expected traffic**
|
||||
- [ ] **Test specific high-traffic URLs** - Homepage, landing pages
|
||||
- [ ] **Monitor during test** - CPU, memory, DB connections, response times
|
||||
|
||||
## Traffic Pattern Types
|
||||
|
||||
### Sustained High Traffic on Multiple URLs
|
||||
**Impact**: Strain on PHP workers, database connections, memory
|
||||
**Preparation**:
|
||||
- Scale horizontally (more app servers)
|
||||
- Increase PHP worker count
|
||||
- Optimize database queries
|
||||
- Implement aggressive caching
|
||||
|
||||
### Large Spikes on Few URLs
|
||||
**Impact**: Bottleneck on specific endpoints, potential cache stampede
|
||||
**Preparation**:
|
||||
- Pre-warm caches for target URLs
|
||||
- Implement cache locking to prevent stampede
|
||||
- Consider static HTML fallback for extreme cases
|
||||
|
||||
### Uncached Content Under Load
|
||||
**Impact**: Every request hits origin, database overload
|
||||
**Preparation**:
|
||||
- Identify why pages bypass cache (cookies, sessions, query params)
|
||||
- Implement partial caching for personalized pages
|
||||
- Use ESI (Edge Side Includes) for dynamic fragments
|
||||
|
||||
## Load Testing with K6
|
||||
|
||||
K6 is an open-source load testing tool for simulating traffic.
|
||||
|
||||
### Installation
|
||||
```bash
|
||||
# macOS
|
||||
brew install k6
|
||||
|
||||
# Linux
|
||||
sudo apt-get install k6
|
||||
|
||||
# Docker
|
||||
docker run -i grafana/k6 run - <script.js
|
||||
```
|
||||
|
||||
### Basic Load Test Script
|
||||
```javascript
|
||||
// load-test.js
|
||||
import http from 'k6/http';
|
||||
import { check, sleep } from 'k6';
|
||||
|
||||
export const options = {
|
||||
vus: 100, // 100 virtual users
|
||||
duration: '5m', // 5 minute test
|
||||
thresholds: {
|
||||
http_req_duration: ['p(95)<500'], // 95% of requests under 500ms
|
||||
http_req_failed: ['rate<0.01'], // Less than 1% failures
|
||||
},
|
||||
};
|
||||
|
||||
export default function() {
|
||||
// Test homepage
|
||||
let homeResponse = http.get('https://example.com/');
|
||||
check(homeResponse, {
|
||||
'homepage status 200': (r) => r.status === 200,
|
||||
'homepage < 500ms': (r) => r.timings.duration < 500,
|
||||
});
|
||||
|
||||
// Test archive page
|
||||
let archiveResponse = http.get('https://example.com/category/news/');
|
||||
check(archiveResponse, {
|
||||
'archive status 200': (r) => r.status === 200,
|
||||
});
|
||||
|
||||
sleep(1); // Wait 1 second between iterations
|
||||
}
|
||||
```
|
||||
|
||||
### Running Load Tests
|
||||
```bash
|
||||
# Basic run
|
||||
k6 run load-test.js
|
||||
|
||||
# With more virtual users
|
||||
k6 run --vus 200 --duration 10m load-test.js
|
||||
|
||||
# Output to JSON for analysis
|
||||
k6 run --out json=results.json load-test.js
|
||||
```
|
||||
|
||||
### Key Metrics to Monitor
|
||||
| Metric | Good | Concerning |
|
||||
|--------|------|------------|
|
||||
| `http_req_duration` (p95) | < 500ms | > 1000ms |
|
||||
| `http_req_failed` | < 1% | > 5% |
|
||||
| `http_reqs` (throughput) | Stable | Declining under load |
|
||||
| `vus` vs response time | Linear | Exponential degradation |
|
||||
|
||||
## Measuring with Query Monitor
|
||||
|
||||
Query Monitor plugin provides detailed performance insights per page.
