osm-server

The repository serves OSM points, lines, and polygons using a flask app. Use this repo if you need higher bandwidth access to OSM data than the overpass enpoints can provide.

Setup

Clone the repository, submitting username and password when prompted:

git clone https://github.com/Lkruitwagen/osm-labelcube-generator.git

Create a new environment

We recommend using conda for package and environment management. Open a command line shell. Create a new environment osm-server:

conda create -n osm-server python=3.6

Activate the environment:

conda activate osm-server

Install git and pip if not already installed:

conda install git pip

Change directory into the repository:

cd osm-server

Install remaining package requirements using pip:

pip install -r requirements.txt

Clone OSM

Follow the tile-server osm instructions here. I found it was very unclear how much memory was needed to clone the entire planet. After exploding a 256gb memory machine with the planet-latest.osm.pbf file I changed tactics to creating 8 different databases, each serving one continent's worth of osm data. The continent extracts can be downloaded here.

I also provisioned a large SDD to hold all the data. I moved my postgresql database to the new disk using these instructions.

Following the switch2osm instructions, create a database for each continent:

CREATE DATABASE <continent>;
CREATE EXTENSION postgis;
CREATE EXTENSION hstore;

When you load in the data it may complain that it can't find hstore. I solved this by changing the hstore schema to public:

ALTER EXTENSION hstore SET SCHEMA "public";

I also had to fiddle extensively with my postgres settings. For reference, I've put the .conf file in this repo: [/postgres.conf]. You can use if it you change the data directory in line 41 and copy it to /etc/postgres/12/main/postgresql.conf.

For each continent extract, load the data in the osm using the data loader shell script, [/data_load.sh], adjusting paths, etc. as necessary:

sh ./data_load.sh

Iterating though the continents, I found that the ~10gb osm.pbg files took about 10 hours to load with osm2pgsql and expanded in many times in size. Loading all the continents took about four days with 64gb RAM and 16 vCPUs.

We'll only use the planet-osm-point, planet-osm-polygon, and planet-osm-line tables after this process. Any node.cache files can be deleted.

Set up the Flask app

I serve osm data in json format using a Flask app with Nginx and Gunicorn. Follow these instructions here, with the modifications below. The DigitalOcean guides are also very helpful.

To use your conda environment in your flask app, you'll need to specify the conda bin paths, and add any environment variables. Create a .env file in this repo:

touch .env
nano .env

Then add the environment variables that you need:

PATH="<path/to/your/conda>/bin:<path/to/your/conda>/envs/osm-server/bin"
DB_URI="postgresql://<your_psqluser>:<your_psqlpassword>@localhost/"
BASIC_AUTH_USERNAME="<a_basicauth_user>"
BASIC_AUTH_PASSWORD="<a_basicauth_password>"

In /etc/systemd/system/<yourapp>.service you'll need to adapt the gunicorn configuration:

[Unit]
Description=Gunicorn instance to serve osm-server flask app
After=network.target
[Service]
User=ubuntu
Group=www-data
WorkingDirectory=<path/to/repo/root>
EnvironmentFile=<path/to/repo/root>.env
ExecStart=<path/to/your/conda>/envs/osm/bin/gunicorn --bind 0.0.0.0:5000 osm$
[Install]
WantedBy=multi-user.target

In order to get Nginx to be able to listen at port 80, I also had to disable apache:

sudo systemctl disable apache2

Congrats! Your flask app is ready to serve you OSM features!

Useage

Let's use Python, requests, geojson, shapely to access some OSM features using our flask app.

import requests, geojson, json
from shapely import geometry, ops

# make a point
carfax = geometry.Point(-1.258305, 51.751906)

# buffer the point to get a box
bbox = geometry.box(*carfax.buffer(0.02).bounds)

# Make it a geojson Feature with the continent name in the properties
ft = geojson.Feature(geometry=bbox, properties={'continent':'europe'})

# set your basicauth username and password
U, P = "<your_basicauth_username>", "<your_basicauth_password>"

# prepare the query, specifying geom_type
url = "http://<your_IP>:5000/query"  
body = {'feature':json.dumps(ft), 'geom_type':'polygons'} # where geometry_type is one of points, lines, or polygons

# execute the query
response = requests.post(url, body, auth=(U,P))
osm_features = json.loads(response)['features']

Lkruitwagen / deepsentinel-osm