Changelog

4.0.0

BREAKING CHANGE — PostgreSQL major version upgrade requires manual migration.

What changed

Remove automatic PostgreSQL migration hooks (postgresqlMigration.enabled): the pre-upgrade dump job and post-upgrade restore job have been removed entirely. These hooks proved too fragile in production due to OOM kills on large query_results tables, FK constraint violations during restore, and the requirement for superuser privileges for --disable-triggers. PostgreSQL major version migration must now be performed manually before running helm upgrade.
Remove rbac-postgres-migration templates: the Role and RoleBinding created for the migration hooks are no longer needed and have been removed.
Remove postgresqlMigration values block: the postgresqlMigration.enabled and postgresqlMigration.storage.pvcName values no longer exist. Remove them from your values.yaml or Helm --set flags before upgrading.

Deprecation notice

Chart versions 3.1.x and 3.2.x are deprecated. Do not use them for new upgrades — the automatic migration hooks in those versions are unreliable. Upgrade directly to 4.0.0 and follow the manual migration steps below.

Manual PostgreSQL migration (required when upgrading from chart ≤ 3.0.x)

If you are running chart 3.0.x (PostgreSQL 15) and upgrading to 4.0.0 (PostgreSQL 18), you must migrate your data manually. The steps below use a plain-SQL dump (.sql) rather than custom format — this gives you full control to strip problematic content before restoring.

Replace <release>, <namespace>, <pg-user>, <pg-database> with your actual values throughout.

Step 1 — Collect credentials

# PostgreSQL superuser password (needed to drop/create DB and restore as postgres)
PG_ADMIN_PASS=$(kubectl get secret <release>-postgresql -n <namespace> \
  -o jsonpath='{.data.postgres-password}' | base64 -d)

# Redash app user password
PG_REDASH_PASS=$(kubectl get secret <release>-postgresql -n <namespace> \
  -o jsonpath='{.data.password}' | base64 -d)

# Redis password
REDIS_PASS=$(kubectl get secret <release>-redis -n <namespace> \
  -o jsonpath='{.data.redis-password}' | base64 -d)

Step 2 — Dump from old PostgreSQL (PG 15, plain SQL format)

# Dump the full schema + data as plain SQL, excluding query_results data.
# query_results is a pure query cache — it's large, causes OOM on restore, and is regenerated automatically.
kubectl exec -n <namespace> <release>-postgresql-0 -- \
  env PGPASSWORD="$PG_REDASH_PASS" pg_dump \
    -U <pg-user> -d <pg-database> \
    --no-owner --no-acl \
    --exclude-table-data=query_results \
    -f /tmp/redash-dump.sql

kubectl cp <namespace>/<release>-postgresql-0:/tmp/redash-dump.sql ./redash-dump.sql

Step 3 — Prepare the dump file (run locally, once)

# PG 12+ removed SET default_with_oids — comment it out
sed -i '' 's/^SET default_with_oids = false;/-- SET default_with_oids = false;/' redash-dump.sql

# Comment out the FK constraint that references query_results.
# Since query_results data is excluded, this constraint will fail on restore.
# We add it back manually after NULLing the dangling references (Step 8).
grep -n 'queries_latest_query_data_id_fkey' redash-dump.sql
# Note the line number(s), then comment them out:
sed -i '' '/queries_latest_query_data_id_fkey/s/^/-- /' redash-dump.sql

Step 4 — Scale down Redash app (keep PostgreSQL and Redis running)

kubectl scale deploy -n <namespace> --replicas=0 --all

# Wait until only the database pods remain
kubectl get pods -n <namespace> -w

Step 5 — Drop and recreate the database

Connect as postgres to the postgres database (not redash) to avoid "database is being accessed by other users" errors.

kubectl exec -n <namespace> <release>-postgresql-0 -- \
  env PGPASSWORD="$PG_ADMIN_PASS" psql -U postgres -d postgres \
    -c "SELECT pg_terminate_backend(pid) FROM pg_stat_activity WHERE datname='<pg-database>' AND pid <> pg_backend_pid();" \
    -c "DROP DATABASE <pg-database>;" \
    -c "CREATE DATABASE <pg-database> OWNER <pg-user>;"

Step 6 — Copy dump into pod and restore

kubectl cp redash-dump.sql <namespace>/<release>-postgresql-0:/tmp/dump.sql

kubectl exec -n <namespace> <release>-postgresql-0 -- \
  env PGPASSWORD="$PG_REDASH_PASS" psql -U <pg-user> -d <pg-database> -f /tmp/dump.sql

ERROR: must be owner of extension plpgsql is harmless — safe to ignore.

