When a timeout occurs, record the URL, exact time, and error, then find out at what stage the request ended. The page may take a long time to find the server, establish a connection, or wait for application data. “The catalog is slow” isn’t enough to determine whether the issue is with the hosting, CMS, or database.
A timeout means the client you’re using didn’t receive a result within its wait limit. Sometimes the HTTP response has already started; sometimes no status is received at all.
Measure the request and save the error
For a single public URL, a developer can run:
curl -sS --max-time 30 -o /dev/null \
-w 'DNS=%{time_namelookup} TCP=%{time_connect} TLS=%{time_appconnect} FIRST=%{time_starttransfer} TOTAL=%{time_total} HTTP=%{http_code}\n' \
'https://example.com/catalog/'Replace the domain with your own. The limit of 30 seconds is chosen for this example; it is not a website speed requirement. Save the output line and the curl message if the request was interrupted. HTTP=000 means the tool didn’t receive an HTTP status, not that the server returned a special response code.
The values are in seconds from the start of the request. The parameters are described in the curl documentation. For a clean analysis, start with the direct URL without redirects: redirects add stages and make comparisons harder.
Identify where the time was spent
| Field | What it shows |
|---|---|
| DNS | When the server address lookup finished |
| TCP | When the connection was established |
| TLS | When the secure HTTPS connection was established |
| FIRST | When the response started arriving |
| TOTAL | When the entire request finished |
For example, the results were DNS=0,02, TCP=0,05, TLS=0,10, FIRST=1,80, and TOTAL=1,90. About 1,70 seconds passed between establishing HTTPS and the start of the response. Loading the rest took about 0,10 seconds. These intervals are calculated by subtracting adjacent timestamps, not by adding all the values.
A long wait for the first byte doesn’t necessarily mean slow SQL. The application may be waiting on a queue, an external service, or the database. You need server logs for that specific request to determine the cause.
Compare the catalog with a simple page
Check a known simple page on the same domain and the problematic catalog using the same client. Make several separate requests with pauses between them; don’t run a load test on a site that is experiencing an issue.
If it takes a long time to establish a connection to both URLs, start by checking the connection and infrastructure. If they connect quickly but only the catalog waits for a response, send the catalog route to the developer. They can investigate how the page is processed and what it depends on.
Say whether the issue happens every time, only after a refresh, or at certain hours. A normal response now doesn’t rule out an outage a minute ago. The time of the original error and the data saved then are important.
Send your hosting provider a message with the measurements
Suppose the catalog timed out before the first response arrived. You can format the message like this:
https://example.com/catalog/ 4 October at 14:32 MSK, the request timed out after 30 seconds; no HTTP status was received. The HTTPS connection was established in 0,10 seconds. The simple /contacts/ page responded normally in the same minute. Please check the application logs, request queue, and resource limits for the catalog at that time.
If you have a request ID, include it. Send the error text and results; don’t start by demanding that they “fix the SQL.” Your hosting provider should identify where the delay occurred; the developer will need their own measurements for the database or external API.
Repeat the check after the fix
Request the same URL using the same method and compare the stage that was previously slow. A longer timeout lets you wait longer, but doesn’t make the page faster by itself. Save the reason given by support and the result of the follow-up check.
For your next technical report, use the website audit. For an explanation of server responses, see the HTTP reference; for server settings, see the Nginx guide.