If the site stops opening shortly after an update, record the publication time, the affected URL, and the response you actually get. This will help the developer check whether the change is related to the error. The timing alone does not prove that the new version is at fault.
First, check the current status as described in the first-response steps for an outage. For further investigation, it is important to find out which version is on the primary domain and what else other people changed at the same time.
Confirm the publication time on the live site
Ask the developer: “When did the change become available to visitors, and how can we identify that version?” The time development finished or files were uploaded may differ from the actual publication time.
Record a clear label, for example, “new menu, published 04.10 at 09:40 MSK.” If the developer uses a version number, include it. The site owner does not need to study all the code, but it is difficult to compare different versions when they are all referred to as “the latest update.”
Ask about any other changes made in parallel. Hosting settings, site addresses, or an external service may also have been changed that day. Record only known actions and who performed them. If the time is approximate, note a time range rather than inventing an exact minute.
Create a brief timeline
Here is an example of the information to send the developer. It is illustrative; replace the times and responses with your own.
| Time, MSK | Fact | Source |
|---|---|---|
| 09:35 | /catalog responds | reChecker check |
| 09:40 | New menu published | Developer’s message |
| 09:45 | /catalog does not respond | reChecker check |
| 09:48 | Catalog did not open; homepage works | Owner’s check |
| 10:10 | Catalog opens again | Owner’s check after the fix |
This timeline shows that the problem was detected after publication. It does not show when the problem began or why. In “Site monitoring”, availability is checked on a five-minute schedule; the status may have changed between checks.
For technical investigation, ask the developer to check the server log—logs—for the specified time range. If the server records time in a different time zone, the developer needs to account for the difference. Keep the reChecker event time and the log time with their source labels.
Compare the affected URL with working URLs
Check the homepage, the original URL, and a similar page. For example, the new catalog does not open, but an older product page does. This narrows down the issue: the specific section needs investigation, rather than assuming the entire server is down.
If the error occurs during an action—such as selecting a product option or submitting a form—record the steps and conditions. A page loading successfully does not confirm that the action works. Save the exact error text and describe what should happen instead.
Send the developer the timeline, URLs, and description. Ask them to identify a confirmed cause or the next way to investigate. “Check whether the new version is related to this response” is more useful than asking them to confirm a culprit you have already chosen. The article on checking a site after an update covers how to check a published site.
After reverting to the previous version, separate two tasks
The decision to revert to the previous version should be made by the person who understands how the site works. After the reversion, first check that the site is restored: test the original URL and the action that produced the error. Do not rely only on the homepage working.
Then clarify what will happen to the new change. Reverting to the old menu may restore the catalog, but the new menu task is still incomplete. Record “site restored” and “update needs to be fixed and published again” separately.
Example summary: “At 10:00, we restored the previous menu; at 10:10, the catalog opens on the primary domain. The developer is checking the connection between the error and the new menu using the logs. Before publishing again, we will check the catalog and a product page with options.” If the cause has not been established, leave it unresolved. For the next release, use the same URLs and conditions under which the outage occurred.