Skip to content

Latest commit

 

History

History
147 lines (119 loc) · 9.7 KB

README.md

File metadata and controls

147 lines (119 loc) · 9.7 KB
Raw

CbTSP-r2r-Docker

Repository for CbTSP docker stack

Updates

01/02/2022 - Externalized cbtsp_director from docker image, and adjusted stack_pass.yml accordingly, fixed a bug in cbtsp_director (string instead of boolean in a pojo)
11/02/2022 - Fixed wrong naming of docker images in "stack_pass.yml" it was referencing a local naming, instead of the github docker registry naming convention
22/02/2022 - Added documentation about GeoJson and PBF files
02/03/2022 - Fixed a bug on Drivers API endpoint (delete and getByEmail not working properly)
14/03/2022 - Fixed an issue where asking for a non-existing driver would result in an internal server error instead of a 404 not found error
14/03/2022 - Fixed an issue where in the intermediate list of points of a lift, the last one was missing
18/03/2022 - Fixed configuration issue as requested by OLTISgroup, added route replan functionality to support triptracking
25/03/2022 - Fixed an issue where asking for a non-existing lift, ride or car would result in an internal server error instead of a 404 not found error
10/05/2022 - Added implementation of integration with Triptracker (POLIMI), implemented functionality for signaling to stop tracking on ride cancellation
18/05/2022 - Updated logic on the "search for lift" operation, now all the possible results are included
03/06/2022 - Fixed an issue regarding Lift search when Departure and Arrival points where inside the walk radius
30/06/2022 - Fixed a bug in the PATCH driver API call

Hello Ride2Rail Docker

This repository contains script files, and external data folders and files for the docker stack of the CbTSP.

cbtsp_planner. cbtsp_backend. cbtsp_director. mongo:3.6.8.

Project Contents

CbTSP is composed by more than one component. It includes:

  1. A relational database
  2. A backend for anagraphics and plan caching
  3. A planner component which is responsible for the effective route calculation and plans (rides) storing and retrieving
  4. An orchestrator which ties all togheter and exposes a set of API to interact with.

These components are delivered as docker images as agreed with partners, they are mostly self contained with exception to some externalized folders and files, which are either configuration files, or persistance folders mapped to their respective containers

Prerequisites

  • A docker engine
  • Docker compose

Install and first run

$ docker login ghcr.io -u USERNAME

  • Clone this repository in your local working directory, you will need all the files of this repo to correctly run the CbTSP stack

  • Pull the images:

$ docker pull ghcr.io/ride2rail/cbtsp_planner:latest
$ docker pull ghcr.io/ride2rail/cbtsp_backend:latest
$ docker pull ghcr.io/ride2rail/cbtsp_director:latest
$ docker pull ghcr.io/ride2rail/mongo:3.6.8

At this point you should have on your local repository, the images listed above. Next steps are:

  • Clone this github repository in your working directory and move into it. 
    cd
git clone https://github.com/Ride2Rail/CbTSP-r2r-Docker.git
  • Make sure that all the scripts (.sh) are set as executables (you probably will not need all of them) 
    chmod +x ./cbtsp-run.sh
chmod +x ./cleanup.sh
  • You can directly bring online the stack by issuing the command: 
    $ docker-compose --env-file ./compose_env.env -f stack_pass.yml up
    if you want to run the stack attached, or: 
    $ docker-compose --env-file ./compose_env.env -f stack_pass.yml up -d
    for detached mode.

