GridSuite local deployment

Local setup

Databases folders configuration

All data must be stored under a common root directory whose location is defined by the environment variable $GRIDSUITE_DATABASES

The following subdirectories must be created with file mode 777 (rwx) :

• cases : working directory for cases-server
• postgres : databases Postgres
• elasticsearch : indexes (documents) Elasticsearch
• init : data files for initialization

$ cd $GRIDSUITE_DATABASES
$ chmod 777 cases postgres elasticsearch init

⚠️ This environment variable must be set and subdirectories created before running any containers with docker-compose !

All databases are created automatically at start as well as the necessary initial data loading (geographical, cgmes boundaries, tsos, ...).

To do this, you must copy the following files in the init directory :

• geo_data_substations.json
• geo_data_lines.json
• business_processes.json
• tsos.json

Clone deployment repository

$ git clone https://github.com/gridsuite/deployment.git
$ cd deployment

Docker compose deployment

Important Docker Compose v2 is mandatory.
See instructions in sub-section

Application profiles (alias)

Note

If you want to use profiles other than the ones associated with the folders (dynamic-mapping, merging, study, suite, technical), you have to use profiles as explained in the next section.

$ cd docker-compose/suite
$ docker compose up

$ cd docker-compose/study
$ docker compose up

$ cd docker-compose/merging
$ docker compose up

$ cd docker-compose/dynamic-mapping
$ docker compose up

Notes : When using docker-compose for deployment, your machine is accessible from the containers thought the ip address 172.17.0.1

Notes : The containers are accessible from your machine thought the ip address 127.0.0.1 (localhost) or 172.17.0.1 and the corresponding port

Notes : These folders (other than explicit-profiles) act now like an alias to docker compose --project-name grid<name> --profile <folder_name> ..., with the difference that they have implicitly a profile active and will be considered like another project stack, so compose commands will not affect others folders state.

Docker-compose profiles

This is the preferred development deployment.
Everything described in this section is inside the folder explicit-profiles.

Here's the summary of the profiles and what services they includes:

Component \ Service	(none)	merging	study	study-light	dynamic-mapping	dynamic-simulation	suite	import	kibana	pgadmin
rabbitmq postgres elasticsearch logstash socat logspout	✅	✅	✅	✅	✅	✅	✅	✅	✅	✅
kibana									✅
pgadmin										✅
apps‑metadata‑server mock‑user‑service gateway actions‑server case‑server config‑notification‑server config‑server filter‑server loadflow‑server network‑conversion‑server network‑store‑server report‑server user‑admin‑server		✅	✅	✅	✅	✅	✅	✅
griddyna‑app dynamic‑mapping‑server					✅	✅	✅
gridmerge‑app balances‑adjustment‑server case‑import‑job case‑validation‑server cgmes‑assembling‑job cgmes‑boundary‑import‑job cgmes‑boundary‑server merge‑notification‑server merge‑orchestrator‑server		✅					✅
gridstudy‑app dynamic‑simulation‑server timeseries‑server			✅			✅	✅
cgmes‑gl‑server odre‑server security‑analysis‑server sensitivity‑analysis‑server shortcircuit‑server voltage‑init‑server			✅				✅
directory‑notification‑server directory‑server explore‑server geo‑data‑server gridexplore‑app network‑map‑server network‑modification‑server single‑line‑diagram‑server study‑notification‑server study‑server			✅	✅		✅	✅
case‑import‑server								✅

To use a profile, you use:

$ cd docker-compose/explicit-profiles
$ docker compose --profile suite <cmd>

You can also combine multiple profiles:

$ docker compose --profile study --profile dynamic-mapping <cmd>

Important

The --profile argument isn't always mandatory with some commands:

• With the commands up & down, services/container who belongs to at least one profile can't be accessed if the profile isn't specified. For example docker compose up study-server would not work because the profile study isn't passed in the CLI. The correct CLI would be docker compose --profile study up study-server.
• With the commands start, stop, restart, the --profile ... has no effect because theses commands affect the containers already created by a previous up command.

