Timeout when get history

[LesBoys43]

URL

https://www.openstreetmap.org/history#map=17/31.345927/121.440822

How to reproduce the issue?

Just open it

Screenshot(s) or anything else?

Always throws an error: 抱歉，您所请求的变更集列表检索时间过长。

[mmd-osm]

It looks like the db server was maxed out on CPU at the time this issue was reported.

https://prometheus.openstreetmap.org/d/Ea3IUVtMz/host-overview?orgId=1&from=2026-02-11T02:57:19.965Z&to=2026-02-11T08:45:30.700Z&timezone=utc&var-instance=snap-01&refresh=1m

[tomhughes]

Yes but that was a symptom rather than the cause - the cause as usual was AI scrapers overwhelming the site.

[mmd-osm]

That's what I thought. Is there some more info available which queries are consuming lots of CPU? I've read somewhere that the nodes/ways/relation pages are frequently scraped. If that's the case here, we could think about showing a limited amount of detail for unauthenticated users. As an example, showing tags for a way only and skipping all node details.

[tomhughes]

Honestly trying to accommodate them is not going to achieve anything - we could support 10x or 100x the query load and they'd just soak it up.

[LesBoys43]

So may we add a CAPTCHA, AI really crazy

[mmd-osm]

Maybe that's finally a good use case for read replicas (openstreetmap/openstreetmap-website#2634), where we're sending requests for unauthenticated users bots to a read replica, to keep the load low on the central database for (authenticated) mappers. So the goal is not accomodate bots, but making everyone else's experience better.

I believe this could be done on a case by case basis, possibly starting with tracepoints and the node/way/relation pages, or whatever else happens to be the pain of the day.