Docker orchestration for UAL GBL (Geoblacklight) development

Overview

This Docker orchestration runs a fully decoupled GeoBlacklight search service that uses a Blacklight (Ruby on Rails) application querying the Solr.

How does it work? 30K ft view...

Here's a little diagram of the GBL and Solr interaction (Mermaid syntax):

  graph LR;


      A[Geoblacklight]==>|QueryRequest|B{Solr};
      B{Solr}==>|DataResponse|A[Geoblacklight];

The GBL app container queries the Solr instance directly. Using the "blacklight-core" metadata, Solr sends back its data response in a format that can be ingested by RoR models.

Setup

Necessary tools:

Terminal access
Git
Docker (either Desktop or just the straight engine if deploying on any of the Linux distros)
Lando
Stache access (for deployments)

Preliminary Steps:

Start by cloning this repository to pretty much anywhere on most filesystems that the Docker daemon has access to.

Build the UAL-GBL Docker images (for GBL and Solr containers):

$ lando start

Re-build image and containers while preserving volumes

$ lando rebuild

Ingest UAL metadata records from ual-geospatial-metadata

$ lando rake ual_docs:reindex

Note: If you're using Docker Desktop on a Mac, and encounter the following error: ERROR ==> connect ENOENT /var/run/docker.sock, try toggling the checkbox for Allow the default Docker socket to be used (requires password) and re-running lando start.

Deployment

We use Capistrano for deployments. Deployments are easiest to run through the Docker container for local development. You must provide the Slack webhook using the SLACK_WEB_HOOK environment variable in the .env file in order to deploy. You won't be able to deploy without it. You can find the webhook stored in Stache under the "Capistrano Slack App Webhook" entry. You must also provide the path to your SSH key in the DEPLOY_SSH_KEY_PATH environment variable in .env. It must be an RSA or ED25519 key. Lastly, the deployment is executed as the 'deploy' user, so you must have your SSH public key added to the deploy users authorized_keys file.

Here are the steps to deploy to production:

Make sure you're on the Library network (on site or using VPN)
Run the Capistrano deployment command
```
lando cap production deploy
```
This will deploy the main branch to production. Alternatively you can provide a branch name to deploy:
```
lando cap production deploy BRANCH=some_branch_name
```

You can check the status of the deployment in the #tess-dev-deployer Slack channel in the UAL Slack workspace.

Optional application container commands

Stop the Docker network:

This is non-destructive. All containers remain stateful, as well as volumes and network.

$ lando stop

Run Rake commands in the containerized application directory:

$ lando rake "<command-to-run>"

# View all available rake tasks
$ lando rake --tasks

# Populate default Solr test fixtures:
$ lando rake "geoblacklight:index:seed[:remote]"

# Clear current Solr index:
$ lando rake "ual_docs:clear_index"

# Ingest UAL test docs:
$ lando rake "ual_docs:load_test_docs"

# Clear current Solr index and ingest all UAL GeoBlacklight metadata from Github:
$ lando rake "ual_docs:reindex"

See Geoblacklight tasks here.

:geoblacklight main namespace for all commands.
- :server Solr & Geoblacklight startup (which this repo doesn't use). This also installs some seed data.
- :webpack Runs the app and Solr together in interactive mode with the foreman gem.
- :index child name for running fixtures.
  - :seed by itself runs local fixtures (spec/fixtures/solr_documents).
    - [:remote] reaches out to the github repo (see below) to get fixtures.
  - :ingest_all Ingests a GeoHydra transformed.json.
  - :ingest default to ingest blacklight.json files from directory data in project root.
    - [:directory] to pass a directory path to ingest files.
- :downloads stuff to do with downloading files.
  - :delete deletes cached downloads.
  - :mkdir create the cache directory, like bash.
  - :precache programmatically add cache files.
- :solr:seed same as geoblacklight:index:seed
- :application_asset_paths echoes out all asset paths, kinda handy.

Inspect GBL application logs for development

# number of lines to watch can be set
$ lando logs

Run Solr queries for development and testing

$ ./make-solr-query.sh <query-something-neat>

# Example (after running seed ingest) - look for geo data containing "Manhattan":
$ ./make-solr-query.sh "q=Manhattan"

Also see the Solr query screen in the Solr admin: https://solr.apache.org/guide/solr/latest/query-guide/query-screen.html. Scroll to the main, lefthand menu. Click on the collection, "blacklight-core". Then scroll down to the query button.

Tear-down

WARNING: This destroys all data, meaning containers and volumes. (It does not remove Docker images, however.)

$ lando destroy

This does not remove images. If you want to remove images you can run docker image prune -a wich will remove all unused images.

To also remove any app related files that are not code (tmp directory files, SQLite file, etc) you can run:

./rm_app_files.sh

Re-indexing metadata

We keep all the GeoBlacklight metadata docs in the ual-geospatial-metadata repository (private repo currently). We have a rake task for clearing the current core docs and reingesting the docs from the ual-geospatial-metadata repository.

Reindexing locally

lando rake "ual_docs:reindex"

Reindexing in production

You can use capistrano to run the same rake task in production:

lando cap production invoke:rake TASK=ual_docs:reindex

Notes

Helpful hints

Blacklight-core metadata and config are stored in the solr/conf directory, which is mounted into the Solr container.

Rails Console

To access rails console run lando rails c to fire up the console

Helpful Rails Console Commands

Command	Purpose
`reload!`	Refresh the build (clear cache)

ualibraries/ual-geoblacklight