-
Notifications
You must be signed in to change notification settings - Fork 3.5k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
osrm-routed consistent Segementation fault on specific requests #653
Comments
A couple of things were fixed in the mean-time. Is this issue still reproducable in the develop branch on your end? Also, please note that builds are done in cmake now instead of scons. |
Hi, I can confirm that it works for me now with latest develop branch. Dennis |
Ok, cool. Thanks for the feedback. |
I am getting this same problem on an Amazon EC2 Ubuntu 13 instance (m2.4xlarge, over 64GB memory). Downloaded source today and north_america_latest from Geofabrik. I've had this problem for over a week trying both north america and the planet, and have not been able to resolve it. Here's a simple run - it works for the first query (which is about 5 blocks in Denver, CO) while it segfaults on the second (which goes from north to south in Denver): [server] running and waiting for requests I followed the basic build instructions in the wiki. Any ideas? |
Hi, |
or if you have ruby installed, you can just use 'rake' which will make the build dir if needed, and build inside it (like suggested above) |
should this be reopened? |
Hi, it works fine, at least for me. I thought rake was replaced by cmake? |
I used mkdir -p build; cd build; cmake ..; make Extract and prepare ran without issue. When you say it works for you, is this on an Amazon EC2? If so what are the machine specs? Thanks for the help! |
no, scons was replaced by cmake. |
@emiltin is right, scons was replaced. rebuilding the source is not necessary in general, but AWS/EC2 has a number of oddities. E.g. you have no control on what physical hardware your virtual machine runs. When compiling, the compiler applies certain optimizations that are CPU specific, i.e. it uses the features that are advertised by the virtual CPU. It appears that this is very dependent on what kind of real hardware you are running. When you downsize/reboot/start your VM, it may very well be started on different hardware. So the previous compiler optimizations become invalid and thus stuff may not work anymore. |
@roblysam: are you on develop branch? This bug is not fixed on master branch yet... |
@drieks I assume I'm on the dev branch, I used the following to get the code: git clone https://github.com/DennisOSRM/Project-OSRM.git |
you must use this clone command: you can switch to develop branch using this command: |
(you are on master, check "git status") |
or |
thanks I will try that |
Does the new build system offer the option to disable architecture specific optimisations as scons did? This may help with some of the issues on EC2. Additionally is there any timetable for including this bugfix in the master branch? |
I have the same problem as Roblysam 5 months ago. [server] running and waiting for requests I use OSRM 3.5 and have 64GB ram, so I do not think thats the problem. What's the solution ? |
When running osrm-routed on Ubuntu 12.04 LTS on an Amazon EC2 instance some requests are consistently causing segmentation faults. For example:
This is using the europe-latest.osm.pbf form GeoFabrik and the latest source code from the Git repository as in the build instructions.
I compiled with the --no-march flag.
The text was updated successfully, but these errors were encountered: