Add postgres service for local development, configure Rails to use it (ARC-181)

.github/workflows/ci.yml

@@ -105,6 +105,18 @@ jobs:
     #    ports:
     #      - 6379:6379
     #    options: --health-cmd "redis-cli ping" --health-interval 10s --health-timeout 5s --health-retries 5
+    services:
+      postgres:
+        image: postgres
+        env:
+          POSTGRES_DB: boxwalker
+          POSTGRES_USER: myuser
+          POSTGRES_PASSWORD: secret
+        options: >-
+          --health-cmd pg_isready
+          --health-interval 10s
+          --health-timeout 5s
+          --health-retries 5
     steps:
       - name: Install packages
         run: sudo apt-get update && sudo apt-get install --no-install-recommends -y libvips node-gyp
@@ -133,6 +145,18 @@ jobs:
     #    ports:
     #      - 6379:6379
     #    options: --health-cmd "redis-cli ping" --health-interval 10s --health-timeout 5s --health-retries 5
+    services:
+      postgres:
+        image: postgres
+        env:
+          POSTGRES_DB: boxwalker
+          POSTGRES_USER: myuser
+          POSTGRES_PASSWORD: secret
+        options: >-
+          --health-cmd pg_isready
+          --health-interval 10s
+          --health-timeout 5s
+          --health-retries 5
     steps:
       - name: Install packages
         run: sudo apt-get update && sudo apt-get install --no-install-recommends -y libvips node-gyp

Gemfile

@@ -54,6 +54,9 @@ gem "resque-scheduler"
 # intentionally NOT used: it depends on sass-rails/sprockets, which conflicts
 # with this app's Propshaft asset pipeline and breaks `assets:precompile`.
 
+# database
+gem "pg"
+
 group :development, :test do
   # See https://guides.rubyonrails.org/debugging_rails_applications.html#debugging-with-the-debug-gem
   gem "debug", platforms: %i[ mri windows ], require: "debug/prelude"

Gemfile.lock

@@ -299,6 +299,12 @@ GEM
     parser (3.3.12.0)
       ast (~> 2.4.1)
       racc
+    pg (1.6.3)
+    pg (1.6.3-aarch64-linux)
+    pg (1.6.3-aarch64-linux-musl)
+    pg (1.6.3-arm64-darwin)
+    pg (1.6.3-x86_64-linux)
+    pg (1.6.3-x86_64-linux-musl)
     popper_js (2.11.8)
     pp (0.6.4)
       prettyprint
@@ -592,6 +598,7 @@ DEPENDENCIES
   importmap-rails
   jbuilder
   kamal
+  pg
   propshaft
   puma (>= 5.0)
   rails (~> 8.1.3)
@@ -735,6 +742,12 @@ CHECKSUMS
   ostruct (0.6.3) sha256=95a2ed4a4bd1d190784e666b47b2d3f078e4a9efda2fccf18f84ddc6538ed912
   parallel (2.1.0) sha256=b35258865c2e31134c5ecb708beaaf6772adf9d5efae28e93e99260877b09356
   parser (3.3.12.0) sha256=21a6d7f755d5a24dfbdc6e6b772e4e879a52e7631a88bc5a3a134606052c9828
+  pg (1.6.3) sha256=1388d0563e13d2758c1089e35e973a3249e955c659592d10e5b77c468f628a99
+  pg (1.6.3-aarch64-linux) sha256=0698ad563e02383c27510b76bf7d4cd2de19cd1d16a5013f375dd473e4be72ea
+  pg (1.6.3-aarch64-linux-musl) sha256=06a75f4ea04b05140146f2a10550b8e0d9f006a79cdaf8b5b130cde40e3ecc2c
+  pg (1.6.3-arm64-darwin) sha256=7240330b572e6355d7c75a7de535edb5dfcbd6295d9c7777df4d9dddfb8c0e5f
+  pg (1.6.3-x86_64-linux) sha256=5d9e188c8f7a0295d162b7b88a768d8452a899977d44f3274d1946d67920ae8d
+  pg (1.6.3-x86_64-linux-musl) sha256=9c9c90d98c72f78eb04c0f55e9618fe55d1512128e411035fe229ff427864009
   popper_js (2.11.8) sha256=f4b0be717fc0d50bdb3dbbc55788525a9e0e8f640b76c9971fc34ee609eadbd2
   pp (0.6.4) sha256=dfcb0fce700c41456265922884f9fe195d7fbb0674a3578e6c0f69588e82b570
   prettyprint (0.2.0) sha256=2bc9e15581a94742064a3cc8b0fb9d45aae3d03a1baa6ef80922627a0766f193

