Enable PostgreSQL as the default database for rhdh-local by NiallTwomey2 · Pull Request #324 · redhat-developer/rhdh-local

NiallTwomey2 · 2026-09-16T12:57:01Z

Description

Enable PostgreSQL as the default database for RHDH Local (RHIDP-16882 / RHIDP-16883).

db is in compose.yaml, image ${POSTGRES_IMAGE:-quay.io/fedora/postgresql-18:latest}
Anonymous volume at /var/lib/pgsql/data
compose stop/start/restart keep data;
compose down then compose up starts a new empty volume (RHDH does not support DB downgrades)
rhdh depends_on: db (healthy)
Default app-config.yaml uses client: pg and POSTGRES_* from default.env
Removed WITH_POSTGRES,app-config.db.yaml,compose-with-db.yaml
wait-for-plugins-and-start.sh waits for $POSTGRES_HOST:$POSTGRES_PORT before starting node
Corporate-proxy overlay sets NO_PROXY=localhost,127.0.0.1
SQLite remains an optional app-config.local.yaml override
RHEL Postgres is opt-in: POSTGRES_IMAGE=registry.redhat.io/rhel10/postgresql-18:latest in .env plus registry login

Which issue(s) does this PR fix or relate to

Relates to RHIDP-16882
Relates to RHIDP-16883
Follow up issue RHIDP-16884 - CI check that the effective RHDH config is client: pg (Postgres, not SQLite) after the stack is healthy.

PR acceptance criteria

Tests updated and passing
Documentation updated
Built-in TechDocs updated if needed

How to test changes / Special notes to the reviewer

Postgres uses an anonymous volume at /var/lib/pgsql/data.

Command	Catalog / plugin DB data
`podman compose stop` then `start` (or `restart`)	Kept — same containers, same volume
`podman compose down` then `podman compose up -d`	Wiped — PostgresSQL catalog is wiped and replaced with a new empty volume (use this when switching RHDH versions) Named volumes such as plugins and RAG are kept.
`podman compose down --volumes` then `podman compose up -d`	PostgreSQL is Wiped, and named volumes (plugins, RAG) are deleted too. A full reset

# Fresh start
podman compose down --volumes --remove-orphans
podman compose up -d
# http://localhost:7007 — rhdh logs: waited for Postgres at db:5432

podman exec -e PGPASSWORD=postgres db \
  psql -U postgres -d postgres -c '\l'
# expect backstage_plugin_catalog and other backstage_plugin_* DBs

# Keep data
podman compose stop
podman compose start

# Wipe Postgres only (new anonymous volume)
podman compose down
podman compose up -d

# Optional proxy:
CORPORATE_PROXY_IMAGE=docker.io/ubuntu/squid:latest \
  podman compose -f compose.yaml -f compose-with-corporate-proxy.yaml up -d
# Squid access.log should have no CONNECT to db:5432

…base

NiallTwomey2 · 2026-09-16T13:28:00Z

+        $TOOL exec -e PGPASSWORD=postgres db \
+        psql -U postgres -d postgres -c '\l'


Run postgreSQL inside the db container.

NiallTwomey2 · 2026-09-16T13:29:46Z

+        if [ -z "$plugin_dbs" ]; then
+          echo "ERROR: Postgres is up but no backstage_plugin_* databases were found" >&2
+          echo "RHDH is probably using SQLite, or never migrated plugins." >&2
+          exit 1
+        fi


If there are no backstage plugins inside $plugin_dbs, then issue an error message and exit 1 as a failure
Scenario: Postgres could be running inside the container, but RHDH never created plugin databases, due to databases never being migrated or SQLite is still marked as the default database.

NiallTwomey2 · 2026-09-16T13:32:51Z

+        if ! echo "$plugin_dbs" | grep -qx 'backstage_plugin_catalog'; then
+          echo "ERROR: missing required database backstage_plugin_catalog" >&2
+          exit 1
+        fi


The backstage plugin catalog must exist within the backstage plugins pulled from "$plugin_dbs", if this is not the case, this will result in an error

NiallTwomey2 · 2026-09-16T13:47:42Z

+      env:
+        TOOL: ${{ inputs.container_tool }}
+      run: |
+        set -euo pipefail


Ensures that if there are any failures in the bash script, it exits immediately

NiallTwomey2 · 2026-09-16T13:53:43Z

+
+        plugin_dbs=$($TOOL exec -e PGPASSWORD=postgres db \
+          psql -U postgres -d postgres -Atc \
+          "SELECT datname FROM pg_database WHERE datname LIKE 'backstage_plugin_%' ORDER BY 1;")


Querying to find existing PostgreSQL column for datname within the system catalog pg_database.
List the database names that begin with backstage_plugin_.

NiallTwomey2 · 2026-09-16T14:18:34Z

    environment:
      - HTTP_PROXY=http://proxy:3128
      - HTTPS_PROXY=http://proxy:3128
+      - NO_PROXY=localhost,127.0.0.1,db


With Postgres now being the default database, RHDH connects to PostgreSQL via port 5432 / TCP port.
Without NO_PROXY, node traffic still sees HTTP_PROXY and can try to access port 5432 through Squid (Squid only operates with HTTP not TCP for PostgreSQL), as a result database connections to PostgreSQL may fail under timeouts.
With NO_PROXY as the default, RHDH connects traffic to db via the Compose service on TCP port 5432.

Suggested change

- NO_PROXY=localhost,127.0.0.1,db

- NO_PROXY=localhost,127.0.0.1

I don't think this is needed. Instead, I would suggest adding db to the internal_net network instead, so it is reachable from rhdh and rhdh won't need to bypass the proxy at all for this. This would actually mimic a real corporate setup: the DB is usually only accessed only from the internal network.

NiallTwomey2 · 2026-09-16T14:36:35Z

+    volumes:
+      - "postgresqldata:/var/lib/pgsql/data"


postgresqldata volume holds PostgreSQL's data files on the volume disk.
Data and files information is retained to the postgresqldata volume, even after the overall db container is deleted.
compose down --volumes will remove the volume and the data.

Suggested change

volumes:

- "postgresqldata:/var/lib/pgsql/data"

volumes:

- "/var/lib/pgsql/data"

One of the use cases of RHDH Local is to test different versions of RHDH, so this might mean switching from a new version to an older one. And since RHDH currently has issues with downgrades, a named volume might be an issue. The compose-with-db.yaml handles this using an anonymous volume and it makes sense in this context:

rhdh-local/compose-with-db.yaml

Line 17 in ea366db

- "/var/lib/pgsql/data"

NiallTwomey2 · 2026-09-16T14:42:35Z

+    healthcheck:
+      test: ["CMD", "pg_isready", "-U", "postgres"]
+      interval: 5s
+      timeout: 5s
+      retries: 5


Every 5 seconds, the live healthcheck, inside the container will check to see if the status of PostgreSQL accepting connections is still successful, which indicates that the database service is healthy for RHDH.

NiallTwomey2 · 2026-09-16T14:43:07Z

      install-dynamic-plugins:
        condition: service_completed_successfully
+      db:
+        condition: service_healthy


RHDH depends on the healthy status connection of the db service, otherwise it will not run, this comes from the healthcheck from the db service.

NiallTwomey2 · 2026-09-16T14:47:03Z

@@ -51,6 +51,23 @@ fi
 # Loaded after the patched default config (SQLite) and before user local config.
 EXTRA_CONFIGS=""
 if [[ "${WITH_POSTGRES:-}" == "true" ]]; then


Waits for PostgreSQL to accept TCP connections

NiallTwomey2 · 2026-09-16T14:49:28Z

+  PG_HOST="${POSTGRES_HOST:-db}"
+  PG_PORT="${POSTGRES_PORT:-5432}"
+  PG_TIMEOUT="${POSTGRES_WAIT_TIMEOUT:-60}"


Specifying the database (PostgreSQL), the port - 5432, and the timeout wait time for connection - 60

rm3l

@NiallTwomey2 Looks like CI is failing on this PR. Could you please take a look?

