OSRM Docker seems to fail periodically from too many requests

[radumas]

Similar to #4723, but doesn't seem reproducible. I just pulled the latest image with sudo docker pull osrm/osrm-backend:latest.

We set up the docker image on a Ubuntu 16.04 LTS VPS (on Amazon EC2). It has 8GB of RAM and the following processors:

*-cpu:0
          description: CPU
          product: Intel(R) Xeon(R) CPU E5-2676 v3 @ 2.40GHz
          bus info: cpu@0
          slot: CPU 1
     *-cpu:1
          description: CPU
          bus info: cpu@1
          slot: CPU 2

We set up the docker container similar to the quickstart guide for Toronto, Canada, using this .lua file to get only transit directions. We're trying to process transit GPS data and the script makes requests for directions for a batch of GPS points in order to snap them to transit stops. Even making requests in a single thread, the process will make about 10 requests or so and the OSRM just stops responding...

Happy to provide more info, not sure what else is helpful for debugging.

[danpat]

What's CPU usage like inside the container? After requests stop being answered, is the CPU idle or busy?

Does it change if you run osrm-routed --threads 1 to limit concurrency?

[radumas]

mildly embarrassing but how I check cpu usage inside container? sudo docker attach just attaches me to see the output of osrm-backend

[radumas]

Hmm... I think this might be related to #4825 somehow, since this problem doesn't appear to occur with v.5.14.3 which I just downgraded to. Will investigate further tomorrow

[danpat]

If you encounter the problem in #4825, you will fail to create a dataset, and you will get no routing results at all - osrm-routed will fail to start.

[radumas]

I will play with different recent versions tomorrow. We definitely get some routing results, but then osrm just appears to stop responding.

[f3d0r]

I'm having a similar issue with this, any updates here?

[NumberPiOso]

I am having the same problem.

I have the set up in ubuntu 18.04 inside WSL 2.

The CPU idles and osrm-routed --threads 1 did not work.

[ogtayhuseynov0]

I am having the same problem.

updates?

[holdmygithub]

I have also faced a similar issue. OSRM stops responding when bombarded with calls. Manually setting to sleep for some time between the calls seems to do the job (50 ms pause works in my case)

[hoerup]

@holdmygithub could you try with latest code and if that helps? 56a427f

[SiarheiFedartsou]

Stale.