To change the running profile('s) without down&up everything (for example you have up the profile suite and you want now use the profile study), you can instead stop & up:

# previously: docker compose --profile suite up -d
$ docker compose --profile suite stop
$ docker compose --profile study up -d

In case you want to do a down for everything, an all profile exist to simplify :

$ docker compose --profile all down

If you still have a container existing in the project, you will have this message during the down:

$ docker compose down
[+] Running 2/1
 ✔ Container abc-server             Removed                      10.6s
 ! Network gridsuite_default        Resource is still in use     0.0s

Technical profile

This profile allows you to launch only the technical services : postgres, elasticsearch, rabbitmq, ...

Software	Version used
Postgres	13.4
RabbitMQ	latest
Elasticsearch	7.9.3

It is used for k8s deployment with Minikube.

$ cd docker-compose/technical
$ docker compose up

Update docker-compose images

To synchronize with the latest images for a docker-compose profile, you need to :

• delete the containers

$ docker compose down

• get latest images

$ docker compose pull

• recreate containers

$ docker compose up

• remove old images

$ docker image prune -f

Applications and Swagger URLs

You can now access to all applications and swagger UIs of the Spring services of the chosen profile:

Applications:

http://localhost:80 // gridexplore
http://localhost:81 // gridmerge
http://localhost:83 // griddyna
http://localhost:84 // gridstudy

Swagger UI:

http://localhost:5000/swagger-ui.html  // case-server
http://localhost:8095/swagger-ui.html  // cgmes-gl-server
http://localhost:8087/swagger-ui.html  // geo-data-server
http://localhost:5003/swagger-ui.html  // network-conversion-server
http://localhost:8080/swagger-ui.html  // network-store-server
http://localhost:5006/swagger-ui.html  // network-map-server
http://localhost:8090/swagger-ui.html  // odre-server
http://localhost:5005/swagger-ui.html  // single-line-diagram-server
http://localhost:5001/swagger-ui.html  // study-server
http://localhost:5007/swagger-ui.html  // network-modification-server
http://localhost:5008/swagger-ui.html  // loadflow-server
http://localhost:5020/swagger-ui.html  // merge-orchestrator-server
http://localhost:5021/swagger-ui.html  // cgmes-boundary-server
http://localhost:5022/swagger-ui.html  // actions-server
http://localhost:5023/swagger-ui.html  // security-analysis-server
http://localhost:5025/swagger-ui.html  // config-server
http://localhost:5026/swagger-ui.html  // directory-server
http://localhost:5028/swagger-ui.html  // report-server
http://localhost:5029/swagger-ui.html  // explore-server
http://localhost:5036/swagger-ui.html  // dynamic-mapping-server
http://localhost:5032/swagger-ui.html  // dynamic-simulation-server
http://localhost:5027/swagger-ui.html  // filter-server
http://localhost:5010/swagger-ui.html  // balances-adjustment-server
http://localhost:5011/swagger-ui.html  // case-validation-server
http://localhost:5033/swagger-ui.html  // user-admin-server
http://localhost:5030/swagger-ui.html  // sensitivity-analysis-server
http://localhost:5031/swagger-ui.html  // shortcircuit-server
http://localhost:5037/swagger-ui.html  // timeseries-server
http://localhost:5038/swagger-ui.html  // voltage-init-server
http://localhost:5039/swagger-ui.html  // case-import-server

RabbitMQ console

RabbitMQ management UI:

http://localhost:15672
default credentials :
   - username : guest
   - password : guest

PgAdmin for PostgreSQL administration

PgAdmin UI:

http://localhost:12080/login
default credentials :
   - username : [email protected]
   - password : admin

To connect to the PostgreSQL database, the postgres container must be up. Then, you can add a new server with the following configurations :

Host name/address : postgres
Port : 5432
Maintenance database : postgres
Username : postgres
Password : postgres

Kibana console for Elasticsearch

Kibana management UI:

http://localhost:5601

In order to show documents in the case-server index with Kibana, you must first create the index pattern ('Management' page) : case-server*

Installing / Updating docker-compose to v2

Docker compose v2 is necessary to be able to use this compose project which uses profiles feature.
If possible, prefer to install it with your package manager if you are on a Unix system.

Note

You need a client (docker-cli) of v19 v20 at least to have the system of cli-plugins. It isn't necessary to update your Docker engine or client else.

You can install Docker Compose v2 with these commands, as instructed in the doc and the migration guide:

curl -LR --create-dirs -o $HOME/.local/bin/docker-compose https://github.com/docker/compose/releases/latest/download/docker-compose-linux-x86_64
#wget -x -O $HOME/.local/bin/docker-compose https://github.com/docker/compose/releases/latest/download/docker-compose-linux-x86_64
chmod +x $HOME/.local/bin/docker-compose
mkdir -p $HOME/.docker/cli-plugins
ln -s $HOME/.local/bin/docker-compose $HOME/.docker/cli-plugins/docker-compose
docker compose version

You must get this output from docker compose now:

Docker Compose version v2.xx.x

Important

The commands shown will install the plugin user-side, so you don't need to remove your old docker-compose v1 if it is installed system-wide.

k8s deployment with Minikube

Minikube and kubectl setup

This setup is heavyweight and matches a realworld deployment. It is useful to reproduce realworld kubernetes effects and features. In most cases, the lighter docker-compose deployment is preferred.

Download the recommended version of minikube and kubectl :

Software	Version recommendation	Last Version	Link
kubectl	1.21+	1.24.X	Download
minikube	1.21+	1.26.X	Download

install minikube and kubectl following instructions for binaries download installation.

Notes: We require minikube 1.21+ for host.minikube.internal support inside containers (if you want to use an older minikube, replace host.minikube.internal with the IP of your host).

Start minikube and activate ingress support:

$ minikube start --memory 24g --cpus=4
$ minikube addons enable ingress

To specify the driver used by minikube and use specific version of kubernetes you could alternatively use :

$ minikube start --memory 24g --cpus=4 --driver=virtualbox --kubernetes-version=1.22.3
$ minikube addons enable ingress

Notes: With last version of minikube, docker is the default driver (was virtualbox before) which could forbid memory definition depending of your user privilegies.

see kubernetes-version param doc for versions support.

Verify everything is ok with:

$ minikube status
$ minikube kubectl cluster-info

Minikube deployment

Get you ingress ip

$ INGRESS_HOST=`minikube ip`
$ echo $INGRESS_HOST

Fill config files with the INGRESS_HOST in k8s/live/local/ :

$ find k8s/live/local/ -type f | xargs sed -i -e "s/<INGRESS_HOST>/${INGRESS_HOST}/g"

Optionally, give an ssh access to the case importing cronjobs by providing your username and your password (or create a dedicated user for this if you want): The import jobs connect through ssh to your machine and automatically import files from the $HOME/opde and $HOME/boundaries

$ find k8s/live/local/ -type f | xargs sed -i -e 's/<USERNAME>/YOURUSERNAME/g'
$ find k8s/live/local/ -type f | xargs sed -i -e 's/<PASSWORD>/YOURPASSWORD/g'

Start technical services with the docker-compose technical profile :

$ cd docker-compose/technical
$ docker compose up -d

Deploy k8s services:

$ kubectl apply -k k8s/live/local

Verify all services and pods have been correctly started:

$ kubectl get all

You can now access to the application and the swagger UI of all the Spring services:

Applications:

http://<INGRESS_HOST>/gridstudy/
http://<INGRESS_HOST>/gridmerge/
http://<INGRESS_HOST>/griddyna/
http://<INGRESS_HOST>/gridexplore/

