For context, due to a maven error on deployement breaking my docker stack running ors 6.7 I upgraded everything to 7.1.
Queries are now much longer, matrices that took a second now require several minutes. And I had to significantly increase the maximum_visited_nodes in ors-config to run those. I have 30G of ram allocated of java heap space.
Could you help me diagnose & fix this ? the logs don’t show anything.
Also weird is the fact that with the exact same docker configuration, v6.7 ran just fine. Response times where close to the ones on the public API.
This somehow rules out the “system resources” and areas/profile options, as we did not change those while upgrading.
We did rebuild the graphs though, but the source pbf and graph configuration is the same as before (I believe, @vchalmel can confirm this).