IBF-system

Status

Component	Build Status
Cypress Integration Tests

Introduction

NOTE: For now some background on IBF-terminology (e.g. triggers) is expected. This can be expanded on later.

This is the repository for the IBF-system. It includes 3 main components.

1. Trigger model development

• This contains the exploratory analysis for developing a trigger-model for a given country and disaster-type.
• It might include (in the future) a lot of shared code between countries and disaster types, and even (automated) tools to aid analysts to develop trigger models.
• The output is a trigger script which determines (per country/disaster type) when and where a trigger is reached.

2. Services (backend)

• The trigger script is subsequently automated through (e.g. a daily running) service.
• Results (as well as other related data) are stored in a database
• Database content is returned through API-calls to some interface

3. Interfaces (frontend)

• Visualization of model results through dashboards
• Dashboards might move from read-only to write-applications, where users can also add (secondary) data through an interface

System design (draft)

Installation

1. Setup env variables:

   cp example.env .env

   Fill in the .env variables with someone who has them.

2. Whitelist your machine IP at the database server

Using Docker

docker-compose -f docker-compose.yml up # for production

docker-compose up # for development

docker-compose -f docker-compose.yml -f docker-compose.override.yml up # for development (explicit)

Without Docker (for local development)

For local development you can also run and start the services and interface without docker:

cp .env services/API-service/.env npm run start

NOTE on local database

Locally, a database-container will start (as opposed to remote servers, which are connected to a database-server). To currently fill this database with data

• run seed script
  • docker-compose exec ibf-api-service npm run seed
  • the 2nd half of the IBF-database-scripts.sql will fail, because of missing IBF-pipeline-output tables
• run the pipeline for all countries
  • docker-compose exec ibf-pipeline python3 runPipeline.py
• run the seed-script again

Migration to seed-scripts is now finished.

• Any new static data needs to be imported using a seed-script + corresponding TypeORM entity
• This includes e.g. geojson data
• The only exception are raster-files, which need to be included in data.zip and transfered to all relevant servers.

Installation result

These commands will install the IBF-system with listeners at,

1. localhost for the web server
2. *development only - localhost:4200 for the web interface

Integration tests

Cypress

We use Cypress for automated integration testing in this project. Installation:

1. sudo apt-get install libgtk2.0-0 libgtk-3-0 libgbm-dev libnotify-dev libgconf-2-4 libnss3 libxss1 libasound2 libxtst6 xauth xvfb
2. In root folder npm install --only=dev
3. Run npm run start:cypress

Releases

See notable changes and the currently release version in the CHANGELOG.

Release Checklist

• Check if the latest integration tests passed on Cypress Dashboard.
• Pick a tag to release. Let's say we want to release the tag v0.27.9 on GitHub.
• Click the 'Edit Tag' button.
• Enter the release title to v0.27.9.
• Optional: Enter the release description from the CHANGELOG file.
• Click the 'Publish Release' button.

The above steps should trigger the release webhook which updates the staging environment to the published release. This takes a while (approx 20 mins) to update.

Deployment

To "test" environment

• Make sure to verify if the environment-settings are appropriately set on the test VM before merging the PR.
• Merged PR's to 'master' branch are automatically deployed to the test-server. (via webhook, see: /tools#GitHub-webhook)

To "stage" environment

• Make sure to verify if the environment-settings are appropriately set on the stage VM before publishing the release.
• When a release is published, it is automatically deployed to the staging-server.

To "production" environment

• Make sure to verify if the environment variables are appropriately set on the VM.
• Manually run the deploy script with the tag which should be deployed for the specific country.

Adding a new country

For adding a new country to the IBF-system, a lot of components are already generic, and thus automized. But also quit some manual steps are needed at the moment. This is intended to be improved in the future. The list below is intended to give a full overview. It is not however meant to be detailed enough to execute each step, without further knowledge. Ask a developer who knows more.