NiallTwomey2 · 2026-09-17T11:14:28Z

@rm3l The CI has passed. Podman does not cause the CI checks to fail anymore.
CI checks could not identify any database inside nested podman containers and exits 125.
Now, after compose exec, the CI checks will use the same path as podman compose up, the database is now acknowledged by the checks - backstage plugin databases are identifiable now.

rm3l · 2026-09-18T13:56:33Z

        echo "[$(date)] RHDH is ready"
        curl -i --insecure http://localhost:7007

+    - name: Assert Backstage plugin databases exist in PostgreSQL


Why do we need this assertion? IMO this is already guaranteed/tested by RHDH/Backstage, so no need to test this here again. And this test might be fragile if later on we decide to change the pluginDivisionMode or if we change the default backstage_plugin_ DB prefix.
If your goal is to make sure we are using PostgreSQL OOTB, I think you can just verify that the effective RHDH configuration has the right info to connect to PostgreSQL. But tbh, I don't think this is needed.

I agree, that assertion is an unneeded fail safe, that makes sense to leave this assertion out.
This assertion was to address the failure of the database, caused by missing backstage plugin databases in a container - RHIDP-16884

rm3l · 2026-09-18T13:59:14Z

+          echo "ERROR: Postgres is up but no backstage_plugin_* databases were found" >&2
+          echo "RHDH is probably using SQLite, or never migrated plugins." >&2


it doesn't necessarily mean that we are using SQLite; depending on the config, we might still be using PostgreSQL but with all plugins sharing a single database.

rm3l · 2026-09-18T14:02:12Z

-  # comment out the following 'database' section to use the PostgreSQL database
+      
  database:
    client: better-sqlite3
    connection: ":memory:"


The goal is to make PostgreSQL the default, so the default app-config should reflect that IMO.

rm3l · 2026-09-18T14:05:32Z

+
+  db:
+    container_name: db
+    image: "${POSTGRES_IMAGE:-registry.redhat.io/rhel10/postgresql-18:latest}" # dclint disable-line service-image-require-explicit-tag


Suggested change

image: "${POSTGRES_IMAGE:-registry.redhat.io/rhel10/postgresql-18:latest}" # dclint disable-line service-image-require-explicit-tag

image: "${POSTGRES_IMAGE:-quay.io/fedora/postgresql-18:latest}" # dclint disable-line service-image-require-explicit-tag

The default DB image should be the public one that doesn't require any auth, and then we should document how people can use a commercially-supported version of it from registry.redhat.io.

