Fix Guides

Something broke. Here's how to fix it.

128 plain-language guides for when a website, app or server stops working. Each one says what the error means, why it happens, and how to fix it.

Site down

All 2 →

Your website won't load, for you or for your customers.

Why is my website down? Check whether it's down for everyone, read the error on the page, then work through domain, hosting and recent changes. Most outages trace back to one of those. Is my website down? How to check Load it on another network, check the status code it returns, and check its DNS. If it fails from everywhere, it's down; if only for you, it's your connection.

Error codes

All 10 →

The page shows a number instead of your site. Here's what each one means.

500 Internal Server Error Your server hit an error it didn't handle. The cause is in the server's error log: usually a code bug, a bad config change, a failed plugin or a lost database. 502 Bad Gateway The server in front of your app got a bad answer, or none, from your app. Usually the app has crashed, isn't running, or is listening on the wrong port. 503 Service Unavailable The server is up but can't take requests right now: overloaded, stuck in maintenance mode, or out of resources. Here's how to tell which. 504 Gateway Timeout Your app took too long to answer and the server in front of it gave up. Usually a slow database query, a slow outside API, or an overloaded server. 404 Not Found The page doesn't exist at that address. Usually a deleted or moved page, a typo in a link, or a site migration that broke URLs. Fix the link or add a redirect. 403 Forbidden The server understood the request but won't allow it. Usually file permissions, a missing index page, or a firewall or security rule blocking the visitor. 400 Bad Request The server rejected the request as malformed. For visitors it's often bad cookies; for apps, the data sent isn't what the server expects. 401 Unauthorized The request had no valid login or key. For people, log in again. For apps and APIs, the token or API key is missing, expired, or wrong. 408 Request Timeout The server closed a connection because the browser didn't finish sending its request in time. Usually a slow or unstable connection, or a large upload. 429 Too Many Requests A rate limit was hit: too many requests too fast. On your site, check rate-limit and bot rules. In your app, slow down calls to that API.

Cloudflare errors

All 8 →

A Cloudflare error page means Cloudflare couldn't get a proper answer from your server.

Cloudflare error: what it means and how to fix it A Cloudflare error page almost always means Cloudflare is fine but couldn't get a proper answer from your own server. The number says which part failed. Cloudflare error 522: connection timed out Error 522: Cloudflare couldn't connect to your server in time. The server is overloaded or offline, or a firewall is silently dropping Cloudflare. Cloudflare error 521: web server is down Error 521: your server refused Cloudflare's connection. The web server is stopped or a firewall is blocking Cloudflare. Start it, allow Cloudflare. Cloudflare error 520 Error 520: your server sent Cloudflare an empty or unexpected response. Usually a crash, a security plugin blocking Cloudflare, or response headers too large. Cloudflare error 524: a timeout occurred Error 524: your server accepted the request but took longer than Cloudflare's 125-second limit to answer. Find the slow page and the slow step in it. Cloudflare error 523: origin is unreachable Error 523: Cloudflare can't find a route to your server. Usually the IP address in your DNS records is wrong or out of date, or the host's network is down. Cloudflare 502 Bad Gateway A Cloudflare-branded 502 usually means your own server returned the 502 and Cloudflare passed it on. Fix the app behind your web server, not Cloudflare. Cloudflare Ray ID error The Ray ID isn't the error. It's a reference number for the request that failed. Use it with the error code to find that exact request in Cloudflare.

SSL and HTTPS

All 5 →

Browser warnings, padlock problems and certificate errors.

"Your connection is not private" The browser couldn't verify your site's certificate: it's expired, for the wrong domain, or not trusted. Fix or renew it on your server. SSL certificate expired Renew the certificate on your server or hosting panel, then fix whatever stopped the automatic renewal so it doesn't happen again in 90 days. HTTPS not working Usually no valid certificate, port 443 closed, or two redirect settings fighting. Check the certificate, then redirects and Cloudflare SSL mode. SSL handshake failed Your server and the browser or Cloudflare couldn't agree on a secure connection: no certificate, wrong SSL mode, or outdated TLS only. Mixed content error Your secure (HTTPS) page loads some files over insecure HTTP. Browsers block them or drop the padlock. Change those links to HTTPS.

Database

All 7 →

The site loads but can't reach its data.

Database connection error Your app can't reach its database: the database is stopped, the address or password is wrong, or a firewall is blocking it. Check them in that order. "Error establishing a database connection" WordPress can't reach its database. Usually the database details in wp-config.php are wrong, or the database server is down. Check those two first. "Too many connections" database error Every database connection is in use. Usually your app opens connections and doesn't close them, or traffic outgrew the limit. Fix leaks first. Database timeout A query or connection took longer than allowed: usually a slow query on a big table, a missing index, locked rows, or an overloaded database. MySQL error Read the error number: 1045 is access denied, 2002 means the server isn't reachable, 1040 is too many connections. The MySQL error log says the rest. Postgres error Read the message: connection refused, authentication failed, too many connections and statement timeout each point to a different fix. MongoDB connection error Usually your server's IP isn't on the database's access list, the connection string or password is wrong, or the cluster is paused. Check the access list first.

Server

All 7 →

The machine your site runs on is struggling, stopped, or full.

Server not responding Nothing answered: the server is off, overloaded, out of memory or disk, or a firewall is blocking it. Check it's running, then its resources, then the network. Server overloaded More work than the server can handle: a traffic spike, bots, a runaway process, or a slow query. Find what's using the CPU before upgrading. Out of memory: process killed The server ran out of memory and Linux killed a process to survive, often your app or database. Confirm it in the system log, then find what used the memory. Disk full With no disk space left, databases stop, logs can't be written and uploads fail. Find what filled it (usually logs, backups or cache) and clear it safely. Server crash Find out what crashed (the whole machine, or just your app or database), then read the logs from just before. Most crashes are memory, disk, or a code change. Nginx error Nginx's error log names the problem. 502 means the app behind Nginx is down; 504 means it's too slow; a failed reload means a config typo. Run nginx -t first. Apache not working Check the Apache error log first: it's where Apache records every problem. If Apache won't start, test the config with apachectl configtest.

WordPress

All 5 →

Blank pages, errors and crashes on WordPress sites.

WordPress white screen of death A blank white page means a PHP or database error WordPress didn't show. Turn on debug logging to see it, then disable plugins or the theme to find the cause. WordPress error Read the exact message: a critical error, a database error, maintenance mode, a 500, or a blank page each have their own fix. Here's how to tell them apart. WordPress 500 error On WordPress, a 500 error is most often a corrupted .htaccess file, then a plugin, theme or PHP memory limit. Rename .htaccess first; it's quick and safe. WordPress memory limit "Allowed memory size exhausted" means PHP hit its memory cap. Raise WP_MEMORY_LIMIT in wp-config.php; on shared hosting your host may need to raise it. WordPress plugin error Find the plugin, then deactivate it. If you can't log in, rename its folder in wp-content/plugins. Then update, replace or report it.

Shopify

All 5 →

When your Shopify store, checkout or theme stops working.

Is Shopify down? Check Shopify's status page first. If Shopify is fine, the problem is usually your domain, a theme change, or an app. Here's how to narrow it down. Shopify checkout not working Place a test order to see exactly where checkout fails. The usual causes are payment settings, shipping rates for the customer's address, an app, or theme code. Shopify Payments not working Check your payment settings for a verification request or alert. Missing documents, an unsupported business type or country, or test mode are common causes. Shopify cart error Cart errors usually come from theme code or an app that changes the cart, or from stock and product settings. Test with an unedited theme to find out which. Shopify theme error Usually a recent code edit, a theme update that dropped your changes, or an app's code. Go back to a working copy of the theme, then redo the change carefully.

WooCommerce

All 3 →

When your WooCommerce store or checkout breaks.

