Backup & restore¶

Two things hold durable state: PostgreSQL (all records) and the attachment volume (uploaded files). Redis is a transient job queue and needs no backup.

PostgreSQL¶

Compose¶

# Backup
docker compose exec -T postgres pg_dump -U tessio tessio > tessio-$(date +%F).sql

# Restore (into an empty database)
docker compose exec -T postgres psql -U tessio -d tessio < tessio-2026-01-01.sql

Kubernetes¶

kubectl exec -n tessio statefulset/tessio-postgresql -- \
  pg_dump -U tessio tessio > tessio-$(date +%F).sql

For a managed Postgres, use your provider's backup tooling.

pgvector

Restores must target a database where the vector and pgcrypto extensions are available. The migrations create them, but a brand-new managed database may need pgvector installed first.

Attachments¶

Uploaded files live under /data/storage inside the api / all-in-one container:

Compose: the tessio_storage volume (tessio_aio_data for all-in-one). Back it up with docker run --rm -v tessio_storage:/data -v "$PWD":/backup alpine tar czf /backup/attachments.tgz -C /data .
Kubernetes: the tessio-storage PVC — snapshot it with your CSI driver, or kubectl cp the directory out of an api pod.

Restore checklist¶

Stop the app (or scale it to zero) so nothing writes during restore.
Restore PostgreSQL from your dump.
Restore the attachment volume.
Start the app — migrations run and are no-ops if the schema already matches.