|
||||
|
||||
### Key Panels
|
||||
|
||||
**Overview Panel**
|
||||
- Total page generation time
|
||||
- Database query time
|
||||
- HTTP API call time
|
||||
|
||||
**Database Queries Panel**
|
||||
- Individual query times
|
||||
- Duplicate queries (same query run multiple times)
|
||||
- Slow queries highlighted
|
||||
- Query origin (which plugin/theme)
|
||||
|
||||
**Object Cache Panel**
|
||||
- Cache hit/miss ratio
|
||||
- Total cache operations
|
||||
- Cache groups usage
|
||||
|
||||
### Custom Timing with Query Monitor
|
||||
```php
|
||||
// Start timer
|
||||
do_action('qm/start', 'my_operation');
|
||||
|
||||
// Your code here
|
||||
expensive_operation();
|
||||
|
||||
// Stop timer
|
||||
do_action('qm/stop', 'my_operation');
|
||||
|
||||
// For loops, use lap:
|
||||
do_action('qm/start', 'loop_operation');
|
||||
foreach ($items as $item) {
|
||||
process_item($item);
|
||||
do_action('qm/lap', 'loop_operation');
|
||||
}
|
||||
do_action('qm/stop', 'loop_operation');
|
||||
```
|
||||
|
||||
### Target Metrics in Query Monitor
|
||||
| Metric | Target | Investigate |
|
||||
|--------|--------|-------------|
|
||||
| Page generation | < 200ms | > 500ms |
|
||||
| Database queries | < 50 | > 100 |
|
||||
| Duplicate queries | 0 | > 5 |
|
||||
| Slowest query | < 50ms | > 100ms |
|
||||
| Object cache hits | > 90% | < 80% |
|
||||
|
||||
## Measuring with PHP Logging
|
||||
|
||||
Manual timing for specific code sections.
|
||||
|
||||
### Basic Timing
|
||||
```php
|
||||
function my_function() {
|
||||
$start_time = microtime(true);
|
||||
|
||||
// Code to measure
|
||||
expensive_operation();
|
||||
|
||||
$end_time = microtime(true);
|
||||
$execution_time = $end_time - $start_time;
|
||||
|
||||
error_log(sprintf(
|
||||
'my_function execution time: %.4f seconds',
|
||||
$execution_time
|
||||
));
|
||||
}
|
||||
```
|
||||
|
||||
### Timing Wrapper Function
|
||||
```php
|
||||
function measure_execution($callback, $label) {
|
||||
$start = microtime(true);
|
||||
$result = $callback();
|
||||
$duration = microtime(true) - $start;
|
||||
|
||||
if ($duration > 0.1) { // Log if > 100ms
|
||||
error_log(sprintf('[SLOW] %s: %.4fs', $label, $duration));
|
||||
}
|
||||
|
||||
return $result;
|
||||
}
|
||||
|
||||
// Usage
|
||||
$data = measure_execution(function() {
|
||||
return expensive_query();
|
||||
}, 'expensive_query');
|
||||
```
|
||||
|
||||
### Memory Measurement
|
||||
```php
|
||||
$mem_start = memory_get_usage();
|
||||
|
||||
// Code that might use lots of memory
|
||||
$large_array = process_data();
|
||||
|
||||
$mem_end = memory_get_usage();
|
||||
$mem_used = ($mem_end - $mem_start) / 1024 / 1024;
|
||||
|
||||
error_log(sprintf('Memory used: %.2f MB', $mem_used));
|
||||
```
|
||||
|
||||
## Measuring with New Relic
|
||||
|
||||
New Relic provides APM (Application Performance Monitoring) for production.
