
Report repair work on the server
Update Habr, which took place last night, prompted to write a short note. During technical works, Habr hangs out a one-page plug, the text on which says about the ongoing work. A stub is sent to all the requested addresses. No redirect: at what address of the article do not go - everywhere the same text about the repair. In this case, the server response is accompanied by the status "HTTP / 1.1 200 OK". Most sites I know do. And if a person, by and large, doesn’t care, then the search engine indexing the site at that moment sees that the content has been updated at the address of the article - the index needs to be updated.
The solution was invented before us and has been standardized for a long time - this is an answer with code 503 , which speaks about the temporary unavailability of the server. If you know what time it will take to work, then this is also correctly reported using the Retry-After header .
Thus, from the point of view of http, hourly downtime should be reported to customers as follows: The search engine when you visit the site will understand that it’s better not to disturb it yet, but you can go in after an hour.
These are all modal windows that the programmers came up with:
[some garbage happened] - [OK] - Yes, this is not a damn OK!
@mad_escape
The solution was invented before us and has been standardized for a long time - this is an answer with code 503 , which speaks about the temporary unavailability of the server. If you know what time it will take to work, then this is also correctly reported using the Retry-After header .
Thus, from the point of view of http, hourly downtime should be reported to customers as follows: The search engine when you visit the site will understand that it’s better not to disturb it yet, but you can go in after an hour.
HTTP/1.1 503 Service Unavailable
Retry-After: 3600