WooCommerce not working WooCommerce says most problems come from outdated software or a conflict with your theme or another plugin. Update, then conflict-test to find which. WooCommerce checkout broken Usually a plugin or theme conflict, page caching on the checkout, or payment and shipping settings. Run a test-mode order, then conflict-test. WooCommerce payment failed Find out whether the payment provider declined it or your site failed to send it. The order notes and the gateway's own dashboard say which.

Hosting

All 11 →

Problems that start with your hosting company or deployment platform.

Vercel deployment failed Open the build logs on the deployment's page: the first error line is the cause. Usually a build command failing, a missing environment variable, or bad config. Vercel error Vercel's error page shows a code like FUNCTION_INVOCATION_FAILED or DEPLOYMENT_NOT_FOUND. It says whether your function crashed, timed out, or is missing. Netlify build failed Read the deploy log from the first error. Usual causes: the build command, a missing environment variable, a different Node version, or the repo link. Netlify not working Check Netlify's status page, then whether the latest deploy succeeded. "Page not found" on refresh usually means a single-page app is missing its rewrite rule. Heroku error Heroku logs an error code with every failure: H10 app crashed, H12 request timeout, H14 no web dynos, R14 memory exceeded. Run heroku logs to see which. AWS error Check the AWS Health Dashboard first, then the service that failed. Most AWS outages for one site are security group rules, permissions, or a resource limit. GoDaddy website down Check whether it's the domain, the hosting, or the site itself. Error codes point to hosting or code; "can't be reached" points to the domain or DNS. Is DigitalOcean down? Check DigitalOcean's status page. If it's fine, your Droplet is the problem: use the Recovery Console to get in even when the network or SSH is down. Bluehost error Read the error first. On Bluehost shared hosting, most errors are a WordPress plugin, a broken .htaccess file, account resource limits, or the domain's DNS. SiteGround error SiteGround's own guide: check if the site loads from elsewhere, then read the error. 403 is permissions or .htaccess, 404 a missing file, 500 configuration. Is Hostinger down? Check Hostinger's status page. If Hostinger is fine, check your plan's resource usage, your domain's DNS, and your site's own errors, in that order.

Slow website

All 3 →

It works, but it's slow, or it gives up before the page arrives.

Why is my website slow? Find out which part is slow: the server producing the page, or the browser loading heavy files. Measure first, then fix the slowest part. Website timeout Something waited too long and gave up: the browser, a proxy such as Cloudflare, or your server. The error tells you which one, and so where the slowness is. Slow time to first byte (TTFB) TTFB is how long until the server starts answering. Over 0.8 seconds is slow. The fix is on the server side: caching, faster queries, or better hosting.

Domain and DNS

All 6 →

Your domain doesn't point where it should, or doesn't point anywhere.

Domain not working Check three things in order: the domain hasn't expired, its nameservers are the ones you expect, and its DNS records point at your current host. DNS not resolving: NXDOMAIN NXDOMAIN means DNS says the name doesn't exist at all. Usually a typo, an expired domain, a missing record for that subdomain, or broken nameservers. DNS error A DNS error means the name couldn't be turned into an address. Check whether it fails everywhere or only on your network, then check the domain's records. DNS propagation: how long it takes A record change reaches everyone within its TTL, often minutes to hours. A nameserver change can take up to 24 hours. Here's how to check and speed it up. Nameserver error The nameservers listed for your domain aren't answering correctly. Usually they don't match your DNS provider, or the provider doesn't have your domain set up. Dangling DNS record (subdomain takeover) A DNS record pointing at a service you no longer use can let someone else claim it and serve content on your subdomain. Delete it.

Security

All 3 →

Files and code on your site that anyone can read or abuse.

.env file exposed If yoursite.com/.env opens in a browser, your passwords and API keys are public. Block it, then change every secret in it. .git folder exposed If yoursite.com/.git/HEAD opens, your site's git repository is public. Block dot folders and check what the history contains. Outdated JavaScript library with known vulnerabilities Old versions of libraries like jQuery have published security holes. Find the version your pages load and update it.