README.md

@@ -15,58 +15,60 @@ docker compose --profile app up
 # Create the blacklight-collection
 /bin/bash ./solr/dev-init.sh 
 ```
-#### Indexing a directory
+#### Start ingest pipeline
 ```shell
+# Add your data under /data/ead/{repo_slug}
+mkdir -p ./data/ead
+cp -r ./sample-ead/scrc ./data/ead/
+
 # Inside the container
 docker compose exec app bash
-DIR=<your directory path> REPOSITORY=<repository id> rake um_arclight:index_dir
+rake arclight:ingest_everything
 
-# Outside the container
-SOLR_URL=http://localhost:8983/solr/blacklight-collection DIR=./<location> REPOSITORY_ID=<repository shortname> rake um_arclight:index_dir
 ```
+
 ### Docker Hybrid development
 Use hybrid for quick rebuilds of Rails (outside a container) without repeatedly spinning up Solr/Zookeeper for each reset.
 ##### Services shell
 
 ```shell
 # Build resque-web and resque images  
 docker compose build
-# Start zookeeper, solr, redis, resque and resque-web containers
-docker compose up 
+# Start zookeeper, solr, postgres, redis, resque and resque-web containers
+docker compose up
 ```
 ##### Rails shell
 
 ```shell
 # Create the blacklight-collection
-/bin/bash ./solr/dev-init.sh 
+/bin/bash ./solr/dev-init.sh
 # Bundle install the gems outside of the container
 bundle install
-# Migrate the database and prepare it for GUI finding aid indexing
-# NOTE: This command is run via bin/docker-entrypoint when resque is brought up
+# You may also need to install vips: https://formulae.brew.sh/formula/vips.
+
+# NOTE: The migration and database preparation command here is run by the migrate service
+# (see Notes - Database and Migration).
 # bin/rails db:prepare 
 # Development Rails server using Solr container
 SOLR_URL=http://localhost:8983/solr/blacklight-collection bin/dev
 ```
-##### Browser Resque Web
-Open http://localhost:5678 in your browser
 
 ##### Command shell
 ```shell
-# List all available tasks
-bin/rails --task
-```
-```shell
-# Index sample-ead
+# Rake task to start ingest pipeline
 rsync -av --progress sample-ead/ data/ead/
-FINDING_AID_DATA=./data bin/rails arclight:ingest_everything 
+SOLR_URL=http://localhost:8983/solr/blacklight-collection \
+  FINDING_AID_DATA=./data bin/rails \
+  arclight:ingest_everything
 ```
 
-### Troubleshooting
-If you encounter this indexing error post-Boxrunner merge
-```
-. . . .rake aborted!
-Bundler::GemRequireError: There was an error while trying to load the gem 'ruby-vips'. (Bundler::GemRequireError)
-Gem Load Error is: Could not open library 'vips.42': dlopen(vips.42, 0x0005): tried: 'vips.42' (no such file), '/System/Volumes/Preboot/Cryptexes/OSvips.42' (no such file), '/usr/lib/vips.42' (no such file, not in dyld cache), 'vips.42' (no such file), '/usr/local/lib/vips.42' (no such file), '/usr/lib/vips.42' (no such file, not in dyld cache).
-```
-I solved it by installing `vips` using `brew install vips`.    
-Then run `bundle install`
+##### Browser Resque Web
+Open http://localhost:5678 in your browser
+
+## Notes
+
+### Database and Migration
+
+The application is now setup to use a single PostgreSQL database (shared by the `app` and `resque` services)
+in both all-Docker and hybrid usages. We use a service called migrate to run `db:prepare`
