Operations (planned)

Planned. Self-hosted deployment is not available yet. Commands starting with provantix are the planned interface and may change; the files, services and standard commands are as written.

You run it; this page is what to keep an eye on and how to do the routine jobs.

Health

systemctl status provantix                       # the service
curl -sS http://127.0.0.1:8787/health            # {"ok":true,"mode":"saas"}
curl -sS http://127.0.0.1:8787/api/licence       # licence health: ok, warning, blocked
df -h /var/lib/provantix                         # disk

Watch these, and alert on them in your own monitoring:

What Alert when
service not active
/health not 200
licence health warning (act before it becomes blocked)
disk free under 20%
certificate under 30 days to expiry
backup the latest is older than your target

Logs

journalctl -u provantix -f                       # follow
journalctl -u provantix --since "1 hour ago" --no-pager

Records and their contents are not written to the log. The log holds startup, migration, mail and error messages.

What to back up

Both halves, together. A database backup without the data directory cannot sign new roots, and a data directory without the databases has no evidence.

  1. The databases: the platform database and one database per organization.
  2. The data directory (LEDGER_DATA_DIR): the deployment's signing key, export keys and per-organization state. The signing key is the most important file. Losing it means roots can no longer be extended under the same key.
  3. The settings file (/etc/provantix/env): it holds the secrets, including LEDGER_PLATFORM_SECRET, which organization tokens are derived from, and LEDGER_CREDENTIAL_KEY if you set it. The credentials of destination connections are encrypted with that key. If it is lost, those credentials cannot be read and each connection has to be connected again; no evidence is lost.
# the platform database
pg_dump -Fc "$PLATFORM_DATABASE_URL" -f platform.dump

# each organization's database
psql "$PLATFORM_DATABASE_URL" -Atc "SELECT tenant_db FROM orgs" | while read db; do
  pg_dump -Fc "${PLATFORM_DATABASE_URL%/*}/$db" -f "$db.dump"
done

# the data directory and the settings file
sudo tar -czf provantix-data.tgz /var/lib/provantix /etc/provantix/env

Store backups off the machine, encrypted, and keep the signing key's backup under stricter control than the rest. Test a restore before you rely on it, and again after each upgrade of this procedure. The planned sudo provantix backup will do these steps and check the result.

Restore

Restore onto the same machine or a rebuilt one: install the same version, stop the service, restore the databases and the data directory, put the settings file back, start the service. A rebuilt or moved machine needs a new activation key from us, because the licence ties a deployment to its machine; see Licensing.

Upgrade

Releases are versioned. The planned upgrade does the same preflight, takes a backup, applies the migrations, restarts, and rolls back if the self-test fails:

curl -fsSLO https://get.provantix.tech/releases/<version>/provantix-<version>-linux-x86_64.tar.gz
# verify the checksum and signature as in "Install", then:
tar -xzf provantix-<version>-linux-x86_64.tar.gz && cd provantix-<version>
sudo ./install.sh upgrade

Migrations run automatically at start, for the platform database and each organization's, and there is no automatic rollback of a migration: the backup is the way back. Upgrade within the support window (the latest two minor versions), and check the compatibility note on the release page for the Node SDK batch format.

Support bundle

When you want help, generate a bundle yourself and send it:

sudo provantix support-bundle --out /tmp/provantix-support.tgz

It holds versions, health, the preflight report, configuration with secrets removed, and counters. It never holds records or content. Open it before you send it; it is a normal archive. We use it to help; it gives us no access.

Provantix has no remote login, shell or tunnel into your deployment. If you want us to look at the system, share your screen or grant access yourself, for that session.

Sanitized logs

The deployment also sends sanitized usage and error logs to Provantix on the same hourly schedule as the beacon, from a fixed allow-list of fields (counts, timings, versions and error classes). Your security team reviews the allow-list before go-live, and the plan is for the same list to be shown in the app so you can see exactly what leaves. Nothing outside the list can be sent.

Uninstall

sudo provantix uninstall                  # removes the service and packages, leaves your data
sudo provantix uninstall --purge-data     # also removes the databases and data directory: irreversible

The default leaves the databases and the data directory untouched. Export what you need to keep first, and tell us so we stop expecting beacons.

When something is wrong

Symptom First thing to check
People cannot sign in; the screen mentions the licence Licensing: can the machine reach Provantix, and is the clock right?
Agents get HTTP 402 The licence is lapsed or revoked: contact us. Batches stay queued.
No sign-in code arrives Email: the relay, the sender, the log line SMTP configured.
The browser warns about the certificate Domains and TLS: expiry or a missing authority.
An alert webhook fails Alerts: private addresses are refused until the allow-list ships.
The service will not start after an upgrade Read journalctl -u provantix; restore from the backup taken by the upgrade.