rm3l · 2026-09-18T14:09:01Z

+    ```

-> **Warning:** If you already run optional Postgres and have a persisted `/var/lib/pgsql/data` volume from an **older major** image, do **not** only bump `POSTGRES_IMAGE` (or the default image major). Follow [Upgrading PostgreSQL](#upgrading-postgresql) first so the volume is upgraded safely.
+You do not need [`compose-with-db.yaml`](https://github.com/redhat-developer/rhdh-local/blob/HEAD/compose-with-db.yaml) or [`app-config.local.yaml`](https://github.com/redhat-developer/rhdh-local/blob/HEAD/configs/app-config/app-config.local.example.yaml) to use Postgres. That overlay remains for CI and back-compat; a normal start does not use it.


You can remove the compose-with-db.yaml file.

rm3l · 2026-09-18T14:10:25Z

+You do not need [`compose-with-db.yaml`](https://github.com/redhat-developer/rhdh-local/blob/HEAD/compose-with-db.yaml) or [`app-config.local.yaml`](https://github.com/redhat-developer/rhdh-local/blob/HEAD/configs/app-config/app-config.local.example.yaml) to use Postgres. That overlay remains for CI and back-compat; a normal start does not use it.

-1. Login to container registry with *Red Hat Login* credentials to use `postgresql` image
+`compose.yaml` sets `WITH_POSTGRES=true` on `rhdh`. On startup, RHDH loads [`app-config.db.yaml`](https://github.com/redhat-developer/rhdh-local/blob/HEAD/configs/app-config/app-config.db.yaml) (`client: pg`) after the SQLite block in [`app-config.yaml`](https://github.com/redhat-developer/rhdh-local/blob/HEAD/configs/app-config/app-config.yaml).


We can simplify this by removing the logic in the script handling WITH_POSTGRES since the default app-config will already contain the right info to connect to PostgreSQL.

rm3l · 2026-09-18T14:11:12Z

+The examples below use `podman` and `podman compose`. If you use Docker, replace `podman` with `docker` (for example `docker login`, `docker compose`, `docker exec`).

-2. Start RHDH with the optional Postgres overlay [`compose-with-db.yaml`](https://github.com/redhat-developer/rhdh-local/blob/main/compose-with-db.yaml). Note that the order of the YAML files is important:
+> **NOTE**: The default image is [`registry.redhat.io/rhel10/postgresql-18`](https://catalog.redhat.com/en/software/containers/rhel10/postgresql-18/6942a60aab9edd836017e3d0). That registry needs a [Red Hat Login](https://access.redhat.com/RegistryAuthentication#getting-a-red-hat-login-2) (`podman login registry.redhat.io`). To skip login, set `POSTGRES_IMAGE` in `.env` (for example `quay.io/fedora/postgresql-18:latest`).


quay.io/fedora/postgresql-18:latest should be the default, otherwise it means that users will require authentication to start using RHDH Local, which would be an issue.

rm3l · 2026-09-18T14:40:02Z

+    volumes:
+      - "postgresqldata:/var/lib/pgsql/data"


Suggested change

volumes:

- "postgresqldata:/var/lib/pgsql/data"

volumes:

- "/var/lib/pgsql/data"

One of the use cases of RHDH Local is to test different versions of RHDH, so this might mean switching from a new version to an older one. And since RHDH currently has issues with downgrades, a named volume might be an issue. The compose-with-db.yaml handles this using an anonymous volume and it makes sense in this context:

rhdh-local/compose-with-db.yaml

Line 17 in ea366db

- "/var/lib/pgsql/data"

…-default

Fortune-Ndlovu

Thanks for the RHIDP-16882 work. The overall approach looks good: Postgres in the default compose file, pg client in app-config, startup wait, and NO_PROXY for db behind the corporate proxy overlay.

Before merge, please align the PR description with the code, add (or explicitly defer) the CI check for plugin databases, and pin the default Postgres image by digest instead of :latest.

Details are on the inline comments.

Fortune-Ndlovu · 2026-09-23T09:57:11Z

+
+  db:
+    container_name: db
+    image: "${POSTGRES_IMAGE:-quay.io/fedora/postgresql-18:latest}" # dclint disable-line service-image-require-explicit-tag


Please default to an image digest (for example quay.io/fedora/postgresql-18@sha256:...) instead of :latest. The tag can change between pulls, which makes local runs and CI less predictable. Users who want the newest image can still set POSTGRES_IMAGE in .env.

Fortune-Ndlovu · 2026-09-23T09:57:11Z

      # The default corporate proxy image is located on registry.redhat.io, which requires authentication.
      CORPORATE_PROXY_IMAGE: docker.io/ubuntu/squid:latest
-      # The default Postgres image is located on registry.redhat.io, which requires authentication.
      POSTGRES_IMAGE: quay.io/fedora/postgresql-18:latest


Use the same digest pin here as in compose.yaml so CI always runs against a known Postgres build. Right now :latest can drift without a code change.

Fortune-Ndlovu · 2026-09-23T09:57:11Z

+The examples below use `podman` and `podman compose`. If you use Docker, replace `podman` with `docker` (for example `docker login`, `docker compose`, `docker exec`).

-1. Login to container registry with *Red Hat Login* credentials to use `postgresql` image
+> **NOTE**: The default image is [`quay.io/fedora/postgresql-18:latest`](https://quay.io/repository/fedora/postgresql-18). No registry login is required for a normal start. To use the commercially supported image, set `POSTGRES_IMAGE=registry.redhat.io/rhel10/postgresql-18:latest` in `.env` and [log in to `registry.redhat.io`](https://access.redhat.com/RegistryAuthentication#getting-a-red-hat-login-2) (`podman login registry.redhat.io`).


Once the default image is digest-pinned in compose, update this note to match (digest in the repo default, :latest only as an optional override in .env).

Fortune-Ndlovu · 2026-09-23T09:57:11Z

-   podman compose -f compose.yaml -f compose-with-db.yaml up -d
-   ```
+- `podman compose stop` / `start` (or `restart`) keep the volume and catalog data.
+- `podman compose down` then `podman compose up` creates a new empty volume. Use this when switching RHDH versions.


This says compose down then up creates a new empty Postgres volume. The PR description "How to test" section says podman compose down && podman compose up -d keeps data. Please make those two places agree so users know when catalog data is kept or wiped.