|
||||
|
||||
### Custom Transaction Naming
|
||||
```php
|
||||
// Name transactions for better grouping
|
||||
if (function_exists('newrelic_name_transaction')) {
|
||||
if (is_single()) {
|
||||
newrelic_name_transaction('single-post');
|
||||
} elseif (is_archive()) {
|
||||
newrelic_name_transaction('archive');
|
||||
}
|
||||
}
|
||||
```
|
||||
|
||||
### Custom Instrumentation
|
||||
```php
|
||||
// Track custom code segments
|
||||
if (function_exists('newrelic_start_transaction')) {
|
||||
newrelic_start_transaction('my_custom_process');
|
||||
|
||||
// Your code
|
||||
process_data();
|
||||
|
||||
newrelic_end_transaction();
|
||||
}
|
||||
|
||||
// Add custom attributes for filtering
|
||||
if (function_exists('newrelic_add_custom_parameter')) {
|
||||
newrelic_add_custom_parameter('post_type', get_post_type());
|
||||
newrelic_add_custom_parameter('user_role', $current_user_role);
|
||||
}
|
||||
```
|
||||
|
||||
### Key New Relic Metrics
|
||||
- **Apdex score** - User satisfaction (target: > 0.9)
|
||||
- **Web transaction time** - Average response time
|
||||
- **Throughput** - Requests per minute
|
||||
- **Error rate** - Percentage of failed requests
|
||||
- **Database time** - Time spent in DB queries
|
||||
|
||||
## Performance Alerting
|
||||
|
||||
Set up alerts for performance degradation.
|
||||
|
||||
### Key Alert Thresholds
|
||||
| Metric | Warning | Critical |
|
||||
|--------|---------|----------|
|
||||
| Response time (p95) | > 1s | > 3s |
|
||||
| Error rate | > 1% | > 5% |
|
||||
| CPU usage | > 70% | > 90% |
|
||||
| Memory usage | > 80% | > 95% |
|
||||
| Database connections | > 80% of max | > 95% of max |
|
||||
|
||||
### Alert Channels
|
||||
- **Immediate**: PagerDuty, Slack, SMS for critical
|
||||
- **Batched**: Email digest for warnings
|
||||
- **Dashboard**: Real-time visibility for operations team
|
||||
|
||||
## Performance Regression Detection
|
||||
|
||||
Catch performance issues before production.
|
||||
|
||||
### Baseline Metrics
|
||||
Establish baseline performance metrics:
|
||||
```
|
||||
Homepage: < 200ms, < 30 queries
|
||||
Archive: < 300ms, < 50 queries
|
||||
Single: < 250ms, < 40 queries
|
||||
Search: < 500ms (with ElasticSearch)
|
||||
```
|
||||
|
||||
### CI/CD Integration
|
||||
```yaml
|
||||
# Example: Run performance check in CI
|
||||
- name: Performance Check
|
||||
run: |
|
||||
RESPONSE_TIME=$(curl -o /dev/null -s -w '%{time_total}' https://staging.example.com/)
|
||||
if (( $(echo "$RESPONSE_TIME > 0.5" | bc -l) )); then
|
||||
echo "Warning: Response time ${RESPONSE_TIME}s exceeds threshold"
|
||||
exit 1
|
||||
fi
|
||||
```
|
||||
|
||||
### Query Count Monitoring
|
||||
```php
|
||||
// Add to theme's functions.php for staging
|
||||
add_action('shutdown', function() {
|
||||
if (!defined('SAVEQUERIES') || !SAVEQUERIES) return;
|
||||
|
||||
global $wpdb;
|
||||
$query_count = count($wpdb->queries);
|
||||
|
||||
if ($query_count > 100) {
|
||||
error_log("[PERFORMANCE] High query count: $query_count");
|
||||
}
|
||||
});
|
||||
```
|
||||
328
skills/wp-performance-review/references/wp-query-guide.md
Normal file
328
skills/wp-performance-review/references/wp-query-guide.md
Normal file
@@ -0,0 +1,328 @@
|
||||
# WP_Query Optimization Guide
|
||||
|
||||
Best practices for efficient database queries in WordPress applications.
