Troubleshooting guides
Step-by-step checks for agents that will not connect, missing heartbeats, source connections, empty discovery, certification, provisioning and billing.
Support
Help yourself first — support is here if you need it
Search the help, follow a troubleshooting guide or look up an error code. Most problems can be solved without waiting for anyone.
Search help
Start here
Error codes
Every agent, connector and control-plane error code, what it means and what to do. The list is checked against the product source so it stays complete.
Documentation
Installation, the agent protocol, the API, the CLI and operator runbooks.
Status and health
Is there a public status page?
Not yet. DataNivra does not operate a public status page today. Use the health checks below, and the agent’s own state in the console.
How do I check control-plane health?
GET /v1/health on your control-plane address returns 200 with status OK when its dependencies answer, and 503 with status DEGRADED otherwise. GET /v1/health/live is a liveness check. Neither needs a login or reveals customer data.
How do I check an agent’s health?
The console shows each agent as ACTIVE, DEGRADED (for example clock skew) or OFFLINE (no heartbeat for 180 seconds by default). On the host, `datanivra-agent check` validates the configuration and the JSON logs carry value-free codes.
Billing FAQ
What does the free trial include?
The Free Trial runs for 14 days and uses the real workflow: a synthetic sandbox plus one agent in your own environment, with limits on sources, jobs and datasets. No sales call is required to start.
What happens when my trial ends?
Your organisation switches to a read-only state and new jobs are blocked. Nothing is deleted immediately: configuration is retained according to the data retention policy, and you can upgrade from Billing & Plan to continue.
How do I pay for Team or Business?
From Billing & Plan in the console. Payment is taken on the payment provider’s hosted checkout pages (Stripe), so card details are never entered on or stored by DataNivra. Prices are shown there, not invented on this site.
Can I cancel or change plans?
Yes. Upgrade, downgrade or cancel from Billing & Plan without contacting anyone. Cancelling stops renewal; your organisation then follows the same read-only and retention rules as an ended trial.
Can we pay by invoice or purchase order?
Enterprise and Private Cloud / Self-Hosted plans are contracted: quote, order form or purchase order, then invoice. See the enterprise page for the procurement path.
What is metered for billing?
Only safe metadata: jobs, processing time, aggregate volumes, agents, sources, datasets and packs. No source rows are ever uploaded for metering or licence enforcement.
Plan details are on the pricing page; contract billing is on the enterprise page.
Security FAQ
What does DataNivra Cloud receive?
Control metadata (ids, states, schema/table/column names, error codes), aggregate metrics (counts, sizes, durations), evidence metadata (checksums, gate outcomes) and secret references — never source values, rows, samples, query results or credentials.
Does the agent need inbound network access?
No. The agent only makes outbound HTTPS calls to the control plane. The control plane never connects into your network.
Where are database credentials and masking keys kept?
In your own secret store. DataNivra stores only references such as env://NAME or vault://path#key; the agent resolves them locally and the values never leave your environment.
What stops the agent sending data by mistake?
EgressGuard checks every outgoing message against allow-listed schemas and content detectors and refuses anything that looks like data. Refusals are logged locally without the payload; list them with `datanivra-agent violations` (codes and hashes only). The control plane applies the same detectors on ingress.
How can we verify these claims ourselves?
The trust page describes checks you can run: canary values, egress firewall rules, proxy inspection, the violation log, read-only proofs and audit-chain verification.
More detail: Security, customer-resident architecture and how to verify our claims yourself.
Still stuck? Contact support
Include the error code, the job, dataset or agent id and the time it happened. Never send credentials, enrollment tokens, secret values or source data — support never needs them.