Thanks for raising this.
Just to clarify:

podman compose down && podman compose up Containers are removed and down wipes the previous volume and removes the db container. The anonymous volume is no longer attached to the running container.
up attaches a new empty volume to the container and creates a new database container for postgreSQL.

podman compose stop then podman compose start (or restart), maintains the same containers, the same anonymous volume. Catalog and plugin data is retained.

podman compose down --volumes then podman compose up, the volume is wiped, as well as PostgreSQL and the named volumes (plugins. RAG) are deleted.

I now have made edits to the PR description to highlight this.

Fortune-Ndlovu · 2026-09-23T09:57:11Z

+    sleep 2
+    elapsed=$((elapsed + 2))
+  done
+  echo "PostgreSQL at reachable at ${PG_HOST}:${PG_PORT}"


Small typo: "PostgreSQL at reachable" should be "PostgreSQL is reachable".

Fortune-Ndlovu · 2026-09-23T09:57:11Z

@@ -1,36 +0,0 @@
-# This Compose file is not usable on its own.


This overlay is removed in the PR, which matches moving db into compose.yaml. Please update the PR description: it still mentions keeping compose-with-db.yaml, WITH_POSTGRES, and app-config.db.yaml. That will confuse reviewers and anyone tracing RHIDP-16882.

Fortune-Ndlovu · 2026-09-23T09:57:11Z

+      host: ${POSTGRES_HOST}
+      port: ${POSTGRES_PORT}
+      user: ${POSTGRES_USER}
+      password: ${POSTGRES_PASSWORD}


Fine for local dev. Worth a short note in the Postgres guide that catalog data now lives in Postgres on disk (until volumes are removed), and that default postgres credentials are not for production.

rm3l · 2026-10-01T07:43:30Z

        env: |
          CORPORATE_PROXY_IMAGE=docker.io/ubuntu/squid:latest
-          POSTGRES_IMAGE=quay.io/fedora/postgresql-18:latest
+          POSTGRES_IMAGE=quay.io/fedora/postgresql-18@sha256:9ff34b7dd85956cb5cfbaeff193c66ddc4f410aceb20848330d6432b18a99bd5


Suggested change

POSTGRES_IMAGE=quay.io/fedora/postgresql-18@sha256:9ff34b7dd85956cb5cfbaeff193c66ddc4f410aceb20848330d6432b18a99bd5

POSTGRES_IMAGE=quay.io/fedora/postgresql-18:latest

I think it is fine to use latest here. The image name specifies postgresql-18, so it shouldn't cause inadvertent major upgrades of the DB, by contract. Or, if we want to use digests, we should make sure this is monitored by Renovate, otherwise it will likely fall behind.
I would suggest keeping latest for now and maybe capturing a followup ticket to handle this consistently, as this is not the single image that might be impacted.

rm3l · 2026-10-01T07:46:26Z

              repo: context.repo.repo,
            });
-
+            


Suggested change

rm3l · 2026-10-01T07:46:58Z

      CORPORATE_PROXY_IMAGE: docker.io/ubuntu/squid:latest
-      # The default Postgres image is located on registry.redhat.io, which requires authentication.
-      POSTGRES_IMAGE: quay.io/fedora/postgresql-18:latest
+      POSTGRES_IMAGE: quay.io/fedora/postgresql-18@sha256:9ff34b7dd85956cb5cfbaeff193c66ddc4f410aceb20848330d6432b18a99bd5


Suggested change

POSTGRES_IMAGE: quay.io/fedora/postgresql-18@sha256:9ff34b7dd85956cb5cfbaeff193c66ddc4f410aceb20848330d6432b18a99bd5

POSTGRES_IMAGE: quay.io/fedora/postgresql-18:latest

Same suggestion. latest is fine.

rm3l · 2026-10-01T07:47:29Z

      CORPORATE_PROXY_IMAGE: docker.io/ubuntu/squid:latest
-      # The default Postgres image is located on registry.redhat.io, which requires authentication.
-      POSTGRES_IMAGE: quay.io/fedora/postgresql-18:latest
+      POSTGRES_IMAGE: quay.io/fedora/postgresql-18@sha256:9ff34b7dd85956cb5cfbaeff193c66ddc4f410aceb20848330d6432b18a99bd5


