Website down: find where the outage comes from and get the site back online
A website outage can show up in several ways: the browser cannot find the address, the page spins forever then times out, or the server responds but with a 502, 503 or 504 error. Each symptom points to a different layer: DNS, domain name, hosting, server or application.
Downtime costs money every minute: lost sales, unreceived forms, damaged image and, if it lasts, lower search rankings. It is therefore essential to locate the failure quickly rather than restarting things at random.
This page gives you a layer-by-layer diagnostic method, with the commands and tools for each check, the most frequent causes and the steps to restore service, on your own or with the help of a professional.
Typical symptoms
- "This site can't be reached" or "server IP address could not be found" (DNS_PROBE_FINISHED_NXDOMAIN): the domain name does not resolve.
- "Connection timed out" (ERR_CONNECTION_TIMED_OUT): the server does not respond at all.
- "Connection refused" (ERR_CONNECTION_REFUSED): the machine responds but no web service is listening.
- 502 Bad Gateway, 503 Service Unavailable or 504 Gateway Timeout: the web server is running but the application behind it does not respond.
- The site shows a parking page from the host or registrar, or an "expired domain" page.
- The site works for some visitors and not others, or on mobile but not on desktop.
Possible causes
Expired domain or changed DNS
The renewal was not paid, the domain is in its redemption period, or the name servers were changed (migration, mistake, compromised registrar account).
Wrong or propagating DNS zone
A or AAAA record pointing to an old IP address, broken CNAME, long TTL prolonging a recent change, or misconfigured DNSSEC.
Suspended hosting
Unpaid invoice, quota exceeded, spam sending or malware detected: the host cuts the account and shows an information page.
Server down or overloaded
Machine stopped, full disk, exhausted memory, denial-of-service attack, or incident at the infrastructure provider.
Web service or application stopped
Nginx, Apache, PHP-FPM, Node.js or the database are no longer running, often after a reboot, an update or a crash. The server responds but returns 502 or refuses the connection.
Expired SSL certificate or blocking firewall
An expired certificate makes the site unreachable over HTTPS for most browsers; a firewall, a WAF or misconfigured anti-DDoS protection can block legitimate visitors.
Application error
Failed deployment, wrong configuration, unfinished database migration: the infrastructure works but the application returns an error or a blank page.
Checks to perform
- 1
Confirm the outage is not local
Test from another network (mobile data) and with an external service such as downforeveryoneorjustme.com or a multi-location testing tool. If the site works elsewhere, the problem comes from your network, your DNS cache or a block on your IP address.
- 2
Check the domain and DNS
whois your-domain.com shows the expiry date and domain status. dig your-domain.com A or nslookup your-domain.com shows which IP address the name points to; compare it with your server's. dig NS your-domain.com confirms the name servers.
- 3
Test the connection to the server
ping ip-address checks that the machine responds; curl -I https://your-domain.com shows the HTTP code and headers, or the exact error (connection refused, timeout, certificate).
- 4
Check the host's status page
Most hosts and cloud providers publish a status page and send incident emails. Also check your customer area: suspension, invoice, security alert.
- 5
Check the services on the server
Over SSH: systemctl status nginx (or apache2), php8.2-fpm, mysql, plus df -h and free -m. A stopped service restarts with systemctl restart service-name.
- 6
Read the logs
/var/log/nginx/error.log or /var/log/apache2/error.log for 502/503/504 errors, the application log (Laravel, Symfony, Node, WordPress debug.log) for application errors, journalctl -xe for system incidents.
- 7
Check the certificate
openssl s_client -connect your-domain.com:443 -servername your-domain.com shows the certificate validity. A browser also shows the exact error when you click the padlock.
Solutions
Renew or recover the domain
Pay the renewal at the registrar, restore the correct name servers and secure the account (password, two-factor authentication, transfer lock).
Fix the DNS zone
Update the A, AAAA, CNAME and MX records to point to the right infrastructure, with a reduced TTL while the fix is in progress.
Lift the hosting suspension
Pay the invoice or address the cause (spam, malware, overage) then request reactivation. If the site was hacked, clean it before putting it back online.
Restart and repair the server
Restart the stopped services, free up disk space, identify the process eating memory, reboot the machine if needed, then address the root cause.
Fix the application
Roll back to the previous deployment, fix the configuration or complete the database migration. A deployment process with rollback capability prevents this kind of outage.
Set up monitoring
A monitoring tool (UptimeRobot, Better Stack, self-hosted Uptime Kuma) alerts you as soon as the site stops responding and watches domain and certificate expiry.
When should you call a professional?
- The site is in production and generates revenue or leads: recovery must be fast and clean.
- You do not know which layer the failure belongs to, or you lack the access (registrar, host, SSH).
- Restarted services go down again, a sign of an underlying cause (disk, memory, attack, bug).
- The outage follows a migration, a DNS change or a deployment whose details you do not fully control.
- You suspect a denial-of-service attack or a server compromise.
How Agencei can help
- 1
Locating the failure
DNS, network, HTTP and certificate tests from several points, server connection, log review: we identify the layer at fault within minutes.
- 2
Getting back online
Action on the faulty layer: DNS fix, service restart, resource clean-up, application rollback or a temporary page if the repair takes time.
- 3
Fixing the cause
Addressing the underlying problem to prevent recurrence: disk, memory, configuration, application bug, security.
- 4
Full verification
Check of all key site features, redirects, HTTPS, emails and Google indexing.
- 5
Monitoring and prevention
Availability alerts, backups and a continuity plan matched to how important the site is.
Frequently asked questions
The site works for me but not for my customers, why?
Usually because of DNS caching: your computer or your ISP still holds the old IP address. This happens after a migration or a DNS change. A test from several locations confirms the situation.
What is the difference between a 502, 503 and 504 error?
A 502 means the web server received an invalid response from the application (often PHP-FPM or Node stopped). A 503 indicates an unavailable service or maintenance. A 504 means the application did not respond in time, often due to overload.
How long does DNS propagation take?
It depends on the records' TTL, often from a few minutes to a few hours, rarely more than 48 hours. Lowering the TTL before a change limits the delay.
My domain has expired, can I get it back?
In most cases yes, during a grace period and then a redemption period defined by the registry, with additional fees. After that, the domain becomes available to anyone.
How can I be warned before this happens again?
External availability monitoring, alerts on domain and certificate expiry, disk and memory monitoring on the server, and regularly tested backups.
Related services
IT Support
Application and server support, incident handling and clear service commitments.
See this serviceIT Maintenance
Monitoring, updates, backups and continuous improvement of your applications in production.
See this serviceServer Migration
Full transfer of a dedicated server or VPS to a new machine, with OS upgrade and a controlled cutover.
See this serviceSecurity
Hacked site cleanup, hardening, vulnerability assessment, WAF, backups and access management.
See this serviceTell us about your project
Describe your need in a few lines: we come back to you with a first analysis and the next steps.