Errors (that apparently shouldn't happen) are happening in graphhopper.sh measurement

Hi,

I’m running the measurement tool over graphhopper data that I imported.

My import was using these config.properties:

graph.flag_encoders=car|turn_costs=true
prepare.ch.weightings=no
prepare.lm.weightings=fastest
prepare.min_network_size=200
prepare.min_one_way_network_size=200
routing.non_ch.max_waypoint_distance = 1000000
graph.dataaccess=RAM_STORE

When I first ran the measurement tool, I got a Java Exception because the graphhopper.sh measurement tool was overwriting my config.properties graph.flag_encoders setting, and setting it to “car” instead of “car|turn_costs=true”. I removed all the overwritten configuration properties in the graphhopper.sh file, so that it would work from my config.properties file, and so it now would run!

JAVA_OPTS="-Xmx30000m -Xms30000m -server" ./graphhopper.sh measurement /path/to/map.osm.pbf

However, I’m getting a lot of errors, perhaps about 20 per second. They look like this:

2017-12-20 09:59:48,141 [main] ERROR com.graphhopper.tools.Measurement$8 - errors should NOT happen in Measurement! -43.589193603154875,172.38357415162565; 52.121747146287866,-3.5735710391455378 => [com.graphhopper.util.exceptions.PointDistanceExceededException: Point 1 is too far from Point 0: 52.121747146287866,-3.5735710391455378]

The message says that “errors should NOT happen in Measurement!” and so I’m now concerned that my imported data is corrupted somehow.

Is that the case? Is there any clue I can be given about what to do to prevent this corruption?

Thanks
Luke

If you set the limit

routing.non_ch.max_waypoint_distance = 1000000

Then you’ll get errors in measurement if there are routes longer than this. This is kind of expected and data is ok IMO

Thank you so much for your quick reply!

I’ve just had that setting explained in this post so that makes sense to me.

Cheers,
Luke