I have a pretty powerful machine (24 CPUs, 192GB RAM, 1TB SSD) on which I want to run ORS with a planet file.
My ORS instance is currently building the graphs… but it seems it’s barely using any CPU. RAM usage is ok (I’ve modified the JAVA OPTS), but my load average is currently 2. I’ve tried setting “ParallelGCThreads=18” to see if it’d use more CPU, but not really.
Is there any way to speed up the building of the graphs since I still have so much RAM and CPU idling?
If you have enabled elevation, it might be possible that ORS is busy with downloading the SRTM height data. For planet, this is quite some data and unfortunately takes time. In your ors-config.json you can look up, which directory you have configured for caching the elevation data. If this directory is slowly filling with tiles, you are probably in this phase. If this is the case, you might try whether downloading the SRTM data manually and placing them in the cache-folder is quicker. Or, if you don’t need elevation, you can switch it off in the config.