Step 7 — Upgrade the chart

helm upgrade <release> <repo>/redash --version 4.0.0 \
  --namespace <namespace> \
  -f values.yaml \
  --wait --timeout 15m

Scale back down immediately after upgrade so migrations can run cleanly:

kubectl scale deploy -n <namespace> --replicas=0 --all

Step 8 — Fix dangling FK references

queries.latest_query_data_id points to query_results rows that were not restored. NULL them out so the FK constraint can be added.

kubectl exec -n <namespace> <release>-postgresql-0 -- \
  env PGPASSWORD="$PG_REDASH_PASS" psql -U <pg-user> -d <pg-database> \
    -c "UPDATE queries SET latest_query_data_id = NULL
        WHERE latest_query_data_id IS NOT NULL
          AND latest_query_data_id NOT IN (SELECT id FROM query_results);"

kubectl exec -n <namespace> <release>-postgresql-0 -- \
  env PGPASSWORD="$PG_REDASH_PASS" psql -U <pg-user> -d <pg-database> \
    -c "ALTER TABLE ONLY public.queries
          ADD CONSTRAINT queries_latest_query_data_id_fkey
          FOREIGN KEY (latest_query_data_id) REFERENCES public.query_results(id);" || true

Step 9 — Run Redash schema migrations

The new Redash version may have schema changes that Alembic needs to apply. Pass the connection URLs explicitly to avoid environment variable lookup issues.

kubectl exec -n <namespace> <release>-postgresql-0 -- \
  env PGPASSWORD="$PG_REDASH_PASS" psql -U <pg-user> -d <pg-database> \
    -c "SELECT version FROM alembic_version;"
# Should show current migration head after upgrade

# Find a running server pod and run migrations
kubectl exec -n <namespace> <redash-server-pod> -- \
  env REDASH_REDIS_URL="redis://:<redis-pass>@<release>-redis-master:6379/0" \
      REDASH_DATABASE_URL="postgresql://<pg-user>:<pg-pass>@<release>-postgresql:5432/<pg-database>" \
  /app/manage.py db upgrade

Step 10 — Scale up and verify

kubectl scale deploy -n <namespace> --replicas=1 --all
kubectl get pods -n <namespace> -w

# Verify data survived
kubectl exec -n <namespace> <release>-postgresql-0 -- \
  env PGPASSWORD="$PG_REDASH_PASS" psql -U <pg-user> -d <pg-database> \
    -c "SELECT count(*) FROM queries;"

# Check server logs
kubectl logs -n <namespace> deploy/<release>-redash-server --tail=50

query_results rows will be repopulated automatically as users re-run their queries.

Known gotchas

Issue	Fix
`SET default_with_oids` error	Comment it out in the dump (removed in PG 12+)
OOM kill during `COPY query_results`	Use `--exclude-table-data=query_results` — it's just cache
FK violation on `queries_latest_query_data_id_fkey`	Comment out in dump; re-add after NULLing dangling refs (Step 8)
`cannot drop currently open database`	Connect to `postgres` DB as `postgres` superuser, not to `redash`
`database being accessed by other users`	Scale down all Redash deploys first (Step 4)
`manage.py` can't find Redis/Postgres	Pass `REDASH_REDIS_URL` and `REDASH_DATABASE_URL` explicitly (Step 9)
`cryptography.fernet.InvalidToken` on data sources	`REDASH_SECRET_KEY` must match the key used by the old Redash to encrypt credentials — ensure it is set correctly in your Helm values