+before both `app` and `resque` run, which helps us avoid race conditions during schema set up.

bin/docker-entrypoint

@@ -1,15 +1,9 @@
 #!/bin/bash -e
 
-# Prepare the database for the long-running services that depend on it:
-#   * the Rails web server (renders catalog pages), and
-#   * the Resque worker -- Package::Generator renders /catalog/:id in-process while
-#     building HTML/PDF, and in development the migration_error=:page_load middleware
-#     runs a pending-migration check on that request. The worker container has its own
-#     (image-baked, non-volume) SQLite DB, so it must be prepared here too or packaging
-#     jobs fail with a 500 on the internal render.
-if { [ "${@: -2:1}" == "./bin/rails" ] && [ "${@: -1:1}" == "server" ]; } || \
-   [[ "${*}" == *resque:work* ]]; then
-  ./bin/rails db:prepare
-fi
+# NOTE: database preparation is intentionally NOT done here. A dedicated,
+# run-once "migrate" service (see compose.yml) owns `db:prepare` so exactly one
+# process touches the schema, avoiding concurrent schema-load races between the
+# web server and the Resque worker. This also mirrors the Kubernetes model where
+# migrations run as a one-shot Job rather than on every pod/container start.
 
 exec "${@}"

compose.yml

@@ -4,10 +4,14 @@ services:
     profiles:
       - app
     depends_on:
+      migrate:
+        condition: service_completed_successfully
       solr:
         condition: service_healthy
       redis:
         condition: service_healthy
+      postgres:
+        condition: service_healthy
     build:
       context: .
     ports:
@@ -16,10 +20,24 @@ services:
       - SOLR_URL=http://solr:8983/solr/blacklight-collection
       - REDIS_URL=redis://redis:6379/0
       - FINDING_AID_DATA=/opt/app-data
+      - DB_HOST=postgres
     env_file:
       - .env
+    command: ./bin/rails server -b 0.0.0.0 -p 3000
     volumes:
       - ./data:/opt/app-data
+  migrate:
+    build:
+      context: .
+    depends_on:
+      postgres:
+        condition: service_healthy
+    environment:
+      - DB_HOST=postgres
+    env_file:
+      - .env
+    command: ./bin/rails db:prepare
+    restart: "no"
   redis:
     image: redis:7
     healthcheck:
@@ -33,20 +51,43 @@ services:
     volumes:
       - redis-data:/data
     command: redis-server --appendonly yes
+  postgres:
+    healthcheck:
+      interval: 30s
+      timeout: 10s
+      start_period: 30s
+      retries: 5
+      test: [ "CMD-SHELL", "pg_isready -h localhost -d boxwalker -p 5432 -U myuser" ]
+    image: 'postgres:latest'
+    environment:
+      - 'POSTGRES_DB=boxwalker'
+      - 'POSTGRES_USER=myuser'
+      - 'POSTGRES_PASSWORD=secret'
+    ports:
+      - '127.0.0.1:5432:5432'
+    volumes:
+      - postgres-data:/var/lib/postgresql
   resque:
     depends_on:
+      migrate:
+        condition: service_completed_successfully
       solr:
         condition: service_healthy
       redis:
         condition: service_healthy
+      postgres:
+        condition: service_healthy
     build:
       context: .
     environment:
-      - RAILS_ENV=development
       - SOLR_URL=http://solr:8983/solr/blacklight-collection
       - REDIS_URL=redis://redis:6379/0
       - QUEUE=*
       - FINDING_AID_DATA=/opt/app-data