Suggested change

POSTGRES_IMAGE: quay.io/fedora/postgresql-18@sha256:9ff34b7dd85956cb5cfbaeff193c66ddc4f410aceb20848330d6432b18a99bd5

POSTGRES_IMAGE: quay.io/fedora/postgresql-18:latest

Same suggestion: latest is fine.

rm3l · 2026-10-01T08:58:59Z

-   You can combine this with other overlays the same way. For example, with the [corporate proxy](corporate-proxy-setup-sim.md) setup:
+- `podman compose stop` / `start` (or `restart`) keep the volume and catalog data.
+- `podman compose down` then `podman compose up` creates a new empty volume, catalog and plugin data is wiped. Use this when switching RHDH versions.
+- `podman compose down --volumes` also deletes other Compose volumes (plugins, RAG, Postgres, and so on).


Suggested change

- `podman compose down --volumes` also deletes other Compose volumes (plugins, RAG, Postgres, and so on).

- `podman compose down --volumes` also deletes other Compose volumes (plugins, Postgres, and so on).

RAG is going to be removed soon.

rm3l · 2026-10-01T10:31:57Z

    environment:
      - HTTP_PROXY=http://proxy:3128
      - HTTPS_PROXY=http://proxy:3128
+      - NO_PROXY=localhost,127.0.0.1,db


Suggested change

- NO_PROXY=localhost,127.0.0.1,db

- NO_PROXY=localhost,127.0.0.1

The install-dynamic-plugins service does not need to reach the DB.

rm3l · 2026-10-01T10:33:44Z

+  echo "Waiting for Postgres at ${PG_HOST}:${PG_PORT} ..."
+  until bash -c "exec 3<>/dev/tcp/${PG_HOST}/${PG_PORT}" 2>/dev/null; do
+    if [ "${elapsed}" -ge "${PG_TIMEOUT}" ]; then
+      echo "Timed out waiting for PostgreSQL at ${PG_HOST}:${PG_PORT} after ${PG_TIMEOUT}s" >&2
+      exit 1
+    fi
+    echo "PostgreSQL is not reachable yet, retrying..."
+    sleep 2
+    elapsed=$((elapsed + 2))
+  done
+  echo "PostgreSQL is reachable at ${PG_HOST}:${PG_PORT}"


Please fix the indentation here, as well as the SonarCloud suggestion to Use '[[' instead of '[' for conditional tests. The '[[' construct is safer and more feature-rich.

rm3l · 2026-10-01T10:45:06Z

    environment:
      - HTTP_PROXY=http://proxy:3128
      - HTTPS_PROXY=http://proxy:3128
+      - NO_PROXY=localhost,127.0.0.1,db


Suggested change

- NO_PROXY=localhost,127.0.0.1,db

- NO_PROXY=localhost,127.0.0.1

I don't think this is needed. Instead, I would suggest adding db to the internal_net network instead, so it is reachable from rhdh and rhdh won't need to bypass the proxy at all for this. This would actually mimic a real corporate setup: the DB is usually only accessed only from the internal network.

sonarqubecloud · 2026-10-01T15:32:55Z

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

benwilcock · 2026-10-02T12:59:46Z

We never wanted PostgreSQL to be on by default so we could keep the resource overheads low. Why do we need this now?

NiallTwomey2 · 2026-10-02T14:10:39Z

@benwilcock In memory SQLite is definitely lower cost, and resource overheads are lower than of a PostgreSQL container, also due the fact that PostgreSQL adds an additional db container to rhdh-local, which contributes to resource overheads being higher.
We are looking into enabling PostgreSQL as the default database for rhdh-local, to address the loss of catalog and plugin data when rhdh-local restarts - podman compose stop/start/restart, which is currently unavailable when memory for rhdh-local is configured with better-sqlite3.