Working directory content and meaning

Once you have cloned this repository in your working directory you should find these folders:

  • FOLDER: planner-data - It contains the padova.pbf which is the openstreetmap file from which, upon first run, the planner builds its internal graph, upon first run, this folder will contain also the persisted planner graph object. For other demos other than "Padova", you must change the padova.pbf file with the one of your interest, before the first run

  • FOLDER: director-data - It contains SpringBoot application.properties and the directore executable (.jar), those files should not be altered unless you are 100% sure of what you are doing, .properties basically contains all the configuration parameters of the orchestrator, including other api's endpoints, ip addresses, ports, usernames, passwords and so on. It's externalized so that if something changes down the line, it may just require editing this file and relaunching the container instead of redeployng a new one. BEWARE THAT: the content of this file may be correlated to other configuration files listed, so in case of any planned change, you should check the correlation on theese other files also

  • FOLDER: mongo-data - IT GETS CREATED ON FIRST RUN it contains the persistance data for the mongo-db container

  • FILE: compose_env.env - This files contains some enviroment variables that are used by the docker compose when launching, those environment variables are common to use for all the containers (it mainly includes subnet definition for the docker network that gets build up during deploy) BEWARE THAT: the content of this file may be correlated to other configuration files listed, so in case of any planned change, you should check the correlation on theese other files also

  • FILE: mongo-init.js - This files contains first run configuration for the mongo-db container like users, passwords, initial db creation and grants BEWARE THAT: the content of this file may be correlated to other configuration files listed, so in case of any planned change, you should check the correlation on theese other files also

  • FILE: stack_pass.yml - It's the main entrypoint for docker-compose it contains the definition of the services, their order of startup, theri mapping to the host system and so on

First and subsequent runs

First time you issue the docker compose command, many things happens:

  • Containers gets created
  • A specific subnetwork in the docker space gets created
  • needed ports of every container gets exposed (in the meaning of docker syntax, which implies that some ports are seen in the docker sub-network but not from the host, i.e the mondo-db port
  • The port for the entrypoint (cbtsp-orchestrator) gets published instead to the host so to be reachable (default API entrypoint: 8888)
  • configuration files per specific container are read and used, this implies:
    1. mongodb-init
    2. orchestrator application properties
    3. planner initialization (creation of the graph object from the .pbf file
    4. planner launch after initialization

Note: first run generally requires more time to startup than subsequent one

Note: some exceptions on the planner launch may happen, this is totally normal and does not implies generally a problem with it, it's just set to be very verbose at the moment

Note: the orchestrator is wrapped in a safety script that blocks it from starting untill everything else is up and running, you should see in the log a message telling you that the launch is delayed more than one times until it's finally safe to start it, once you see the springboot logo in the logs, you should be ready to go

Operating the stack

Generally speaking you should only need to use the proper docker-compose command to start and stop the whole stack, and everything should work fine, and data persist trhough restarts

To start:

	$ docker-compose --env-file ./compose_env.env -f stack_pass.yml up
	or
	$ docker-compose --env-file ./compose_env.env -f stack_pass.yml up -d

To stop:

	$ docker-compose --env-file ./compose_env.env -f stack_pass.yml stop

CAVEATS: If for any reason, you need to stop only the cbtsp-planner container, BEWARE THAT planning data is persisted in the backend and not in the planner which operates in memory, the component responsible for checking consistency of the planner is the cbtsp-orchestrator at its stratup, so if you manually restart the cbtsp-planner, remember to restart AFTER THAT also the cbtsp-planner In any case, after a stack restart you should see the last component starting (the orchestrator) checking for plans to restore into the planner memory, you should see in the logs how many plans need to be reinserted into the planner and should be able to follow throug while the orchestrator proceeds to ask the planner to recreate all the cached plans

You want the clean slate?

In case you want /need to start fresh, there's a handy script that you should find in your working directory if you cloned this reposistory, it's name is:

	$ sudo ./cleanup.sh

This script is very basic and you should be able to simply inspect it to understand what it does, but if for any reason you want to do a manual cleanup, this is the correct procedure:

  1. Stop all the cbts related running containers and delete them and their associated volumes
  2. Delete the docker network associated with the stack (the name may vary depending on your working dir name)
  3. Completely delte the mongo-data folder (you might need administrative credentials) 
    $ sudo rm -rf ./mongo-data
  4. Delete ONLY the Graph.obj file in the planner-data folder Beare not to delete the file named {City}.pbf
  5. Issue the docker compose up command as instructed in the first run chapter.

TO BE UPDATED...