Introduction
A classic frustrating situation: the website is live on the server, the admin sees a normal response, but some users say that everything is red and won’t load.
What went wrong
The issue didn't seem like a typical website bug. Everything might have looked fine based on server and local checks, but users on other networks were experiencing timeouts, TLS errors, or strange HTTP responses. So, we needed to separate three areas: the website itself, DNS/SSL, and regional availability.
Diagnostics
- HTTP status codes were checked for the list of URLs;
- Redirects and TLS have been verified;
- IPv4 connectivity has been verified;
- parallel curl checks were performed;
- Behavior has been tested via proxies/mirrors;
- It has been separately verified that the origin server is up and running.
What has been done
- A list of URLs to monitor has been compiled;
- a parallel status check was performed;
- Added output of codes and response times;
- Proxies and mirrors have been checked;
- The following error codes are listed, indicating the type of issue: 200/301/403/499/500/502/504;
- A plan for further monitoring has been prepared.
Result
Instead of just thinking, “The site isn’t working sometimes,” we now have a list of symptoms. This makes it clear whether the problem lies with the server, DNS/SSL, routing, a proxy, or specific ISPs.
What to monitor next
In such cases, you need to monitor the site from multiple locations, not just run a single check from your work laptop. A single green "curl" result doesn't prove that the site is accessible to all customers.