NiallTwomey2 added 2 commits September 16, 2026 13:38

feat: enable PostgreSQL as the default database in rhdh-local

8502bac

docs: changes to README.md to describe PostgreSQL as the default data…

f4c17cc

…base

openshift-ci Bot requested review from Fortune-Ndlovu and zdrapela September 16, 2026 12:57

NiallTwomey2 commented Sep 16, 2026

View reviewed changes

Merge branch 'dev' into RHIDP-16883-postgres-default

7acf8b2

rm3l reviewed Sep 17, 2026

View reviewed changes

fix: use compose exec for Postgres CI assert on podman

6e416dd

rm3l requested changes Sep 18, 2026

View reviewed changes

NiallTwomey2 added 4 commits September 21, 2026 12:35

fix: address review on default Postgres, drop overlay and CI assert

b72771e

ci: include dev in nightly matrix and keep Fedora Postgres pin

79a73c9

Merge remote-tracking branch 'upstream/dev' into RHIDP-16883-postgres…

84e58c7

…-default

fix: drop duplicate topReleaseBranches after merge

7b9576e

NiallTwomey2 requested a review from rm3l September 21, 2026 14:17

docs: anonymous Postgres volume is ephemeral after compose down

2bfd19b

Fortune-Ndlovu requested changes Sep 23, 2026

View reviewed changes

ci: pin default Postgres image to a digest

ddc01d7

NiallTwomey2 requested a review from Fortune-Ndlovu September 23, 2026 19:47

rm3l requested changes Oct 1, 2026

View reviewed changes

fix: addressing fixes to proxy, docs, and Postgres wait

6f0cf71

NiallTwomey2 requested a review from rm3l October 2, 2026 12:53

		$TOOL exec -e PGPASSWORD=postgres db \
		psql -U postgres -d postgres -c '\l'

	- NO_PROXY=localhost,127.0.0.1,db
	- NO_PROXY=localhost,127.0.0.1

		echo "ERROR: Postgres is up but no backstage_plugin_* databases were found" >&2
		echo "RHDH is probably using SQLite, or never migrated plugins." >&2

	image: "${POSTGRES_IMAGE:-registry.redhat.io/rhel10/postgresql-18:latest}" # dclint disable-line service-image-require-explicit-tag
	image: "${POSTGRES_IMAGE:-quay.io/fedora/postgresql-18:latest}" # dclint disable-line service-image-require-explicit-tag

		@@ -1,36 +0,0 @@
		# This Compose file is not usable on its own.

	POSTGRES_IMAGE=quay.io/fedora/postgresql-18@sha256:9ff34b7dd85956cb5cfbaeff193c66ddc4f410aceb20848330d6432b18a99bd5
	POSTGRES_IMAGE=quay.io/fedora/postgresql-18:latest

	POSTGRES_IMAGE: quay.io/fedora/postgresql-18@sha256:9ff34b7dd85956cb5cfbaeff193c66ddc4f410aceb20848330d6432b18a99bd5
	POSTGRES_IMAGE: quay.io/fedora/postgresql-18:latest

	- `podman compose down --volumes` also deletes other Compose volumes (plugins, RAG, Postgres, and so on).
	- `podman compose down --volumes` also deletes other Compose volumes (plugins, Postgres, and so on).

Conversation

NiallTwomey2 commented Sep 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Which issue(s) does this PR fix or relate to

PR acceptance criteria

How to test changes / Special notes to the reviewer

Uh oh!

Choose a reason for hiding this comment

Uh oh!

NiallTwomey2 Sep 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

NiallTwomey2 Sep 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

rm3l Oct 1, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

NiallTwomey2 Sep 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

rm3l left a comment

Choose a reason for hiding this comment

Uh oh!

NiallTwomey2 commented Sep 17, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

NiallTwomey2 Sep 21, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Fortune-Ndlovu left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

NiallTwomey2 Sep 23, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

NiallTwomey2 commented Sep 16, 2026 •

edited

Loading

NiallTwomey2 Sep 16, 2026 •

edited

Loading

NiallTwomey2 Sep 16, 2026 •

edited

Loading

rm3l Oct 1, 2026 •

edited

Loading

NiallTwomey2 Sep 16, 2026 •

edited

Loading

NiallTwomey2 Sep 21, 2026 •

edited

Loading

NiallTwomey2 Sep 23, 2026 •

edited

Loading

rm3l Oct 1, 2026 •

edited

Loading

benwilcock commented Oct 2, 2026 •

edited

Loading

NiallTwomey2 commented Oct 2, 2026 •

edited

Loading