drolbr
Server load
Hello Team!
Thank you for your hard work and dedication! Having overpass api in open access is like having superpowers!
I recently released this website that renders roads from OSM: https://twitter.com/anvaka/status/1218971717734789120 and it received unusual amount of attention (~70K visitors over past few days).
Coincidentally I noticed that I'm getting timeouts from Overpass API more frequently than I used to.
I just wanted to check that I'm not impacting you negatively with my traffic. Is there still more room to accept even more requests?
I'm hitting the following endpoints:
- https://overpass.kumi.systems/api/interpreter
- https://overpass-api.de/api/interpreter
- https://overpass.openstreetmap.ru/cgi/interpreter
Is there a way to check queue size, cpu load, network, number of timeouts, etc. on those backends?
Just adding a few remarks... In general, it's always a good idea to check the Usage policy for the server resources you're using beforehand: https://wiki.openstreetmap.org/wiki/Overpass_API#Public_Overpass_API_instances
Servers are typically funded by donations, membership fees, etc. (e.g. by FOSSGIS e.V.) and we expect you to be nice and adhere to the respective usage policies outlined on the wiki page. As the code is open source, you're very welcome to run your own instance. You can of course also donate here: https://www.fossgis.de/verein/spenden/
I should note that overpass.kumi.systems/api/interpreter is operated by a third party, i.e. you need to ask them in case of any issues with their service.
Just a small nitpick re: https://twitter.com/anvaka/status/1216285791346802688 The name of the backend is in fact Overpass API, not overpass turbo api. overpass turbo is the name of a very popular web frontend for Overpass API. There's really no such thing as overpass turbo API.
Website for Overpass API: https://overpass-api.de
The name of the backend is in fact Overpass API, not overpass turbo api.
Yes! My bad. This should be fixed now. Please let me know if you notice this anywhere else!
you're very welcome to run your own instance
I'm curious, how much does it cost to run an instance?
check the Usage policy for the server resources you're using beforehand
Totally agree. I read that wiki page a few times. From a client perspective - is there a json api similar to https://overpass-api.de/api/status that would let me check before fetching?
A suitable machine to run your own instance on starts at ~ 65 EUR per month, e.g. https://www.hetzner.de/dedicated-rootserver/ex52-nvme. (Not affiliated with Hetzner but using them myself - among other things for running a commercial Overpass service.)