+      - DB_HOST=postgres
+#      - DB_NAME=boxwalker
+#      - DB_USER=myuser
+#      - DB_PASSWORD=secret
     env_file:
       - .env
     volumes:
@@ -106,6 +147,7 @@ services:
       ZOO_SERVERS: server.1=0.0.0.0:2888:3888;2181
 volumes:
   data:
+  postgres-data:
   solr-data:
   redis-data:
   zk-data:

config/database.yml

@@ -1,40 +1,48 @@
-# SQLite. Versions 3.8.0 and up are supported.
-#   gem install sqlite3
-#
-#   Ensure the SQLite 3 gem is defined in your Gemfile
-#   gem "sqlite3"
-#
+---
 default: &default
-  adapter: sqlite3
-  max_connections: <%= ENV.fetch("RAILS_MAX_THREADS") { 5 } %>
-  timeout: 5000
+  adapter: postgresql
+  encoding: unicode
+  pool: 5
+  host: <%= ENV.fetch("DB_HOST", "127.0.0.1") %>
+  username: <%= ENV.fetch("DB_USERNAME", "myuser") %>
+  password: <%= ENV.fetch("DB_PASSWORD", "secret") %>
+  # Skip GSSAPI/Kerberos negotiation on connect. libpq defaults to
+  # gssencmode=prefer, which hangs ~34s in hybrid dev when the host has an
+  # expired Kerberos ticket (e.g. UMich ADSROOT). Postgres in Docker doesn't
+  # use GSSAPI, so disabling it is correct in every environment.
+  gssencmode: <%= ENV.fetch("DB_GSSENCMODE", "disable") %>
+  # Detect dead/half-open sockets quickly instead of blocking on the OS TCP
+  # timeout (~34s). This matters in hybrid dev (host Ruby -> Postgres in Docker
+  # Desktop), whose network proxy silently drops idle connections. The keepalive
+  # probes also keep pooled connections warm so they never go stale.
+  connect_timeout: <%= ENV.fetch("DB_CONNECT_TIMEOUT", 5) %>
+  keepalives: 1
+  keepalives_idle: 10
+  keepalives_interval: 5
+  keepalives_count: 3
 
 development:
   <<: *default
-  database: storage/development.sqlite3
+  database: <%= ENV.fetch("DB_NAME", "boxwalker") %>
 
-# Warning: The database defined as "test" will be erased and
-# re-generated from your development database when you run "rake".
-# Do not set this db to the same as development or production.
 test:
   <<: *default
-  database: storage/test.sqlite3
+  database: <%= ENV.fetch("DB_NAME", "boxwalker") %>
 
-# Store production database in the storage/ directory, which by default
-# is mounted as a persistent Docker volume in config/deploy.yml.
 production:
   primary:
     <<: *default
-    database: storage/production.sqlite3
+    database: <%= ENV.fetch("DB_NAME", "boxwalker") %>
   cache:
     <<: *default
-    database: storage/production_cache.sqlite3
+    database: <%= ENV.fetch("CACHE_DB_NAME", "boxwalker_cache") %>
     migrations_paths: db/cache_migrate
   queue:
     <<: *default
-    database: storage/production_queue.sqlite3
+    database: <%= ENV.fetch("QUEUE_DB_NAME", "boxwalker_queue") %>
     migrations_paths: db/queue_migrate
   cable:
     <<: *default
-    database: storage/production_cable.sqlite3
+    database: <%= ENV.fetch("CABLE_DB_NAME", "boxwalker_cable") %>
     migrations_paths: db/cable_migrate
+

config/environments/development.rb

@@ -26,7 +26,7 @@
   end
 
   # Change to :null_store to avoid any caching.
-  config.cache_store = :memory_store
+  config.cache_store = :null_store
 
   # Store uploaded files on the local file system (see config/storage.yml for options).
   config.active_storage.service = :local