I help support a large company that uses OSM data, pulled via a specific
query at xxxx://www.overpass-api.de/api/xapi
<http://www.overpass-api.de/api/xapi>. This query runs about once a month.
thank you for asking back.
First about the status: when I last checked some days ago then there
have been still almost a million requests for /xapi . This is in
contrast to 500.000 requests for all other API endpoints combined.
Yesterday figures are 1067494 for /xapi and 937498 for all other API
The bigger surprise came when I have taken a sample of 100 requests from
xapi calls. I turned out that while they are from various apps still all
100 of them were less desirable bulk load. At that point I decided to
wait until somebody comes up with a real request to get the xapi call
back. This has happened right now.
For that reason I have configured the service now as follows:
- xapi works again
- xapi?map now returns 400 Bad request as I have turned off that in the
I think that a longtime solution will be to take the referrer and user
agent into account when scheduling requests. But this has not been
I would like to conclude with some remarks to hedge rumors on this thread:
The OSMF is not in any way involved in operations. Paul's mail describes
the precise situation. Please note that the whole purpose of third party
services (like Overpass API) for secondary functionality is to keep that
burden off the OSMF.
Nothing is held confidentially on purpose. There are two limitations
that might give this impression:
- I'm on vacation right now and until SotM and relatively unwillig to
answer questions immediately.
- The log files may contain personal data and thus I won't hand them out
talk mailing list