Self-hosted troubleshooting

Use this page if you followed Self-hosted Rejourney and something fails or behaves oddly. Commands are run from the repository root (where docker-compose.selfhosted.yml lives).

Start Here

Pick the symptom that best matches what you are seeing.

90-second triage

• ./scripts/selfhosted/deploy.sh status shows API and upload relay running.
• DNS for dashboard, API, ingest, and www points at this server.
• .env.selfhosted is the original file for this deployment.
• PUBLIC_API_URL and PUBLIC_INGEST_URL use the public HTTPS hosts.
• If using the Web SDK, the app origin is in Web allowed domains.
• If using external S3, the ingest-upload container can reach the S3 endpoint.

Fast Checks

Service status

./scripts/selfhosted/deploy.sh status

API logs

./scripts/selfhosted/deploy.sh logs api

Upload relay logs

./scripts/selfhosted/deploy.sh logs ingest-upload

Worker logs

./scripts/selfhosted/deploy.sh logs ingest-worker
./scripts/selfhosted/deploy.sh logs retention-worker
./scripts/selfhosted/deploy.sh logs alert-worker

1. Install or update fails before or during bootstrap

Symptoms

• bootstrap exits non-zero
• app services never become healthy
• status shows API or workers waiting on bootstrap
• install or update exits with Database authentication failed before bootstrap.

Checks

docker compose -f docker-compose.selfhosted.yml --env-file .env.selfhosted logs bootstrap

Common causes:

• bad DATABASE_URL
• credentials mismatch (e.g. from an earlier failed deployment)
• missing STORAGE_ENCRYPTION_KEY
• invalid S3 credentials
• broken external S3 endpoint URL
• on ARM64, missing image support (set DOCKER_DEFAULT_PLATFORM=linux/amd64 or use ./scripts/selfhosted/deploy.sh, which sets it when unset)

Recovery:

1. If you still have the original .env.selfhosted, restore it and run:

./scripts/selfhosted/deploy.sh update

1. If you do not need old data, wipe and reinstall:

./scripts/selfhosted/deploy.sh reset
./scripts/selfhosted/deploy.sh install

Schema / migration messages: On a normal install, the database starts empty and bootstrap sets everything up. If you restored Postgres from a backup into a new server but migration metadata is missing, or you pointed the stack at the wrong database, bootstrap may exit with an error about an inconsistent database instead of overwriting your data. Unless you are doing advanced recovery, fix DATABASE_URL and restore a consistent backup, or start from a clean volume. For deliberate migrate-only recovery, some setups use REJOURNEY_ALLOW_ORPHAN_DB_MIGRATE_ONLY=1 in .env.selfhosted (see maintainer docs or support before using this).

Fix

1. If you have the original .env.selfhosted, restore it and rerun:

./scripts/selfhosted/deploy.sh update

1. If you do not have the original .env.selfhosted, wipe and reinstall:

./scripts/selfhosted/deploy.sh reset
./scripts/selfhosted/deploy.sh install

update reruns schema, seed, and storage-endpoint sync. reset removes self-hosted containers and data volumes so a fresh install can generate new credentials safely.

2. Sessions are counted but Replay stays empty

What this usually means now

With the current architecture, this is usually one of two things:

• ingest-upload could not store the artifact bytes
• ingest-worker could not process an uploaded artifact

The device no longer uploads directly to MinIO/S3, so bucket reachability from the phone is no longer the main suspect.

Checks

./scripts/selfhosted/deploy.sh logs ingest-upload
./scripts/selfhosted/deploy.sh logs ingest-worker
./scripts/selfhosted/deploy.sh logs api

Look for:

• artifact.upload_received
• artifact.upload_stored
• artifact.retry
• artifact.failed
• session.reconciled
• session.finalized

Common causes

• wrong S3 credentials in .env.selfhosted
• external S3 bucket missing
• external S3 endpoint unreachable from Docker network
• upload relay unhealthy
• worker stuck retrying failed artifacts

Fix

• verify S3_* values
• if you changed storage config, rerun:

./scripts/selfhosted/deploy.sh update

3. Dashboard loads, but auth or API calls fail

Checks

• dashboard host DNS points to the server
• API host DNS points to the server
• ingest host DNS points to the server
• ports 80 and 443 are open
• Let's Encrypt has issued certificates

Inspect:

./scripts/selfhosted/deploy.sh logs traefik
./scripts/selfhosted/deploy.sh logs api

4. Web SDK config returns 403

The Web SDK enforces the project's allowed browser origins before it returns configuration or accepts ingest.

Checks:

• The app's browser origin is listed in the project's Web allowed domains.
• Include the host and port for local or staging apps, for example localhost:3100.
• PUBLIC_API_URL points at the public API host that browsers can reach.
• PUBLIC_INGEST_URL points at the public ingest upload relay host that browsers can reach.

After changing domain or ingest values in .env.selfhosted, rerun:

./scripts/selfhosted/deploy.sh update

5. TLS or certificate issues

Traefik manages certificates automatically.

Checks

dig example.com
dig api.example.com
dig ingest.example.com
dig www.example.com

Make sure all names resolve to the server running the stack.

If DNS was wrong during first install, fix DNS and rerun:

./scripts/selfhosted/deploy.sh update

6. External S3 works in CLI, but Rejourney cannot upload

Remember the upload path is server-side.

The important network path is:

• ingest-upload container -> your S3 endpoint

Test from the server by reviewing relay logs and confirming the endpoint/bucket/keys in .env.selfhosted.

If you changed them, rerun:

./scripts/selfhosted/deploy.sh update

7. Built-in MinIO install, but artifacts still fail

Checks

./scripts/selfhosted/deploy.sh logs minio
./scripts/selfhosted/deploy.sh logs minio-setup

The minio-setup one-shot should create the bucket named by S3_BUCKET.

If you changed the bucket name after first install, run:

./scripts/selfhosted/deploy.sh update

8. Billing pages show disabled billing

That is expected unless Stripe keys are configured.

The stack no longer disables billing because it is "self-hosted". It disables billing because Stripe is unconfigured.

If you do not set Stripe keys:

• billing UI stays in the self-hosted/unlimited state
• Stripe checkout and webhooks stay disabled

9. Storage endpoint in Postgres is wrong after changing .env.selfhosted

Run:

./scripts/selfhosted/deploy.sh update

The update path reruns bootstrap and resyncs the active storage_endpoints row.

10. Need to stop services without losing data

Use:

./scripts/selfhosted/deploy.sh stop

This stops containers only. It does not remove volumes.

11. Need deeper logs for one service

./scripts/selfhosted/deploy.sh logs api
./scripts/selfhosted/deploy.sh logs ingest-upload
./scripts/selfhosted/deploy.sh logs ingest-worker
./scripts/selfhosted/deploy.sh logs web