Streaming Admin API

Redpanda Streaming: Admin API

The Redpanda Admin API is an HTTP interface served on port 9644 (default) with base path /v1. It lets you manage brokers, partitions, cluster configuration, security, features, tiered storage, and diagnostics. All operations that rpk performs against the cluster go through this API.

Starting in Redpanda v25.3, new endpoints are served as ConnectRPC services alongside the legacy /v1 REST endpoints — both on port 9644. Legacy endpoints remain fully supported.

Quickstart

ADMIN=http://localhost:9644

# List all brokers in the cluster
curl "$ADMIN/v1/brokers"

# Get cluster health overview
curl "$ADMIN/v1/cluster/health_overview"

# Get cluster config schema (describes every cluster config property)
curl "$ADMIN/v1/cluster_config/schema"

# Get current cluster config (non-default values only)
curl "$ADMIN/v1/cluster_config"

# Get the full cluster config including defaults
curl "$ADMIN/v1/cluster_config?include_defaults=true"

# --- With Basic auth (when authentication is enabled) ---
curl -u admin:password "$ADMIN/v1/brokers"

# --- With Bearer token ---
curl -H "Authorization: Bearer <token>" "$ADMIN/v1/brokers"

# Decommission broker 3
curl -u admin:password -X PUT "$ADMIN/v1/brokers/3/decommission"

# Check decommission progress
curl -u admin:password "$ADMIN/v1/brokers/3/decommission"

# Alter cluster config: set log_compression_type
curl -u admin:password -X PUT "$ADMIN/v1/cluster_config" \
  -H "Content-Type: application/json" \
  -d '{"upsert": {"log_compression_type": "snappy"}, "remove": []}'

# Get features and cluster logical version
curl "$ADMIN/v1/features"

# Get license information
curl -u admin:password "$ADMIN/v1/features/license"

Base URL and Port

http://<broker-address>:9644/v1/<path>

All brokers in the cluster serve the Admin API. Any broker can answer read requests; write operations that mutate cluster state are forwarded internally to the controller leader. You can target any broker with any request.

Authentication

Authentication is optional — it mirrors the authentication configuration of the cluster. When enabled:

• Basic auth: curl -u <user>:<password> ...
• Bearer token: curl -H "Authorization: Bearer <token>" ...
• mTLS: Provide a client certificate and key when TLS is configured with require_client_auth: true.

Most write endpoints and some read endpoints require superuser privileges. Some endpoints are public (no auth required even when auth is enabled).

See auth-and-connection.md for full details.

Endpoint Groups

Key Operations by Use Case

Broker Lifecycle

# List all brokers
curl "$ADMIN/v1/brokers"

# Get a specific broker
curl "$ADMIN/v1/brokers/1"

# Put broker into maintenance mode (drains leadership before restart)
curl -u admin:password -X PUT "$ADMIN/v1/brokers/2/maintenance"

# Remove from maintenance mode
curl -u admin:password -X DELETE "$ADMIN/v1/brokers/2/maintenance"

# Decommission (permanently remove broker from cluster)
curl -u admin:password -X PUT "$ADMIN/v1/brokers/3/decommission"

# Recommission (cancel a decommission in progress)
curl -u admin:password -X PUT "$ADMIN/v1/brokers/3/recommission"

# Check pre-restart safety
curl "$ADMIN/v1/broker/pre_restart_probe"

Partition Operations

# Get partitions hosted on this node (node-local only)
curl "$ADMIN/v1/partitions"

# Get detailed state for a specific partition (namespace/topic/partition)
curl "$ADMIN/v1/partitions/kafka/my-topic/0"

# Get all partitions across the cluster
curl "$ADMIN/v1/cluster/partitions"

# Move partition replicas to different nodes
curl -u admin:password -X POST "$ADMIN/v1/partitions/kafka/my-topic/0/replicas" \
  -H "Content-Type: application/json" \
  -d '[{"node_id": 1, "core": 0}, {"node_id": 2, "core": 0}, {"node_id": 3, "core": 0}]'

# Cancel an ongoing reconfiguration
curl -u admin:password -X POST "$ADMIN/v1/partitions/kafka/my-topic/0/cancel_reconfiguration"

# Trigger on-demand partition rebalance
curl -u admin:password -X POST "$ADMIN/v1/partitions/rebalance"

# Check partition balancer status
curl "$ADMIN/v1/cluster/partition_balancer/status"

Cluster Configuration

# Full config schema — inspect property types, defaults, needs_restart, visibility
curl "$ADMIN/v1/cluster_config/schema" | jq '.["log_retention_ms"]'

# Read a single config property
curl "$ADMIN/v1/cluster_config?key=log_retention_ms"

# Alter config
curl -u admin:password -X PUT "$ADMIN/v1/cluster_config" \
  -H "Content-Type: application/json" \
  -d '{"upsert": {"log_retention_ms": 86400000}, "remove": []}'

# Dry-run a config change (validates without applying)
curl -u admin:password -X PUT "$ADMIN/v1/cluster_config?dry_run=1" \
  -H "Content-Type: application/json" \
  -d '{"upsert": {"log_retention_ms": -1}, "remove": []}'

