You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
However, osrm-extract seems to stop on the generating edge-expanded edges step:
...
[extractor] writing street name index ... ok, after 0.537011s
[info] Writing turn lane masks...
[info] done (0.000127)
[info] extraction finished after 4976.04s
[info] Generating edge-expanded graph representation
[info] - 341223 restrictions.
[info] Importing n = 382120223 nodes
[info] - 1094928 bollard nodes, 294824 traffic lights
[info] and 402322465 edges
[info] Graph loaded ok and has 402322465 edges
. 10% . 20% . 30% . 40% . 50% . 60% . 70% . 80% . 90% . 100%
[info] Node compression ratio: 0.17176
[info] Edge compression ratio: 0.213348
. 10% . 20% . 30% . 40% . 50% . 60% . 70% . 80% . 90% . 100%
[info] Generated 402321631 nodes in edge-expanded graph
[info] generating edge-expanded edges
It's been like this for several hours now, usually it completes within one to two hours. It also doesn't seem to do much: load avergae is 1.00, only one CPU seems to be used.
This happens with a current master build (commit aa5f8dc), stxxl v1.3.1 on a CentOS 7.2.1511 box.
This was fine with the build we used before this one (master d3a6b5a)
Looks related to turn analysis being stuck in a loop. @MoKob can you have a look on Monday?
@rkcpi what you already can do is attach gdb to the running process and manually do some sampling: break and note the function you're in. I guess this works best with debug symbols:
cmake .. -DCMAKE_BUILD_TYPE=Debug
This way you should be able to localize where the loop is happening exactly.
We just merged a fix that should, hopefully solve the hang you are expecting. Can you report back and close here, if the newest master solves your problem?
We're currently trying to use osrm-extract on the following map: http://download.geofabrik.net/europe-latest.osm.pbf.
However, osrm-extract seems to stop on the
generating edge-expanded edges
step:It's been like this for several hours now, usually it completes within one to two hours. It also doesn't seem to do much: load avergae is 1.00, only one CPU seems to be used.
This happens with a current master build (commit aa5f8dc), stxxl v1.3.1 on a CentOS 7.2.1511 box.
This was fine with the build we used before this one (master d3a6b5a)
Might be related to #2852?
The text was updated successfully, but these errors were encountered: