Safe Maintenance and Incident Response

First Response

When PHP breaks a site, collect facts before changing many things.

• Check PHP error logs.
• Identify the newest code, plugin, theme, or PHP version change.
• Confirm whether admin, front end, REST, cron, or CLI is affected.
• Preserve a backup before database changes.

Disable a Broken Plugin

If WP Admin is unavailable, rename the plugin directory over SSH or file manager access.

wp-content/plugins/my-plugin
wp-content/plugins/my-plugin.disabled

This is safer than editing WordPress core or deleting plugin files.

Maintenance Mode

During planned work, use maintenance mode or a staging site. Avoid running risky migrations directly on a live site during peak traffic.

Rollback Plan

Every deployment should have a rollback path:

Area	Rollback
Plugin code	Restore previous release
Theme code	Restore previous theme build
Database	Restore backup or reverse migration
Configuration	Restore known-good constants and settings

Common Pitfalls

• Making several emergency changes before identifying the cause
• Debugging by editing core files
• Running destructive SQL without a backup
• Assuming a plugin conflict is fixed because the symptom disappeared once

Deep WordPress Application

This lesson is most useful when applied to real WordPress code rather than isolated PHP examples. Safe Maintenance and Incident Response affects how a site behaves under plugins, themes, editors, logged-in users, guests, cached pages, REST requests, and production traffic.

Debugging and production operations require evidence: logs, reproducible symptoms, clear rollback plans, and minimal emergency changes.

A practical implementation should answer four questions before code is written:

• Which WordPress request context will run this code?
• Which API or hook owns the behavior?
• Which data is trusted, and which data must be normalized?
• What should happen when the expected data, permission, or dependency is missing?

WordPress API Anchor

For this topic, a common API or integration point is WP_DEBUG_LOG. The exact function may vary by feature, but the design principle is stable: use WordPress APIs before inventing custom plumbing.

A common hook or lifecycle point for this topic is shutdown. Confirm the hook runs in the request type you care about before attaching expensive or state-changing logic.

Focused Code Pattern

if (defined('WP_DEBUG') && WP_DEBUG) {
    error_log('MyPlugin reached checkpoint: ' . current_time('mysql'));
}

This pattern should still be adapted to the exact feature. Add capability checks for private data, nonces for browser-submitted state changes, and output escaping when rendering values.

Data Flow Walkthrough

Stage	What To Decide	WordPress Example
Source	Where the value comes from	Request data, option, post meta, user meta, REST parameter, remote API
Trust	Whether the value can be used directly	Treat external, request, and old saved values as untrusted
Normalize	How PHP converts the value into a safe shape	absint(), sanitize_text_field(), sanitize_key(), custom allow-list
Authorize	Who can read or change it	current_user_can() with a specific capability
Persist	Where the value belongs	Option, post meta, user meta, taxonomy term, custom table
Render	How the value leaves PHP	esc_html(), esc_attr(), esc_url(), wp_kses_post()

Applied Practice

Reproduce a harmless warning in development, locate it in the log, fix it, and confirm the log stays clean afterward.

When practicing, do not stop when the happy path works. WordPress code becomes reliable when the failure paths are equally deliberate.

Expanded Decision Guide

Decision	Prefer This	Avoid This
Where code lives	A plugin for durable behavior, a theme for presentation	Editing WordPress core or vendor plugin files
How data enters	A named request handler with validation	Reading superglobals deep inside templates
How data is stored	WordPress APIs with clear keys and types	Unstructured arrays with undocumented meanings
How output is printed	Escape at the final output context	Trusting saved data because it came from the database
How failures behave	Return early, log safely, show useful messages	Fatal errors, blank pages, or exposed stack details
How changes ship	Small reviewed changes with rollback notes	Large untested edits directly on production

Implementation Checklist

Use this checklist when applying the lesson in a real WordPress codebase.

• Identify whether the code belongs in a plugin, child theme, block, mu-plugin, or deployment script.
• Name functions, classes, options, actions, filters, and CSS hooks with a project-specific prefix.
• Confirm which hook should run the code and whether that hook fires in admin, front end, AJAX, REST, cron, or CLI contexts.
• Validate every value that comes from a request, database option, custom field, remote API, cookie, or file.
• Sanitize before storing data and escape at the exact output boundary.
• Check the narrowest useful capability before reading private data or changing state.
• Add nonces or signed requests for state-changing browser actions.
• Keep database queries narrow by requesting only the fields and post types needed.
• Reset global WordPress state after custom queries, site switching, or temporary filters.
• Add a short manual test note for the main success path and at least one failure path.

Extended Example Pattern

The following pattern is intentionally small. It demonstrates how to keep request handling, normalization, and output separate enough to review.

add_action('admin_post_myplugin_example_action', 'myplugin_handle_example_action');

function myplugin_handle_example_action(): void {
    if (! current_user_can('manage_options')) {
        wp_die(esc_html__('You are not allowed to do that.', 'my-plugin'));
    }

    check_admin_referer('myplugin_example_action', 'myplugin_nonce');

    $label = isset($_POST['label'])
        ? sanitize_text_field(wp_unslash($_POST['label']))
        : '';

    if ('' === $label) {
        wp_safe_redirect(add_query_arg('status', 'missing-label', wp_get_referer() ?: admin_url()));
        exit;
    }

    update_option('myplugin_example_label', $label, false);

    wp_safe_redirect(add_query_arg('status', 'saved', wp_get_referer() ?: admin_url()));
    exit;
}

Troubleshooting Matrix

Symptom	Likely Cause	First Check
Callback never runs	Wrong hook name, priority, or load context	Confirm the hook fires in this request type
Data saves incorrectly	Missing unslash, sanitization, or type normalization	Log sanitized values in development only
Output looks broken	Escaped for the wrong context or escaped too early	Check whether the output is text, HTML, URL, or attribute
Permission bug	Role check is too broad or capability is missing an object ID	Use current_user_can() with the specific operation
Slow page	Query, HTTP request, or loop work runs on every request	Profile with Query Monitor and add caching or batching
Works locally only	PHP version, plugin dependency, rewrite rule, or environment setting differs	Compare environment versions and enabled plugins

Practice Exercise

1. Recreate the smallest practical example from this lesson in a local WordPress site.
2. Add one valid input and one invalid input.
3. Confirm the invalid input fails safely without changing stored data.
4. Confirm the valid input is sanitized before storage.
5. Render the stored value in at least two contexts, such as text and attribute output.
6. Add a capability check and verify a lower-privilege user cannot perform the action.
7. Temporarily enable debug logging and confirm no PHP warnings appear.
8. Remove temporary logs before treating the example as complete.

Review Questions

• What WordPress hook or API makes this lesson reliable in the right request context?
• Which values in the example are untrusted at the moment they enter PHP?
• Where should validation happen, and where should output escaping happen?
• What user capability is required for the action, and why is that the narrowest useful choice?
• What is the expected behavior when the required data is missing or malformed?
• What would need to change before this code runs safely on a large production site?

Production Notes

Production WordPress PHP should be easy to audit under pressure. Keep the control flow direct, keep side effects visible, and prefer small functions that name the business rule they enforce. If a future maintainer cannot identify the request source, permission check, data normalization, and output boundary in a few minutes, the code is too implicit.