Skip to main content
Version: User Guides (BYOC)

VectorDB Audit Logs

Audit logging allows administrators to track and monitor user-driven operations and API calls on Zilliz Cloud clusters. This feature provides a detailed record of vector db activities, including vector searches, query execution, index management, and other data operations.

📘Notes
  • Audit logging is available only for Dedicated clusters in Enterprise projects or higher plan tiers.

  • Audit logging is supported only for Zilliz Cloud clusters running Milvus 2.5.x.

  • Audit logging can be forwarded to AWS S3, Azure Blob Storage, or Google Cloud Storage.

  • Enabling audit logging incur charges. For details, see Audit Log.

Overview

Audit logging tracks a wide range of operations on the data plane, including:

  • Search and Query Operations: Vector searches, hybrid searches, and query operations.

  • Data Management: Index creation, collection creation, partition management, and entity operations like insert, delete, and upsert.

  • System Events: User access attempts, authorization checks, and other predefined actions.

📘Notes

Cluster-level data jobs such as migration, backup, and restore do not generate audit logs. To view these activity records, refer to View Activities.

Audit logs are forwarded directly to a user-designated object storage bucket at regular intervals. Logs are stored in a structured file path and naming format for easy access and management:

  • File Path: /<Cluster ID>/<Log type>/<Date>

  • File Naming Convention: <File name><File name suffix> in the format HH:MM:SS-$UUID, where HH:MM:SS represents the time in UTC and $UUID is a unique random string. Example: 09:16:53-jz5l7D8Q.

Below are examples of audit log entries forwarded to a bucket:

  • Create Collection

    {
    "action": "CreateCollection",
    "cluster_id": "in01-0045a626277eafb",
    "connection_uid": 456912553983082500,
    "database": "default",
    "interface": "Grpc",
    "log_type": "AUDIT",
    "params": {
    "collection": "test_audit",
    "consistency_level": 2
    },
    "status": "Receive",
    "timestamp": 1742983070463,
    "trace_id": "216a8129c06fd3d93a47bd69fa0a65ad",
    "user": "key-hwjsxhwppegkatwjaivsgf"
    }
  • Create Index

    {
    "action": "CreateIndex",
    "cluster_id": "in01-0045a626277eafb",
    "connection_uid": 456912553983082500,
    "database": "default",
    "interface": "Grpc",
    "log_type": "AUDIT",
    "params": {
    "collection": "test_audit"
    },
    "status": "Receive",
    "timestamp": 1742983070645,
    "trace_id": "4402e7bfc498dd06be1408c7e6a7954d",
    "user": "key-hwjsxhwppegkatwjaivsgf"
    }
  • Drop Index

    {
    "action": "DropIndex",
    "cluster_id": "in01-0045a626277eafb",
    "connection_uid": 456912553983082500,
    "database": "default",
    "interface": "Grpc",
    "log_type": "AUDIT",
    "params": {
    "collection": "test_audit"
    },
    "status": "Receive",
    "timestamp": 1742983073378,
    "trace_id": "066ec33c3f55d3edbf7d01c6270024e2",
    "user": "key-hwjsxhwppegkatwjaivsgf"
    }

Refer to the Audit Log Reference for a detailed list of supported actions and corresponding log fields.

📘Notes

Audit logging will be directly forwarded to the object storage bucket configured during the data plane deployment.

To export the logs to your logging system for further analysis, contact us.