|
||||
|
||||
## WP_Query Review Checklist
|
||||
|
||||
When reviewing WP_Query code, verify:
|
||||
|
||||
- [ ] `posts_per_page` is set (not -1, not missing)
|
||||
- [ ] `no_found_rows => true` if not paginating
|
||||
- [ ] `fields => 'ids'` if only IDs needed
|
||||
- [ ] `update_post_meta_cache => false` if meta not used
|
||||
- [ ] `update_post_term_cache => false` if terms not used
|
||||
- [ ] Date limits on archive queries (recent content only)
|
||||
- [ ] `include_children => false` if child terms not needed
|
||||
- [ ] No `post__not_in` with large arrays
|
||||
- [ ] No `meta_query` on `meta_value` (use taxonomy or key presence)
|
||||
- [ ] Results cached with `wp_cache_set` if repeated
|
||||
|
||||
## Core Principles
|
||||
|
||||
1. **Always set limits** - Never use `posts_per_page => -1`
|
||||
2. **Validate inputs** - Check IDs before querying
|
||||
3. **Limit scope** - Use date ranges, taxonomies to narrow results
|
||||
4. **Skip unnecessary work** - Use `no_found_rows`, `fields`
|
||||
5. **Pre-fetch related data** - Avoid N+1 queries
|
||||
6. **Cache results** - Use object cache for repeated queries
|
||||
|
||||
## Essential Query Arguments
|
||||
|
||||
### Limiting Results
|
||||
|
||||
```php
|
||||
// REQUIRED: Always limit results
|
||||
'posts_per_page' => 10,
|
||||
|
||||
// Skip counting total posts if not paginating
|
||||
'no_found_rows' => true, // Skips SQL_CALC_FOUND_ROWS
|
||||
|
||||
// Return only IDs when you don't need full post objects
|
||||
'fields' => 'ids',
|
||||
|
||||
// Suppress filters when you need raw results
|
||||
'suppress_filters' => true,
|
||||
```
|
||||
|
||||
### Date-Based Limiting
|
||||
|
||||
For sites with years of content, limit queries to relevant time ranges:
|
||||
|
||||
```php
|
||||
// Limit to recent content
|
||||
$query = new WP_Query([
|
||||
'post_type' => 'post',
|
||||
'posts_per_page' => 10,
|
||||
'date_query' => [
|
||||
[
|
||||
'after' => '3 months ago',
|
||||
'inclusive' => true
|
||||
]
|
||||
]
|
||||
]);
|
||||
|
||||
// Specific date range
|
||||
'date_query' => [
|
||||
'after' => '2024-01-01',
|
||||
'before' => '2024-12-31'
|
||||
]
|
||||
|
||||
// Dynamic date filtering via pre_get_posts
|
||||
add_action('pre_get_posts', function($query) {
|
||||
if (!is_admin() && $query->is_main_query() && $query->is_category('news')) {
|
||||
$query->set('date_query', [
|
||||
'after' => date('Y-m-d', strtotime('-3 months'))
|
||||
]);
|
||||
}
|
||||
});
|
||||
```
|
||||
|
||||
### Taxonomy Query Optimization
|
||||
|
||||
```php
|
||||
// Exclude child terms to reduce query complexity
|
||||
'tax_query' => [
|
||||
[
|
||||
'taxonomy' => 'category',
|
||||
'field' => 'term_id',
|
||||
'terms' => [6],
|
||||
'include_children' => false // Important for performance
|
||||
]
|
||||
]
|
||||
|
||||
// Use term_id instead of slug (avoids extra lookup)
|
||||
'field' => 'term_id', // Faster than 'slug' or 'name'
|
||||
```
|
||||
|
||||
### Meta Query Optimization
|
||||
|
||||
Meta queries are expensive - minimize usage:
|
||||
|
||||
```php
|
||||
// ❌ AVOID: Multiple meta conditions
|
||||
'meta_query' => [
|
||||
'relation' => 'AND',
|
||||
['key' => 'color', 'value' => 'red'],
|
||||
['key' => 'size', 'value' => 'large'],
|
||||
