graphhopper-selfhost
Installation
SKILL.md
Self-hosted OSM routing
Take a user from "I have a .pbf file" to "I can curl a route", and diagnose the three places this normally breaks: the import, the CH preparation, and the first request.
Ask these before touching anything
- Which engine? Stock GraphHopper (roads) or OpenRailRouting (railways)? They are different jars with different configs.
- How big is the pbf? Memories and import time scale with it.
- What is
-Xmxgoing to be? If the machine has 8 GB, a national extract will not fit. - Is the jar already built? If not, they need Maven and (for the web UI) Node.
Do not start an import that cannot finish. Ask first.
Step 1 — Get a jar
| Engine | Where | Notes |
|---|---|---|
| OpenRailRouting | github.com/geofabrik/OpenRailRouting releases, or build from source |
Apache-2.0. Build needs git submodule init && git submodule update && mvn clean install. Produces railway_routing-<version>.jar. |
| Stock GraphHopper | github.com/graphhopper/graphhopper releases |
graphhopper-web-<version>.jar. |