Swagger UI:

http://<INGRESS_HOST>/case-server/swagger-ui.html
http://<INGRESS_HOST>/cgmes-gl-server/swagger-ui.html
http://<INGRESS_HOST>/geo-data-server/swagger-ui.html
http://<INGRESS_HOST>/network-conversion-server/swagger-ui.html
http://<INGRESS_HOST>/network-store-server/swagger-ui.html
http://<INGRESS_HOST>/network-map-server/swagger-ui.html
http://<INGRESS_HOST>/odre-server/swagger-ui.html
http://<INGRESS_HOST>/single-line-diagram-server/swagger-ui.html
http://<INGRESS_HOST>/study-server/swagger-ui.html
http://<INGRESS_HOST>/network-modification-server/swagger-ui.html
http://<INGRESS_HOST>/loadflow-server/swagger-ui.html
http://<INGRESS_HOST>/merge-orchestrator-server/swagger-ui.html
http://<INGRESS_HOST>/cgmes-boundary-server/swagger-ui.html
http://<INGRESS_HOST>/actions-server/swagger-ui.html
http://<INGRESS_HOST>/security-analysis-server/swagger-ui.html
http://<INGRESS_HOST>/config-server/swagger-ui.html
http://<INGRESS_HOST>/directory-server/swagger-ui.html
http://<INGRESS_HOST>/balances-adjustment-server/swagger-ui.html
http://<INGRESS_HOST>/case-validation-server/swagger-ui.html
http://<INGRESS_HOST>/dynamic-simulation-server/swagger-ui.html
http://<INGRESS_HOST>/filter-server/swagger-ui.html
http://<INGRESS_HOST>/report-server/swagger-ui.html
http://<INGRESS_HOST>/user-admin-server/swagger-ui.html
http://<INGRESS_HOST>/sensitivity-analysis-server/swagger-ui.html
http://<INGRESS_HOST>/shortcircuit-server/swagger-ui.html
http://<INGRESS_HOST>/timeseries-server/swagger-ui.html
http://<INGRESS_HOST>/voltage-init-server/swagger-ui.html
http://<INGRESS_HOST>/case-import-server/swagger-ui.html

How to use a local docker image into Minikube?

Build and load a local image into Minikube:

$ mvn clean install jib:dockerBuild -Djib.to.image=local/<pod>
$ minikube image load local/<pod>

Then add it to your deployment before (re)deploy:

$ vi <pod>-deployment.yaml
+  image: docker.io/local/<pod>:latest

Check the pod has started with your local image:

$ kubectl describe pod <pod-instance> | grep "Image:"
  Image:         docker.io/local/<pod>:latest

Multiple environments with customized prefixes

To deploy multiple environments we can use customized prefixed databases (Postgres), queues (rabbitMq) and indexes (elasticsearch).

You must follow those steps:

1. Edit the docker-compose/.env file and specify the prefix by defining the DATABASE_PREFIX_NAME property

DATABASE_PREFIX_NAME=dev_

2. Edit the k8s/base/config/common-application.yml file and specify the prefix by defining the environement property

powsybl-ws:
  environment: dev_

⚠️ do not forget to include underscore '_'

After this configuration :

• every services which use a Postgres database will call to dev_{dbName} database.
• every services which provide or read a rabbitMq queue will call to dev_{queueName} queue.
• every services which use a elasticsearch index will call to dev_{indexName} index.

Databases creation and data initialization

Considering databases are created automatically as well as the necessary initial data (geographical, cgmes boundaries, tsos, ...), the following part concerns only the databases recreation and/or the update of the initial data. All actions can be done from a docker-compose profile.

Databases creation

$ docker compose exec postgres /create-postgres-databases.sh

Data initialization

First update the data files in the directory $GRIDSUITE_DATABASES/init

$ docker compose exec postgres /init-geo-data.sh
$ docker compose exec postgres /init-merging-data.sh

Note: For RTE geographic data (lines and substations), alternately, you can use the odre-server swagger UI (see the URL above) to automaticaly download and import those data in your database. You have to execute those REST requests :

• .../substations
• .../lines

Be sure to have at least odre-server and geo-data-server containers running.

Working with Spring services

In order to use your own versions of Spring services with docker-compose, you have to generate your own Docker images (using jib:dockerBuild Maven goal) and modify the docker-compose.yml to use these images.

Docker image is generated using the following command in the considered service folder:

mvn jib:dockerBuild -Djib.to.image=<my_image_name>

Once the image has been generated, you have to modify the name of the image to use in docker-compose.yml file, for the considered service:

services:
...
  my-service:
    image: <my_image_name>:latest
...

Now, when using docker compose up, your custom Docker image will be used.