Skip to main content
Use this reference at the preparation step of the Kubernetes or Docker Compose procedure. It applies to releases that include formbricks-authzed-prepare; verify that executable in the selected release image before maintenance. An upgrade from v5 to v6 first needs the dependencies in 6.0: Upgrade guide.

Preparation contract

  • Run only the selected target image, with AUTHZED_CONSISTENCY=fully_consistent.
  • The command checks the configuration and SpiceDB health, runs the app’s database migrations, prepares the permission schema, processes pending permission updates, repairs relationships and verifies the result.
  • A different non-empty schema is blocked. Review formbricks-authzed schema check and the release’s schema changes before passing --expected-current-digest <remoteDigest>.
  • Exit 0 and upgrade_prepared permit the next deployment step. Any nonzero result keeps maintenance active.

What runs where

You run the maintenance steps: stopping the app and workers, pausing controllers, taking backups and restarting. The preparation command in the release image runs the database migrations and prepares the permission data. formbricks-authzed-prepare does not stop anything, take backups or restart services for you. The --backup-confirmed and --writers-stopped flags are your confirmation that you’ve done those steps; the command doesn’t check them. Do them first. The command reuses the image’s existing migration runner and formbricks-authzed upgrade prepare and upgrade check implementations. It does not automatically prune unexplained relationships.

Before the maintenance window

  • Rehearse with an isolated copy of the installation. Measure migrations, backfill, verification, and restoration time; use those measurements to communicate the maintenance window.
  • Pin the target Formbricks image by digest and the matching chart/Compose files. Preserve the current images, chart, values, secrets, and volume names. Do not combine this change with a PostgreSQL major upgrade, a storage migration, or unrelated Hub/Cube changes.
  • Treat the Helm chart version and application image version as separate pins. A chart-only patch can keep the same Formbricks appVersion; verify both values and the rendered image before upgrading.
  • Prepare private SpiceDB connectivity, dedicated datastore credentials, its reviewed version, and completed datastore migrations. Retain existing operator ownership; never install a second operator over an existing cluster-wide operator. See SpiceDB operations.
Identify all writers, including Formbricks replicas, Hub/Hub workers, workflow workers, scheduled jobs, import jobs, integrations, and administrative scripts. Include workloads outside the application namespace or Compose project that write to the same databases. Identify Argo CD/Flux, image updaters, autoscalers, and other controllers that can recreate those workloads. Record their original state for restoration. Do not print Secret contents or expanded Compose configuration into shared CI logs. Prepare and render the exact target manifests before maintenance.

Docker and one-click

Keep the same Compose project name, database images, and volumes. Prepare a target Compose file that retains customized SMTP, storage, networking, and Traefik settings. Pin formbricks, formbricks-migrate, authzed-ops, and authzed-initialize to the same target image digest. Do not copy a fresh-install Compose file over an existing installation without reviewing those differences. After stopping the old project’s writers and verifying external writers/controllers are stopped, start only the required infrastructure from the reviewed target configuration. Never use down -v. With infrastructure already healthy, run the existing operations service without starting dependencies:
Supply your normal --project-name, --env-file, and -f options when customized. The operator service must receive the application database URL, migration credentials when different, and the six AuthZed client variables. The distributed default uses the bundled PostgreSQL database; customize it when your installation uses a managed database. No database or SpiceDB port needs public exposure. For a pre-provisioned external SpiceDB database, apply the release-matched external Compose override before starting spicedb-migrate or running this preparation. Keep COMPOSE_FILE in .env so validation, preparation, checks, and cutover all use the same override; if you pass -f, specify both files in the same order every time. Its URI and any private CA mount must reach both SpiceDB migration and runtime services. Confirm docker compose config --services omits authzed-db-bootstrap, then use docker compose ps -a to verify a completed migration and healthy SpiceDB. The database administrator provisions the database and scoped login once, outside Compose; no administrator URL belongs in the application stack. Include the external SpiceDB database in the coordinated backup before preparation and in the rollback plan. After successful preparation, a one-click installation also sets FORMBRICKS_AUTHZED_V6_MIGRATION_ACKNOWLEDGED=true before its explicit v6 update. Do not invoke an unattended update or move a floating image tag until the release-specific installation tests pass.