Email

All 8 →

Your site's emails aren't arriving, or are bouncing back.

Emails not sending from my website Check whether the email left your site at all, then whether it was rejected or landed in spam. Most failures are missing SPF/DKIM or wrong sending settings. SMTP error SMTP errors start with a code: 4xx means temporary (try again later), 5xx means permanent (fix something first). The text after the code says what. Emails bouncing A bounce means the receiving server rejected the email. Read the bounce message: a bad address, a full mailbox, or an authentication or reputation problem. SendGrid error Check SendGrid's activity feed for the message, then the error. Common causes: an unverified sender identity, a wrong or revoked API key, or bounces. Mailgun error Most Mailgun problems come from an unverified domain (a daily limit and "sent via" labels), a wrong API key, or sandbox domain sending rules. SPF too many DNS lookups SPF allows at most 10 DNS lookups. Past that, the SPF check returns an error and your email can fail. Remove includes you no longer use. DMARC not set up Without a DMARC record, receiving servers aren't told what to do with mail that fails SPF or DKIM. Gmail requires it for bulk senders. DKIM not set up DKIM signs your emails so receivers can tell they really came from you. Turn it on in your email service and publish its DNS record.

API

All 6 →

The part of your app that talks to servers has stopped getting answers.

API not working Look at the status code and response body the API returns: 4xx means the request is wrong (login, data, limits); 5xx means the API's server failed. CORS error The browser blocked your page from reading another site's response because that site didn't allow your origin. Fix it on the server, not the front end. "Failed to fetch" / fetch failed "Failed to fetch" means no usable response arrived: a network error, a blocked request (often CORS or mixed content), or a bad URL. The console says which. API returning 500 A 500 from an API means its server code failed. If it's your API, the error is in your server logs. If it's someone else's, check their status page. API timeout The API took longer than your app or a proxy was willing to wait. Find whether it's the network, the API being slow, or one slow operation inside it. API rate limit You've sent more requests than the API allows in a time window. Slow down, respect Retry-After, cache and batch requests, or move to a higher plan.

JavaScript

All 6 →

Errors in the code that runs in your visitors' browsers.

JavaScript error Open the browser console (F12). The red error names the problem, the file and the line. That's the starting point for every JavaScript error. "Cannot read property of undefined" Code tried to read a property of something that's undefined or null, usually data that hasn't loaded yet or an element that isn't on the page. "undefined is not a function" Code tried to call something as a function that isn't one: a misspelled name, a method that doesn't exist on that kind of value, or a library that didn't load. Script not loading Check the browser's Network tab for the script: a 404 means a wrong path, a blocked request means a Content Security Policy, mixed content or an ad blocker. React error If a React app goes blank, an error during rendering removed the UI. The console shows which component failed. Error Boundaries stop one bug blanking it all. Vue error Read the console: Vue warnings name the component. Most Vue errors are reading data before it loads or a template referring to something undefined.

Broken features

All 8 →

The site loads, but one part of it doesn't work.

Login not working If the password is right but you're sent back to the login page, the session cookie isn't sticking: HTTPS, domain or caching. Otherwise, read the error. Checkout broken Place a test order and see where it stops. Failed payments are usually provider keys, test mode or webhooks; a frozen checkout is usually a script error. Form not submitting Watch the Network tab when you press submit: no request means a script problem; a failed request means the server or a spam filter; success means email. Signup not working Find where signup stops: the form, the account creation on the server, or the confirmation email. Each needs a different fix. Redirect loop Two settings keep sending visitors back and forth. Most often an HTTPS redirect on your server fighting Cloudflare's Flexible SSL, or www rules that disagree. Images not loading Check the image's address in the Network tab: 404 is a wrong path or missing file, 403 is permissions or hotlink protection, blocked is mixed content. File upload failing Most failed uploads hit a size limit: in the web server (Nginx returns 413), in PHP (upload_max_filesize, 2 MB by default), or in your app. Raise them together. Site search not working Check whether search returns an error, nothing at all, or old results. Errors point to the server or a search service; old results point to an index or cache.

