Replace bitnami postgresql with integrated one

[fhacloid]

Existing Issue

Fixes #387.

Why do we need this PR?

This PR will remove the dependency to bitnami chart and replace the postgresql with a simple statefulSet with one master database.

Changes proposed in this pull request

• Removed bitnami chart dependency
• Added a postgresql statefulSet enabled when postgresql.enabled == true
• Updated values.

Contributor Checklist

• Variables are documented in the README.md
• Which branch are you merging into?
  • master is for changes related to the current release of the concourse/concourse:latest image and should be good to publish immediately

Reviewer Checklist

This section is intended for the core maintainers only, to track review progress. Please do not
fill out this section.

• Code reviewed
• Topgun tests run
• Back-port if needed
• Is the correct branch targeted? (master or dev)

[taylorsilva]

Looks good overall! Only nit-type of comments and things to change. Ran our k8s integration test suite against it, it only failed on one test related to the externalUrl. Once that change, and the other minor things, are reverted/fixed, this will be good to merge.

[fhacloid]

Updated with all your comments with fixups, if approve I will rebase and let you merge :)

[taylorsilva]

Doesn't match the default in values.yaml. I think the default should be the major tag 17, not a variant tag.

[taylorsilva]

Why is this the default value of tls here?

[taylorsilva]

Remove reference to bitnami chart

[taylorsilva]

doesn't match default in README.

[fhacloid]

@taylorsilva I updated with your last nits I didn't saw.

We did some more tests on our side and added 3 settings:

• extraArgs add args to the cmd of postgres
• configOverride override the postgresql.conf
• pvcNameOverride added a way to override the name of the pvc template. This helps with migrating (you can point to the existing vpc that what created with the bitnami charts).

We updated the pvc name to match the one that was in the bitnami chart, so migrating via the PVC should work, I would advise to dump the database anyway to be sure. ;)

[fhacloid]

I get an error on _helpers.tpl at line 81:
helm lint --values values.yaml
==> Linting .
[ERROR] templates/: template: concourse/templates/_helpers.tpl:102:16: executing "concourse.statefulSet.apiVersion" at <include "concourse.kubeVersion" .>: error calling include: template: concourse/templates/_helpers.tpl:81:59: executing "concourse.kubeVersion" at <"">: invalid value; expected string

Error: 1 chart(s) linted, 1 chart(s) failed
error: Recipe lint failed on line 4 with exit code 1

Does that ring a bell on your side ?

[fhacloid]

Oh, dumb issue, pushing a fix

[fhacloid]

We did some thorough testing for the migration, and it works.

Here are the migration steps if you want to re-use the same pvc:

1. Backup your database in case something bad happends
2. Switch to the new chart version, adapt values as you see fit
3. On postgres 17, the data and pvc must be mounted at /var/lib/postgresql/data, but on the bitnami image, the volume was at /opt/bitnami and the data was at /opt/bitnami/data.

If you remount the same pvc, the data dir will not be detected by the entrypoint. So you must move the content of the data folder in the correct place.

Also, the config for both postgresql.conf and pg_hba.conf was managed using a configmap, the postgres container uses a default one on the data folder, for this migration we can recreate the default config.

For that override the command:

postgresql:
  commandOverride:
      - /bin/bash
      - -ec
      - |
          if [ -d "/var/lib/postgresql/data/data" ]; then
            echo "Migrating existing data directory..."
            mv /var/lib/postgresql/data/data/* /var/lib/postgresql/data/ || true
            rmdir /var/lib/postgresql/data/data || true

