Skip to main content
Run a backup before source-load changes that can rewrite ClickHouse data — switching sources to replace mode, changing target schemas, or changing connector credentials for production sources — and on a regular schedule. All commands run from the installer directory.

​
Back up

./platinur backup [dir]
The target directory defaults to backups/ next to the CLI. Each backup produces two timestamped files:
  • platinur-state-<stamp>.tar.gz — an archive of the PLATINUR_HOME state directory: config, the portal metadata database, generated models, and dashboards.
  • platinur-env-<stamp> — a copy of .env. It contains secrets (PLATINUR_SECRET_KEY, the admin and ClickHouse passwords, the registry token), so store it safely.
Keep both. PLATINUR_SECRET_KEY encrypts stored credentials — state restored without the matching .env means re-entering every source credential.

​
ClickHouse data

The state backup does not include warehouse data. In the bundled profile, ClickHouse data lives in the platinur_clickhouse_data Docker volume (prefixed by PLATINUR_PROJECT_NAME if you changed it); snapshot it with your volume tooling for a full rollback point. External-ClickHouse deployments (EXTERNAL_CLICKHOUSE=true) run no local ClickHouse — back up the warehouse on your side with your cluster’s or ClickHouse Cloud’s own tooling. ./platinur backup covers Platinur state only.

​
Restore

./platinur restore /path/to/platinur-state-<stamp>.tar.gz
Restore stops the application services (control plane, worker, Prefect runner, Evidence), replaces PLATINUR_HOME with the archive’s contents, restarts the stack, and waits for the portal to come back. The replaced state is kept next to it as <state-dir>.pre-restore-<stamp> — delete it once you have verified the restore. Restore requires a .env in the installer directory and does not restore it for you. On a fresh host, copy the matching env backup into place first: 
cp /path/to/platinur-env-<stamp> .env
chmod 600 .env
Restore does not touch ClickHouse data. If warehouse data also changed, restore the bundled volume snapshot separately before restarting production workloads.

​
Restore drill

A backup you have never restored is a guess. Periodically restore the latest backup (on a spare host, or the same host during a maintenance window) and verify:
  1. ./platinur doctor passes every check.
  2. You can log in to the portal with the restored credentials.
  3. Configured sources and generated models are present.

​
Support bundle

./platinur support-bundle
This collects a host-side diagnostics archive, platinur-support-<stamp>.tar.gz, in the installer directory: the container list, the last 400 log lines from each service, and env.redacted — your .env with every line containing PASSWORD, SECRET, TOKEN, or API_KEY removed. This is the bundle to attach when reporting host or startup problems. For application-level diagnostics, the portal’s Monitoring page has its own one-click redacted support bundle.