test(get): reject range with part number (#2725)

Co-authored-by: houseme <housemecn@gmail.com>

.agents/skills/pr-creation-checker/SKILL.md

@@ -26,6 +26,8 @@ Use this skill before `gh pr create`, before `gh pr edit`, or when reviewing whe
 - Review the diff and summarize what changed.
 - Call out unrelated edits, generated artifacts, logs, or secrets as blockers.
 - Mark risky areas explicitly: auth, storage, config, network, migrations, breaking changes.
+- Scan the diff for newly added string literals and confirm whether they duplicate values already defined as constants/enums/typed wrappers in the same module or shared modules.
+- Treat introducing a new hardcoded literal where a project constant already exists as a likely regression risk; require either a refactor to reuse the constant or an explicit exception explanation in the PR body.
 
 3. Verify readiness requirements
 - Require `make pre-commit` before marking the PR ready.
@@ -82,6 +84,7 @@ Use this skill before `gh pr create`, before `gh pr edit`, or when reviewing whe
 - Return `BLOCKED` if required template sections are missing.
 - Return `BLOCKED` if the title/body is not in English.
 - Return `BLOCKED` if the title does not follow the repository's Conventional Commit rule.
+- Return `BLOCKED` if the diff introduces string literals that should use existing constants but did not.
 
 ## Reference
 

.agents/skills/pr-creation-checker/references/pr-readiness-checklist.md

@@ -12,3 +12,4 @@
 - Confirm non-applicable sections are filled with `N/A`.
 - Confirm the PR body does not include local absolute paths unless explicitly required.
 - Confirm multiline GitHub CLI commands use `--body-file`.
+- Confirm new hardcoded string literals were not introduced for values already represented by existing constants/enums (including protocol labels, error identifiers, headers, and metric names), or record a justified exception.

.cargo/config.toml

@@ -0,0 +1,26 @@
+# Copyright 2024 RustFS Team
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# RustFS Cargo configuration
+
+# Enable tokio_unstable cfg for dial9-tokio-telemetry support
+# This allows dial9 to hook into Tokio's internal runtime events
+[build]
+# Enable Tokio unstable features required by dial9-tokio-telemetry for runtime tracing.
+# See: https://docs.rs/tokio/latest/tokio/#unstable-features
+rustflags = ["--cfg", "tokio_unstable"]
+
+# Enable frame pointers for CPU profiling (Linux only, optional but recommended)
+# Uncomment the following line for better CPU profiling data
+# rustflags = ["--cfg", "tokio_unstable", "-C", "force-frame-pointers=yes"]

.docker/compose/README.md

@@ -36,6 +36,24 @@ To start a 4-node cluster for distributed testing:
 docker compose -f .docker/compose/docker-compose.cluster.yaml up -d
 ```
 
+### Script-Based 4-Node Validation (Recommended)
+
+Use the local validation script when you need local-source image build, failover checks,
+and benchmark workflow in one command:
+
+```bash
+# Default mode: WAIT_PROBE_MODE=service
+# This avoids false negatives where /health/ready remains 503 locally
+# while the service path is already available.
+./scripts/run_four_node_cluster_failover_bench.sh
+```
+
+Strict mode is available when you explicitly want `/health/ready == 200` as the gate:
+
+```bash
+WAIT_PROBE_MODE=ready ./scripts/run_four_node_cluster_failover_bench.sh
+```
+
 ### (Deprecated) Minimal Observability
 