Getting help

All 3 →

Something's wrong and you don't know where to start.

My website is broken. What do I do? Stay calm and work in order: confirm it's broken for everyone, write down the exact error and time, undo the last change, then check domain, hosting and logs. How to fix a website Name the exact symptom, find which layer it lives in (domain, hosting, code, outside service), then fix that layer. Here's how to work it out. How to debug a website Reproduce the problem, read the evidence (browser console, network requests, server logs), form one guess at a time, and test it. Change one thing at a time.

Website builders

All 7 →

Problems on hosted website builders, where you can't see the server.

Squarespace error Check Squarespace's status page first. If it's fine, the usual problem is a third-party domain whose DNS records don't match what Squarespace asks for. Wix not working Check Wix's status page for incidents, then whether your site is published and your domain is connected. Most Wix outages on one site are the domain. Webflow error Check Webflow's status page, then whether the latest changes were published to your custom domain, and whether the domain's DNS points to Webflow. Framer error Most Framer site problems are the custom domain: wrong or extra DNS records, AAAA records blocking SSL, or a Cloudflare proxy. Check DNS against Framer. Ghost error On self-hosted Ghost, run ghost doctor to check the setup, and ghost run to see the real error when Ghost won't start. On Ghost(Pro), check Ghost's status. Carrd not loading If your Carrd site loads on its carrd.co address but not your domain, the domain's DNS records or publishing are wrong. SSL only arrives once DNS is right. Notion site error Check the page is still published as a Notion Site and the custom domain add-on is still connected. Unpublishing doesn't always make a page private.

Headless CMS

All 4 →

When the system your content lives in stops serving it.

Strapi error Read the server log and the error's status code. If you log in to the admin panel and get sent straight back, the admin URL and cookie path don't match. Contentful error Check Contentful's status page, then the status code your site gets back: 401 is the access token, 404 the space or entry, 429 a rate limit. Sanity error If your site or Studio can't load Sanity data from a new address, add it to your project's allowed CORS origins. Otherwise check the token and dataset. Prismic error Check the access token if your repository is private, the repository name, and that content is published. Then read the status code from the failed request.

Frameworks

All 9 →

Errors from the framework your site or backend is built with.

Next.js error Find where it happens: at build time (read the build log), on the server (your host's function logs), or in the browser (the console). Each has its own fix. Node crash By default, an uncaught exception makes Node.js print the stack trace and exit. Read that trace, fix the error, and run Node under a process manager. Express error Express handles errors through middleware with four arguments (err, req, res, next). Make sure async errors reach it, and log the error before responding. Laravel error Read storage/logs/laravel.log for the full error. Keep APP_DEBUG false in production. Common causes: .env settings, cache, permissions, or a missed migration. Django error Keep DEBUG off in production and read your server logs for the traceback. A 400 Bad Request on every page usually means ALLOWED_HOSTS is missing your domain. Rails error Read the production log in your app's log folder for the exception. Common causes: pending migrations, missing credentials or environment variables, assets. Nuxt error Nuxt errors come from several places: Vue rendering, server and client startup, and the Nitro server. Find which one from the error page and logs. Remix error In Remix, a route's ErrorBoundary shows when its component, loader or action throws. Read the server log for loader and action errors. Gatsby build failed Read the first error in the build output; it names the file and step. Usual causes: data sourcing, gatsby-node.js code, a plugin, or the Node version.

Apps and backends

All 2 →

Your mobile app or backend has stopped working.

App crashing or not loading Is the app itself crashing (an unhandled error), or can't it reach its backend? Crash reports show the first; your API's status shows the second. Backend down: server unreachable Check whether the backend answers at all: no response points to the server, DNS or firewall; an error code points to the app or its database.

Can't find yours?

One click checks your site and gives you a report of what's wrong with yours, not just what usually goes wrong. Your first report is free.

Run your free report