corpusops/setups.graphhopper

Python

Initialise your development environment

Idea is to have a pair of two graphhoppers (ying/yang) running the same configuration balanced using a loadbalancer.
Every once and then, a colocated daemon will detach one of the daemons and refresh the routing from the OSM last extract and then refresh the first (ying) instance upon termination.

changelog

20211214: upgrade graphhopper known version:
- from: https://github.com/graphhopper/graphhopper/commit/a8c654b05b2df1df0bf89f5dc2e17ee48bf2e1eb
- to:

First clone

git clone --recursive https://github.com/corpusops/setups.graphhopper.git

Install docker and docker compose

if you are under debian/ubuntu/mint/centos you can do the following:

.ansible/scripts/download_corpusops.sh
.ansible/scripts/setup_corpusops.sh
local/*/bin/cops_apply_role --become \
    local/*/*/corpusops.roles/services_virt_docker/role.yml

... or follow official procedures for docker and docker-compose.

Update corpusops

You may have to update corpusops time to time with

./control.sh up_corpusops

Doc for deployment on environments

See here

update cron pseudo code

GRAPHHOPPER_CACHE_BACKEND_EXPIRY=604800 # world (weekly)
GRAPHHOPPER_CACHE_BACKEND_EXPIRY=86400 # europe (daily)
GRAPHHOPPER_CACHE_BACKEND_EXPIRY=43200 # france (halfdaily)
On garde un fichier json minimaliste pour garder un etat des executions ancers from loadbalancers themselves

.ansible/scripts/call_ansible.sh -vvvvv .ansible/playbooks/lb.yml -e "{lb_servers: prod_loc-lb, graphhopper_servers: prod_loc-graphhopper}"

Switcher docs

wellknown playbooks

setup frontal playbook

cd /srv/graphhopper
.ansible/scripts/call_ansible.sh -vvvvv .ansible/playbooks/lb.yml \
 -e "{lb_servers: prod_loc-lb, graphhopper_servers: prod_loc-graphhopper}"

setup a specific region graphhopper instance

cd /srv/graphhopper
.ansible/scripts/call_ansible.sh -vvvvv .ansible/playbooks/graphhopper.yml \
 -e "{graphhopper_servers: prod_loc-fr}"