Adding country with disaster type Floods

1. IBF-API-Service
   • Users:
     • Add user for country to src/scripts/users.json
     • Add country to admin-user
   • Country
     • Add country in src/scripts/countries.json
   • Upload through 'npm run seed' from API-service
2. Data for database (look at existing countries and files for examples in terms of format)
   • Save admin-area-boundary file (.shp) for agreed upon admin-level as geojson (with extension .json) with the right column names in services/API-service/src/scripts/git-lfs/
   • Save Glofas_stations_locations_with_trigger_levels_<country_code>.csv in the same folder
   • Save Glofas_station_per_admin_area_<country_code>.csv in the same folder
     • which (e.g.) 'districts' are triggered if station X is triggered?
     • note: this should include all admin-areas. If not mapped to any station, use 'no_station'
   • Potentially add extra code in seed-scripts (seed-amin-area.ts / seed-glofas-station.ts / etc.) to process new data correctly.
   • Run seed script of IBF-API-service
   • NOTE: we are in a migration, where we want to move new data as much as possible to this new seed-script set up. So also for other data, not mentioned here, the goal is to upload this via seed-scripts as well. Some other data that is not yet included in seed-script
     • COVID risk data (.csv) > uploaded through specifically created endpoint
3. Geodata for IBF-pipline and IBF-geoserver (look at existing countries and files for examples in terms of format)
   • Save in services/IBF-pipeline/pipeline/data in the right subfolder ..
     • Flood extent raster (for at least 1 return period) + an 'empty' raster of the same exten/pixel size. (.tif)
     • Population (.tif)
     • Grassland + cropland (.tif)
   • When deploying to other environments (local/remote) this data needs to be transfered (e.g. as data.zip through WinSCP or similar)
4. IBF-pipeline
   • add country_code to .env (for development settings, replace by ONLY that code)
   • add country-specific settings to settings.py (e.g. right links to abovementioned data)
     • with model = 'glofas'
   • add country-specific settings to secrets.py
   • add dummy-trigger-station to glofasdata.py with forecast-value that exceeds trigger-value
   • Run runPipeline.py (python3 runPipeline.py) to test pipeline.
5. Geoserver
   • Manually create new stores+layers in Geoserver interface of test-vm
     • flood_extent__<country_code> for each lead-time
     • population_<country_code>
     • grassland_<country_code>
     • cropland_<country_code>
   • Test that the specifics layers are viewable in the dashboard now
   • When done, commit the (automatically) generated content in IBF-pipeline/geoserver-workspaces to Github
   • This will prevent you from having to do the same for another server, or if your content is lost somehow
6. IBF-dashboard
   • Test dashboard by logging in through admin-user or country-specific user
7. Specifics/Extras
   • Whatsapp:
     • create whatsapp group
     • paste link in IBF-pipeline/pipeline/lib/notifications/formatInfo.py
   • EAP-link
     • create bookmark in Google Docs at place where Trigger Model section starts
     • paste link (incl bookmark) in countries seed-script
     • paste link (excl bookmark) in IBF-pipeline/pipeline/lib/notifications/formatInfo.py
   • Logo's
     • Get logo(s) (.png)
     • Paste in IBF-dashboard/app/assets/logos + add reference to each logo in countries seed-script
     • Paste in IBF-pipeline/pipeline/lib/notifications/logos/email-logo-<country_code>.png
     • Upload logo to mailchimp + retrieve shareable link + copy this in IBF-pipeline/pipeline/lib/notifications/formatInfo.py
   • Mailchimp segment
     • Add new tag '<country_code>' to at least 1 user
     • Create new segment '<country_code>' defined as users with tag '<country_code>'.
     • Get segmentId of new segment
     • Paste this in IBF-pipeline/pipeline/secrets.py
   • EAP-actions
     • Summarize actions from EAP-document + define 1 Area of Focus per EAP-action
     • Add to API-service/seed-data/EAP-actions.json
     • run 'npm run seed' from API-service

Adding country with disaster type Heavy rainfall

• Follow the 'flood' manual above as much as possible, with notable exceptions
• Input data database
  • Rainfall_station_locations_with_trigger_levels.csv > currently not included in seed-script yet, but manually uploaded (through runSetup.py)
• Input dat pipeline
  • There is no equivalent input to the flood extent raster. This is created in the pipeline.
• Add country in IBF-pipeline settings.py with model = 'rainfall'
• Save geoserver output as rainfall_extent__

Glossary

Term	Definition (we use)
version	A 'number' specified in the SemVer-format: 0.1.0
tag	A specific commit or point-in-time on the git-timeline; named after a version, i.e. v0.1.0
release	A fixed 'state of the code-base', published on GitHub
deployment	An action performed to get (released) code running on an environment
environment	A machine that can run code (with specified settings); i.e. a server or VM, or your local machine