['key' => 'price', 'compare' => '>=', 'value' => 100]
|
||||
]
|
||||
|
||||
// ✅ BETTER: Use taxonomies for filterable attributes
|
||||
// Register 'color' and 'size' as taxonomies instead
|
||||
|
||||
// ✅ ALTERNATIVE: Offload to ElasticSearch
|
||||
// Configure ElasticPress to index post meta
|
||||
```
|
||||
|
||||
## Pre-fetching & Cache Priming
|
||||
|
||||
### Update Meta Cache in Bulk
|
||||
|
||||
```php
|
||||
// After running WP_Query, prime the meta cache
|
||||
$query = new WP_Query($args);
|
||||
$post_ids = wp_list_pluck($query->posts, 'ID');
|
||||
|
||||
// Prime meta cache (single query vs N queries)
|
||||
update_postmeta_cache($post_ids);
|
||||
|
||||
// Prime term cache
|
||||
update_object_term_cache($post_ids, 'post');
|
||||
|
||||
// Prime user cache for authors
|
||||
$author_ids = wp_list_pluck($query->posts, 'post_author');
|
||||
cache_users($author_ids);
|
||||
```
|
||||
|
||||
### WP_Query Built-in Cache Updates
|
||||
|
||||
```php
|
||||
// These are on by default but be aware:
|
||||
'update_post_meta_cache' => true, // Updates meta cache after query
|
||||
'update_post_term_cache' => true, // Updates term cache after query
|
||||
|
||||
// Disable if you don't need meta/terms (saves queries)
|
||||
'update_post_meta_cache' => false,
|
||||
'update_post_term_cache' => false,
|
||||
```
|
||||
|
||||
## Caching Query Results
|
||||
|
||||
### Object Cache for Repeated Queries
|
||||
|
||||
```php
|
||||
function get_featured_posts() {
|
||||
$cache_key = 'featured_posts_v1';
|
||||
$posts = wp_cache_get($cache_key, 'my_plugin');
|
||||
|
||||
if (false === $posts) {
|
||||
$query = new WP_Query([
|
||||
'post_type' => 'post',
|
||||
'posts_per_page' => 5,
|
||||
'meta_key' => 'featured',
|
||||
'meta_value' => '1'
|
||||
]);
|
||||
|
||||
$posts = $query->posts;
|
||||
wp_cache_set($cache_key, $posts, 'my_plugin', HOUR_IN_SECONDS);
|
||||
}
|
||||
|
||||
return $posts;
|
||||
}
|
||||
|
||||
// Invalidate when posts are updated
|
||||
add_action('save_post', function($post_id) {
|
||||
wp_cache_delete('featured_posts_v1', 'my_plugin');
|
||||
});
|
||||
```
|
||||
|
||||
### Static Variable Caching (Request-Scoped)
|
||||
|
||||
```php
|
||||
function get_expensive_data($id) {
|
||||
static $cache = [];
|
||||
|
||||
if (!isset($cache[$id])) {
|
||||
$cache[$id] = expensive_computation($id);
|
||||
}
|
||||
|
||||
return $cache[$id];
|
||||
}
|
||||
```
|
||||
|
||||
## ElasticSearch Offloading
|
||||
|
||||
For complex searches, offload to ElasticSearch via ElasticPress:
|
||||
|
||||
### When to Offload
|
||||
|
||||
- Full-text search queries
|
||||
- Complex meta queries
|
||||
- Faceted search / filtering
|
||||
- Large result sets with sorting
|
||||
- Aggregations / analytics
|
||||
|
||||
### When NOT to Offload
|
||||
|
||||
- Simple primary key lookups
|
||||
- Transactional data requiring ACID
|
||||
- Data requiring immediate consistency
|
||||
- Small data sets
|
||||
|
||||
```php
|
||||
// ElasticPress automatically intercepts WP_Query
|
||||
// for search queries when configured
|
||||
|
||||
// Force MySQL for specific queries
|
||||
$query = new WP_Query([
|
||||
'ep_integrate' => false, // Skip ElasticSearch
|
||||
// ... args
|
||||
]);
|
||||
```
|
||||
|
||||
## Analyzing Queries with EXPLAIN
|
