Deploying OLS to a server

[hkir-dev]

Describe the bug

I have a set of application specific ontologies and I want to setup a local instance of the OLS 4 to explore those ontologies.

I deployed OLS to a server following the docker compose based steps. Deployment completed successfully but frontend is trying to access Solr from localhost not through my server ip.

To Reproduce Steps to reproduce the behavior:

1. On a server (172.27.20.150) completed the docker compose based OLS installation. All OLS modules installed to the same server using the docker compose up command.
2. In the frontend make a search
3. UI made request to http://localhost:8080/api/v2/entities?search=VAp&size=5&lang=en&exactMatch=false&includeObsoleteEntities=false however Solr is deployed at 172.27.20.150 and I can see a meaningful response at http://172.27.20.150:8080/api/v2/entities?search=VAp&size=5&lang=en&exactMatch=false&includeObsoleteEntities=false

Additionally, I have a set of ontologies to deploy to local OLS instance. I added all of them to the dataload/configs folder. However, only the ontology I provided with parameter export OLS4_CONFIG=./dataload/configs/my_ont1.json is available for search. Is there a way to add multiple ontologies or load all ontologies in the dataload/configs folder?

[haideriqbal]

did you start your backend as well in docker? from the SS it appears that frontend can't communicate with the backend...

[hkir-dev]

Yes, I just run the following commands in the server:

git clone https://github.com/EBISPOT/ols4.git
cd  ols4
// manually added some configs here
export OLS4_CONFIG=./dataload/configs/my_ont1.json
JAVA_OPTS="-Xms5G -Xmx25G"  docker compose up

[haideriqbal]

okay! you can try by changing the env value for API URL in this file https://github.com/EBISPOT/ols4/blob/dev/frontend/.env#L4 to your corresponding link instead of localhost as it tries to connect to localhost whereas your backend is using your server IP and rerun the docker... i suggest using following docker command:

docker compose up --force-recreate --build --always-recreate-deps --attach-dependencies ols4-solr ols4-neo4j ols4-backend ols4-frontend

[hkir-dev]

Great, thanks. Do you have any suggestions for this part:

Additionally, I have a set of ontologies to deploy to local OLS instance. I added all of them to the dataload/configs folder. However, only the ontology I provided with parameter export OLS4_CONFIG=./dataload/configs/my_ont1.json is available for search. Is there a way to add multiple ontologies or load all ontologies in the dataload/configs folder?

[haideriqbal]

You can load multiple ontologies by defining them in a single json file something like this:

{
  "ontologies": [
      {
          "id": "upheno",
          "ontology_purl": "https://github.com/obophenotype/upheno-dev/releases/latest/download/upheno.owl"
      },
      {
          "id": "uniprotrdfs",
          "ontology_purl": "https://ftp.uniprot.org/pub/databases/uniprot/current_release/rdf/core.owl"
      },
      {
          "id": "hra",
          "ontology_purl": "https://cdn.humanatlas.io/digital-objects/collection/hra-ols/latest/graph.ttl",
                  "hierarchical_property": [
              "http://purl.org/ccf/ccf_part_of"
              ],
                  "label_property": [
              "http://purl.org/ccf/ccf_pref_label"
              ]
      },
      {
          "id": "dcat",
          "ontology_purl": "https://www.w3.org/ns/dcat.ttl"
      },
      {
          "id": "biolink",
          "ontology_purl": "https://github.com/biolink/biolink-model/raw/master/project/owl/biolink_model.owl.ttl",
          "homepage": "https://biolink.github.io/biolink-model/"
      }
]
}

and then export this file as OLS4_CONFIG