            # We recreate the default config file, but you can also override
            # the postgres config using the values provided in the new chart.
            echo "listen_addresses = '*'" > /var/lib/postgresql/data/postgresql.conf
            echo "max_connections = 100" >> /var/lib/postgresql/data/postgresql.conf
            echo "shared_buffers = 128MB" >> /var/lib/postgresql/data/postgresql.conf
            echo "dynamic_shared_memory_type = posix" >> /var/lib/postgresql/data/postgresql.conf
            echo "max_wal_size = 1GB" >> /var/lib/postgresql/data/postgresql.conf
            echo "min_wal_size = 80MB" >> /var/lib/postgresql/data/postgresql.conf
            echo "log_timezone = 'Etc/UTC'" >> /var/lib/postgresql/data/postgresql.conf
            echo "datestyle = 'iso, mdy'" >> /var/lib/postgresql/data/postgresql.conf
            echo "timezone = 'Etc/UTC'" >> /var/lib/postgresql/data/postgresql.conf
            echo "lc_messages = 'en_US.utf8'" >> /var/lib/postgresql/data/postgresql.conf
            echo "lc_monetary = 'en_US.utf8'" >> /var/lib/postgresql/data/postgresql.conf
            echo "lc_numeric = 'en_US.utf8'" >> /var/lib/postgresql/data/postgresql.conf
            echo "lc_time = 'en_US.utf8'" >> /var/lib/postgresql/data/postgresql.conf
            echo "default_text_search_config = 'pg_catalog.english'" >> /var/lib/postgresql/data/postgresql.conf
            echo "local   all             all                                     trust" > /var/lib/postgresql/data/pg_hba.conf
            echo "host    all             all             127.0.0.1/32            trust" >> /var/lib/postgresql/data/pg_hba.conf
            echo "host    all             all             ::1/128                 trust" >> /var/lib/postgresql/data/pg_hba.conf
            echo "local   replication     all                                     trust" >> /var/lib/postgresql/data/pg_hba.conf
            echo "host    replication     all             127.0.0.1/32            trust" >> /var/lib/postgresql/data/pg_hba.conf
            echo "host    replication     all             ::1/128                 trust" >> /var/lib/postgresql/data/pg_hba.conf
            echo "host all all all scram-sha-256" >> /var/lib/postgresql/data/pg_hba.conf
          fi
          exec docker-entrypoint.sh postgres

5. Delete the postgresql statefulset
6. Apply the helm chart
7. There may be a collation issue based on the previous postgresql version you have, if you get this message:

2025-09-04 12:33:45.704 UTC [609] WARNING: database "atc" has a collation version mismatch
2025-09-04 12:33:45.704 UTC [609] DETAIL: The database was created using collation version 2.36, but the operating system provides version 2.41.
2025-09-04 12:33:45.704 UTC [609] HINT: Rebuild all objects in this database that use the default collation and run ALTER DATABASE atc REFRESH COLLATION VERSION, or build PostgreSQL with the right library version.

You will need to alter the DB, you can do it by adding a lifecycle script like so:

postgresql:
  lifecycle:
    postStart:
      exec:
        command:
          - /bin/sh
          - -ec
          - |
            for i in $(seq 1 30); do
              pg_isready -U ${POSTGRES_USER:-{{ .Values.postgresql.auth.user }}} \
                         -d "dbname=${POSTGRES_DB:-{{ .Values.postgresql.auth.database }}}" \
                         -h 127.0.0.1 -p 5432 && \
              psql -U ${POSTGRES_USER:-{{ .Values.postgresql.auth.user }}} \
                   -d ${POSTGRES_DB:-{{ .Values.postgresql.auth.database }}} \
                   -c "ALTER DATABASE atc REFRESH COLLATION VERSION;" && break
              echo "Database not ready yet, retrying in 2s..."
              sleep 2
            done

8. kubectl rollout restart the postgresql statefulset

Warning

In our tests, we weren't able to find the logs of the postStart script

also:

This is the migration setup we tried on our side, please be very careful, back up your db and test the migration steps on a staging env before making it in production.

[fhacloid]

@taylorsilva Waiting for review :)

[fhacloid]

@taylorsilva still up to review, whenever you get the time ;)

[taylorsilva]

LGTM - Running this PR against our E2E test suite and then will merge.

[taylorsilva]

I've published v19 of the chart. Thanks for all your work here!

[fhacloid]

You're welcome, feel free to ping if you need help on a related subject !

It was a pleasure to contribute ;)