JanusGraph discovery must reach the Gremlin server, not stop at the product name

JanusGraph stores graph data and exposes a query service through Gremlin Server. The web surface may show a title or a diagnostic page while the actual graph authority sits behind an HTTP or WebSocket endpoint. A product count gives an inventory lead, not a statement about authentication or graph contents.

On November 15, 2026, a count-only ZoomEye probe used sub_type=all, page 1, and page size 1. app="JanusGraph" returned 0, title="JanusGraph" returned 20, and http.body="JanusGraph" returned 237. The title and body matches may include documentation and management pages. They do not prove that a graph database is accepting queries on that host.

Find the query endpoint

JanusGraph Server is configured through a Gremlin Server YAML file. It can serve HTTP and WebSocket connections, and the security handler must be configured deliberately. Find the server configuration in the deployment, then compare the active port, authentication handler, TLS setting, and allowed origin policy with the exposed endpoint.

An HTTP or WebSocket endpoint that accepts arbitrary Gremlin can read and modify the graph. Treat query access as data-plane access, not as a read-only telemetry feature. If the service supports authorization, verify that the authenticated subject cannot escape its tenant or traverse graph edges outside its assigned scope.

Review the graph behind it

JanusGraph can use external storage and index backends. The graph service may hold credentials for Cassandra, HBase, Elasticsearch, or another system. A management-plane compromise can therefore expose more than the graph vertices. Inventory the storage backend, the index backend, the service account, and the network path between them.

Backups need the same treatment. A graph backup can include identities, relationships, properties, and query history. Confirm that the backup destination is not served by the public web endpoint and that restoration uses a separate credential boundary.

Validate a candidate

For each title or body match, identify the owner and confirm the service with a permitted, non-invasive request. Do not submit a query that modifies data or enumerates sensitive vertices. Record the endpoint, TLS, authentication handler, server version, backend type, and owner. If the endpoint cannot be verified, record it as unconfirmed.

The application and body counts are useful because they separate a broad product mention from a narrower page signature. Keep the exact queries and time. A zero application count is a fingerprint gap, not a zero-deployment conclusion.

What to operate

Put the Gremlin service behind a network boundary and an identity-aware proxy where possible. Require TLS for remote clients. Disable unused protocols. Log authentication identity, query hash or bounded query metadata, result count, and failure reason, not full graph values.

JanusGraph's value is the graph, and the graph is often a map of business relationships. Inventory begins with the product name but ends with the query endpoint and the backend credentials behind it.

References

• JanusGraph Server configuration
• JanusGraph operations and authentication
• JanusGraph configuration reference