# Check which nodes need restart after a config change
curl "$ADMIN/v1/cluster_config/status"

Health and Diagnostics

# Cluster health overview (is_healthy, leaderless/under-replicated counts, nodes down)
curl "$ADMIN/v1/cluster/health_overview"

# Get CPU profile (wait 5 seconds for samples)
curl "$ADMIN/v1/debug/cpu_profile?wait_ms=5000"

# Disk stat for data directory
curl "$ADMIN/v1/debug/storage/disk_stat/data"

# Controller log status (last_applied_offset, committed_index)
curl "$ADMIN/v1/debug/controller_status"

# Self-test: start disk + network benchmarks
curl -u admin:password -X POST "$ADMIN/v1/debug/self_test/start" \
  -H "Content-Type: application/json" -d '{}'
# Query results
curl "$ADMIN/v1/debug/self_test/status"

ConnectRPC Endpoints (v25.3+)

# ConnectRPC endpoints use POST and fully-qualified service paths
curl -u admin:password \
  -X POST "http://localhost:9644/redpanda.core.admin.v2.ShadowLinkService/FailOver" \
  -H "Content-Type: application/json" \
  -d '{"name": "<shadow-link-name>", "shadowTopicName": "<shadow-topic-name>"}'

ConnectRPC endpoints also accept binary Protobuf bodies with Content-Type: application/proto. SDKs can be generated from the Buf Schema Registry at buf.build/redpandadata/core.

Enterprise Features

Most of Redpanda's enterprise differentiators are configured and operated through this same Admin API — cluster config (PUT /v1/cluster_config), the cloud_storage endpoints, the RBAC /v1/security/roles* endpoints, and the ConnectRPC ShadowLinkService. These require a valid Enterprise license.

# Is an enterprise feature enabled without a valid license?
curl -u admin:secret "$ADMIN/v1/features/enterprise"   # license_status, violation, features[]
curl -u admin:secret "$ADMIN/v1/features/license"       # loaded license details

# Enable Continuous Data Balancing (Enterprise)
curl -u admin:secret -X PUT "$ADMIN/v1/cluster_config" \
  -H "Content-Type: application/json" \
  -d '{"upsert": {"partition_autobalancing_mode": "continuous"}, "remove": []}'

# Shadow Linking / DR failover (ConnectRPC, v25.3+)
curl -u admin:secret -X POST "$ADMIN/redpanda.core.admin.v2.ShadowLinkService/FailOver" \
  -H "Content-Type: application/json" -d '{"name": "dr-east"}'

Features and their cluster-config / topic keys: Tiered Storage (cloud_storage_enabled, redpanda.remote.*), Cloud Topics (cloud_topics_enabled), Iceberg Topics (iceberg_enabled, redpanda.iceberg.mode/.delete/.partition.spec/.target.lag.ms/.invalid.record.action), Continuous Data Balancing (partition_autobalancing_mode=continuous, partition_autobalancing_*, core_balancing_continuous), Shadow Linking (ShadowLinkService), Remote Read Replicas (cloud_storage_enable_remote_read, redpanda.remote.readreplica), Audit Logging (audit_*), RBAC (/v1/security/roles*), OIDC/Kerberos (sasl_mechanisms, oidc_*, sasl_kerberos_*), Server-Side Schema ID Validation (enable_schema_id_validation), Leadership Pinning (default_leaders_preference, redpanda.leaders.preference), FIPS (fips_mode, node config). See enterprise-features.md for every nested key, grounded in source.

Reference Directory

• endpoints-overview.md: Full endpoint-group map with HTTP methods, paths, and what each returns — grounded in the Swagger 1.2 JSON specs.
• auth-and-connection.md: Base URL, port 9644, TLS configuration, Basic auth, Bearer tokens, mTLS, and the auth_level model (public / authenticated / superuser).
• brokers-and-partitions.md: Broker lifecycle (list, decommission, recommission, maintenance mode) and partition operations (state, move replicas, rebalance, force recover). Full curl examples and response schemas.
• cluster-config.md: Reading the config schema, getting and altering cluster configuration, the needs_restart flag, the difference between cluster config and node config, and the per-node status endpoint.
• debug-endpoints.md: The /v1/debug/* endpoints — cpu_profile, disk_stat, partition state, self-test, controller status, sampled memory profile — plus the debug_bundle HTTP API. curl examples and response field meanings.
• enterprise-features.md: Redpanda Enterprise differentiators driven through the Admin API, with every nested config key and topic property grounded in source: Tiered Storage / shadow indexing, Topic Recovery / Whole Cluster Restore, Remote Read Replicas, Cloud Topics, Iceberg Topics, Continuous Data Balancing (+ intra-broker core balancing), Leadership Pinning, Audit Logging, RBAC roles endpoints, OIDC/OAuthBearer/Kerberos auth, Server-Side Schema ID Validation, FIPS, and the ShadowLinkService ConnectRPC API for cross-cluster disaster recovery. Includes the license-violation remediation table (which config key disables each feature).