||||
|
||||
Use EXPLAIN to identify slow queries:
|
||||
|
||||
```sql
|
||||
EXPLAIN SELECT * FROM wp_posts
|
||||
WHERE post_type = 'post'
|
||||
AND post_status = 'publish'
|
||||
ORDER BY post_date DESC
|
||||
LIMIT 10;
|
||||
```
|
||||
|
||||
### Key EXPLAIN Indicators
|
||||
|
||||
| Column | Good Value | Bad Value |
|
||||
|--------|------------|-----------|
|
||||
| `type` | `const`, `eq_ref`, `ref`, `range` | `ALL` (full table scan) |
|
||||
| `key` | Named index | `NULL` (no index used) |
|
||||
| `rows` | Small number | Large number |
|
||||
| `Extra` | `Using index` | `Using filesort`, `Using temporary` |
|
||||
|
||||
### Common Optimization Actions
|
||||
|
||||
```sql
|
||||
-- Add index for frequently queried meta key
|
||||
ALTER TABLE wp_postmeta ADD INDEX meta_key_value (meta_key, meta_value(50));
|
||||
|
||||
-- Add composite index
|
||||
ALTER TABLE wp_posts ADD INDEX type_status_date (post_type, post_status, post_date);
|
||||
```
|
||||
|
||||
## Query Monitor Integration
|
||||
|
||||
Use Query Monitor plugin to identify:
|
||||
|
||||
1. **Slow queries** - Queries over threshold
|
||||
2. **Duplicate queries** - Same query run multiple times
|
||||
3. **Queries by component** - Which plugin/theme caused each query
|
||||
4. **Query count** - Total queries per page load
|
||||
|
||||
### Target Metrics
|
||||
|
||||
| Metric | Target | Concern |
|
||||
|--------|--------|---------|
|
||||
| Total queries | < 50 | > 100 |
|
||||
| Duplicate queries | 0 | > 5 |
|
||||
| Slowest query | < 50ms | > 100ms |
|
||||
| Total query time | < 100ms | > 500ms |
|
||||
|
||||
## Common Query Patterns
|
||||
|
||||
### Get Latest Posts (Optimized)
|
||||
|
||||
```php
|
||||
$query = new WP_Query([
|
||||
'post_type' => 'post',
|
||||
'post_status' => 'publish',
|
||||
'posts_per_page' => 10,
|
||||
'no_found_rows' => true,
|
||||
'update_post_term_cache' => false, // If not displaying categories
|
||||
]);
|
||||
```
|
||||
|
||||
### Get Posts by IDs (Optimized)
|
||||
|
||||
```php
|
||||
$query = new WP_Query([
|
||||
'post__in' => $post_ids,
|
||||
'posts_per_page' => count($post_ids),
|
||||
'orderby' => 'post__in', // Preserve ID order
|
||||
'no_found_rows' => true,
|
||||
'ignore_sticky_posts' => true
|
||||
]);
|
||||
```
|
||||
|
||||
### Check if Posts Exist (Optimized)
|
||||
|
||||
```php
|
||||
// Don't fetch full posts just to check existence
|
||||
$query = new WP_Query([
|
||||
'post_type' => 'product',
|
||||
'posts_per_page' => 1,
|
||||
'fields' => 'ids',
|
||||
'no_found_rows' => true,
|
||||
'update_post_meta_cache' => false,
|
||||
'update_post_term_cache' => false
|
||||
]);
|
||||
$has_products = $query->have_posts();
|
||||
```
|
||||
|
||||
### Count Posts (Optimized)
|
||||
|
||||
```php
|
||||
// Use wp_count_posts() for simple counts
|
||||
$counts = wp_count_posts('post');
|
||||
$published = $counts->publish;
|
||||
|
||||
// For filtered counts, use found_posts
|
||||
$query = new WP_Query([
|
||||
'post_type' => 'post',
|
||||
'posts_per_page' => 1, // Minimize actual retrieval
|
||||
'category_name' => 'news',
|
||||
// no_found_rows must be FALSE to get found_posts
|
||||
]);
|
||||
$total = $query->found_posts;
|
||||
```
|
||||
Reference in New Issue
Block a user