Helm and GitOps

Prepare SpiceDB before the application upgrade. Keep the current database/storage settings and existing Secrets. During maintenance, use an explicit, temporary Kubernetes Job with the target Formbricks image, the same database/AuthZed environment bindings, and this command:
Save it as formbricks-prepare-job.yaml. Set the image digest, and add resource limits, imagePullSecrets and network policy if your cluster needs them. Don’t make the root filesystem read-only: the migrations write to the container. For an external SpiceDB, set AUTHZED_ENDPOINT to its address and AUTHZED_INSECURE to "false". Preparation uses these settings:
  • DATABASE_URL
  • AUTHZED_ENABLED=true, AUTHZED_ENDPOINT, AUTHZED_TOKEN, AUTHZED_SYSTEM_KEY, AUTHZED_INSECURE, and AUTHZED_CONSISTENCY=fully_consistent
  • WEBAPP_URL, in every case. Preparation runs the database migrations, and the OAuth resource-link data migration (first released in 5.4.0) derives each MCP OAuth client’s resource link from it. If you upgrade from before 5.4.0 and have MCP OAuth clients, preparation stops with upgrade_database_migration_failed when WEBAPP_URL is missing.
Never put secrets on a command line or commit their values. If migrations use a different database credential, bind MIGRATE_DATABASE_URL as well as DATABASE_URL; the target image’s migration runner uses it when set. If the configuration is still incomplete, the Job reports upgrade_configuration_invalid with the names of the variables to fix. Wait for Job completion and inspect its sanitized result before running helm upgrade or resuming an Argo CD/Flux reconciliation. Preparation is not a Helm migration hook or a permanent chart component. Delete the Job and any temporary environment Secret after the upgrade is done.

When the permission schema changes

Some minor releases change the permission schema, for example 6.0 → 6.1. Preparation then refuses to replace the existing schema until you confirm which one you’re replacing:
  1. Run the same Job with command: ["formbricks-authzed"] and args: ["schema", "check"], using the target image, and a different name such as formbricks-v6-schema-check. It exits with code 2 (drifted), which is expected.
  2. Copy the remoteDigest value from its log.
  3. Run the preparation Job with the digest added: args: ["--backup-confirmed", "--writers-stopped", "--expected-current-digest", "<remoteDigest>"]. The remoteDigest value already starts with sha256:; copy it as it is.
On Docker Compose, run docker compose --profile authzed-ops run --rm --no-deps authzed-ops schema check, then add --expected-current-digest <remoteDigest> to the preparation command. Patch releases never change the schema.

Exit checklist

Keep maintenance active until all of these are true:
  • formbricks-authzed-prepare exited 0, including its two clean full audits, and the v6 migration acknowledgement is set only for this prepared upgrade.
  • The pods run the target image, and a fresh formbricks-authzed upgrade check exits 0.
  • Allowed and denied users, a scoped API key, organization/workspace lists, and a grant revocation work from a fresh browser session; an old tab recovers after a hard refresh.
Only then resume controllers and writers and remove the ingress maintenance page.

Results and recovery

Success prints one JSON result and exits 0:
Blocked preparation exits 2, without a restartRequired field. It never authorizes restarting workloads, including when arguments were rejected before any work ran. The code identifies the failed step: A blocked result can carry more fields, never values from your data:
  • "reason":"timeout" and timeoutSeconds: the step ran out its time limit. It did not report a failure, so the code names the step, not a problem in it. Raise the limit and run preparation again.
  • cause and retryable: the step’s own error code, such as authzed_unavailable, and whether that error is transient. Retry a "retryable":true result once its cause, such as SpiceDB availability, has recovered.
Preparation has five sequential timed steps, each with a 90-minute default limit. The example Job allows eight hours to cover their combined limits and startup/result overhead. Size both limits from your rehearsal; a longer Job deadline does not increase the per-step limit. Set FORMBRICKS_AUTHZED_PREPARE_STEP_TIMEOUT_SECONDS (60 to 86400) to change it. The graph step rewrites every organization’s relationships and then audits them all, so its runtime grows with the number of organizations. Newer images audit several organizations concurrently; an older sequential run took about 35 minutes for 50,000 organizations. Size the limit from your own rehearsal, and keep the Job’s activeDeadlineSeconds above the sum of its steps: a Job deadline stops the container before the wrapper can print a result. Interrupted containers can exit with a signal code rather than JSON; any nonzero exit remains blocked. The command never restarts a workload on success, failure, or interruption. Retry preparation with the same image after correcting the cause, provided the interrupted migration is safe to rerun. Never mark a failed Prisma migration applied without investigating it.

Diagnose migration failures privately

The wrapper deliberately does not offer raw output in shared Job logs: migration errors can contain credentials, SQL, or customer identifiers. In a restricted PostgreSQL operator session, inspect failed Prisma migration records in _prisma_migrations (migration_name, started_at, finished_at, rolled_back_at, and logs). Treat logs as sensitive; do not paste it into CI, tickets, or Slack. For failures before Prisma records the attempt, check connectivity and permissions in that restricted session. If a migration must be retried, first establish whether partial changes need repair, then use the target image’s node /home/nextjs/packages/database/dist/scripts/apply-migrations.js under maintenance. Capture any diagnostic output only in operator-controlled storage; never enable it on normal startup. If a migration fails shows how to find the real error, check what the failed migration left behind, and clear its failed record before retrying. After database changes, helm rollback or changing the image is not necessarily a safe rollback. Remain in maintenance and either complete the v6 preparation or restore the coordinated backup and matching v5 images/configuration. Resuming writes before restoration risks losing those new writes.