3.2.0

Upgrade Redash from v24.04.0-dev to v25.8.0 (latest stable)
Update default image repository from redash/preview to redash/redash for stable releases
Upgrade PostgreSQL dependency from ^15.2.0 to ^18.2.0 (latest: 18.2.4)
Upgrade Redis dependency from ^19.1.0 to ^24.1.0 (latest: 24.1.3)
NEW: Added optional automatic PostgreSQL migration hooks for major version upgrades (15→18)
- Enable with postgresqlMigration.enabled: true and configure a PVC for storage
- See README for details
See upgrade notes in README

CRITICAL Upgrade Notes:

Always backup your PostgreSQL database before upgrading
Redash schema migrations will run automatically via Helm hooks (handles Redash app schema changes)
PostgreSQL version upgrade (15→18):
- Option 1: Enable automatic migration hooks (requires PVC) - see README
- Option 2: Manual migration using pg_dump/pg_restore - see README
PostgreSQL 15 → 18 is a major version jump requiring data migration
Redis 19 → 24 upgrade should be automatic, but test thoroughly
Test in staging environment first, especially if upgrading from v24.x
Review Bitnami PostgreSQL and Redis chart changelogs for breaking changes

3.0.1 (unreleased)

Change scheduler deployment strategy type to Recreate. (#121)

3.0.0

Initial release supporting Redash v10.x
See upgrade notes in README

2.4.0

Final v2.x release (except for critical fixes)
Final release to support Redash v8.x
Redis password is now required
Kubernetes minimum version increased to v19.x
Helm 2 depreciated, will be removed in a future version
Supports stable Ingress API
Add support for SQL Alchemy pool pre-ping configuration
Add support for configurable worker pod labels
Expand configurability of install/upgrade hooks

2.3.0

Added externalPostgreSQLSecret / externalRedisSecret
Depreciated envSecretName (plan to remove in 3.0.0 chart)
Updated docs to make defaults clearer

2.2.0

Update docs to Helm Docs 1.4+ format
Removed duplicated env params that already come from "redash.env"
Updated Redash environment variables and associated docs
Updated CI to use Helm v3.4.1

2.1.0

Added redash.samlSchemeOverride
Added envSecretName
Expanded service configuration: .service.annotations and .service.loadBalancerIP
Moved postgresql and redis charts to Bitnami repo and update to latest patch release versions
Updated CI to use Helm v2.16.12 and v3.3.4
Updated CI to drop k8s v1.15, add v1.17, v1.18 and v1.19
Improved CI release publishing flow

2.0.0

Made secrets required, rather than auto-generating to avoid them changing on upgrade
Variable additions and template updates to bring in line with Helm 3 chart template
Add support for external secrets
Extended CI testing for multiple k8s and Helm versions
Extended CI testing for testing major version upgrades
Moved install and upgrade logic to Helm hooks
Added basic connectivity test hook for helm test
Created values for each environment variable accepted by Redash

1.2.0

Upgrade Redash to 8.0.2.b37747
Upgrade PostgreSQL chart (the old version used depreciated APIs) and image tag
Upgrade Redis chart

1.1.0

Initial release of chart on getredash namespace
Upgrade Redash to 8.0.1.b33387
Add support for external Redis server
Add CI linting/testing and release automation

Pre-release

For pre-release versions please see the pull request where this was originally developed.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Changelog

4.0.0

What changed

Deprecation notice

Manual PostgreSQL migration (required when upgrading from chart ≤ 3.0.x)

3.2.0

3.0.1 (unreleased)

3.0.0

2.4.0

2.3.0

2.2.0

2.1.0

2.0.0

1.2.0

1.1.0

Pre-release

FilesExpand file tree

CHANGELOG.md

Latest commit

History

CHANGELOG.md

File metadata and controls

Changelog

4.0.0

What changed

Deprecation notice

Manual PostgreSQL migration (required when upgrading from chart ≤ 3.0.x)

3.2.0

3.0.1 (unreleased)

3.0.0

2.4.0

2.3.0

2.2.0

2.1.0

2.0.0

1.2.0

1.1.0

Pre-release