Trust & Security
Telemetry Privacy Notice
VectorFlow Pulse is the anonymous telemetry receiver for VectorFlow. This notice describes exactly what heartbeat payloads contain, what they never contain, how long data is retained, and how to turn telemetry off.
What it collects
When telemetry is enabled in a VectorFlow instance, the instance sends a JSON payload to this server once every 24 hours. The payload contains:
- A random instance ID (generated on first opt-in, no link to any user or organisation)
- VectorFlow version
- Agent count, pipeline counts (active / paused / draft)
- Authentication method in use (“credentials” or “oidc”)
- Deployment mode (“docker”, “helm”, “bare”, “unknown”)
- Timestamps
What it does NOT collect
- Hostnames or IP addresses — the client IP is hashed at ingest and never stored as-is.
- Pipeline names, configurations, VRL code, or pipeline contents of any kind.
- User emails, names, or IDs.
- Source / sink endpoint URLs (Kafka, S3, syslog, etc.).
- Any data flowing through pipelines.
How it is used
Aggregate counts only. We use the data to understand which VectorFlow versions are in the wild, typical fleet sizes, and popular deployment modes — nothing more. No payload is ever shared per-instance publicly or with third parties.
Retention
Raw heartbeat rows are retained for 90 days, after which they are deleted. Derived aggregate statistics (version adoption curves, weekly active-instance counts) are kept indefinitely but contain no per-instance identifiers.
Opting out
In VectorFlow, go to Settings → Telemetry and toggle telemetry off. The instance will stop sending heartbeats immediately. No data is sent during or after the opt-out; previously received heartbeats age out within 90 days per the retention policy above.
Source
The VectorFlow client-side telemetry code is open source under AGPL-3.0. The full source of the Pulse receiver is available privately to operators on request.
Questions about this notice? privacy@vectorflow.sh
Return to Trust & Security overview.