 ```bash

.docker/compose/docker-compose.cluster.local-build.yml

@@ -0,0 +1,144 @@
+# Copyright 2024 RustFS Team
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+services:
+  node1:
+    image: ${RUSTFS_IMAGE:-rustfs/rustfs:local-4node}
+    build:
+      context: ../..
+      dockerfile: Dockerfile.source
+    hostname: node1
+    environment:
+      - RUSTFS_VOLUMES=http://node{1...4}:9000/data/rustfs{0...3}
+      - RUSTFS_ADDRESS=:9000
+      - RUSTFS_CONSOLE_ENABLE=true
+      - RUSTFS_ACCESS_KEY=${RUSTFS_ACCESS_KEY:-rustfsadmin}
+      - RUSTFS_SECRET_KEY=${RUSTFS_SECRET_KEY:-rustfsadmin}
+      - RUSTFS_OBS_ENDPOINT=${RUSTFS_OBS_ENDPOINT:-http://host.docker.internal:4318}
+      - RUSTFS_OBS_LOGGER_LEVEL=${RUSTFS_OBS_LOGGER_LEVEL:-info}
+      - RUSTFS_UNSAFE_BYPASS_DISK_CHECK=${RUSTFS_UNSAFE_BYPASS_DISK_CHECK:-true}
+    extra_hosts:
+      - "host.docker.internal:host-gateway"
+    volumes:
+      - node1_data_0:/data/rustfs0
+      - node1_data_1:/data/rustfs1
+      - node1_data_2:/data/rustfs2
+      - node1_data_3:/data/rustfs3
+    ports:
+      - "9000:9000"
+    networks:
+      - rustfs-cluster-net
+
+  node2:
+    image: ${RUSTFS_IMAGE:-rustfs/rustfs:local-4node}
+    build:
+      context: ../..
+      dockerfile: Dockerfile.source
+    hostname: node2
+    environment:
+      - RUSTFS_VOLUMES=http://node{1...4}:9000/data/rustfs{0...3}
+      - RUSTFS_ADDRESS=:9000
+      - RUSTFS_CONSOLE_ENABLE=true
+      - RUSTFS_ACCESS_KEY=${RUSTFS_ACCESS_KEY:-rustfsadmin}
+      - RUSTFS_SECRET_KEY=${RUSTFS_SECRET_KEY:-rustfsadmin}
+      - RUSTFS_OBS_ENDPOINT=${RUSTFS_OBS_ENDPOINT:-http://host.docker.internal:4318}
+      - RUSTFS_OBS_LOGGER_LEVEL=${RUSTFS_OBS_LOGGER_LEVEL:-info}
+      - RUSTFS_UNSAFE_BYPASS_DISK_CHECK=${RUSTFS_UNSAFE_BYPASS_DISK_CHECK:-true}
+    extra_hosts:
+      - "host.docker.internal:host-gateway"
+    volumes:
+      - node2_data_0:/data/rustfs0
+      - node2_data_1:/data/rustfs1
+      - node2_data_2:/data/rustfs2
+      - node2_data_3:/data/rustfs3
+    ports:
+      - "9001:9000"
+    networks:
+      - rustfs-cluster-net
+
+  node3:
+    image: ${RUSTFS_IMAGE:-rustfs/rustfs:local-4node}
+    build:
+      context: ../..
+      dockerfile: Dockerfile.source
+    hostname: node3
+    environment:
+      - RUSTFS_VOLUMES=http://node{1...4}:9000/data/rustfs{0...3}
+      - RUSTFS_ADDRESS=:9000
+      - RUSTFS_CONSOLE_ENABLE=true
+      - RUSTFS_ACCESS_KEY=${RUSTFS_ACCESS_KEY:-rustfsadmin}
+      - RUSTFS_SECRET_KEY=${RUSTFS_SECRET_KEY:-rustfsadmin}
+      - RUSTFS_OBS_ENDPOINT=${RUSTFS_OBS_ENDPOINT:-http://host.docker.internal:4318}
+      - RUSTFS_OBS_LOGGER_LEVEL=${RUSTFS_OBS_LOGGER_LEVEL:-info}
+      - RUSTFS_UNSAFE_BYPASS_DISK_CHECK=${RUSTFS_UNSAFE_BYPASS_DISK_CHECK:-true}
+    extra_hosts:
+      - "host.docker.internal:host-gateway"
+    volumes:
+      - node3_data_0:/data/rustfs0
+      - node3_data_1:/data/rustfs1
+      - node3_data_2:/data/rustfs2
+      - node3_data_3:/data/rustfs3
+    ports:
+      - "9002:9000"
+    networks:
+      - rustfs-cluster-net
+
+  node4:
+    image: ${RUSTFS_IMAGE:-rustfs/rustfs:local-4node}
+    build:
+      context: ../..
+      dockerfile: Dockerfile.source
+    hostname: node4
+    environment:
+      - RUSTFS_VOLUMES=http://node{1...4}:9000/data/rustfs{0...3}
+      - RUSTFS_ADDRESS=:9000
+      - RUSTFS_CONSOLE_ENABLE=true
+      - RUSTFS_ACCESS_KEY=${RUSTFS_ACCESS_KEY:-rustfsadmin}
+      - RUSTFS_SECRET_KEY=${RUSTFS_SECRET_KEY:-rustfsadmin}
+      - RUSTFS_OBS_ENDPOINT=${RUSTFS_OBS_ENDPOINT:-http://host.docker.internal:4318}
+      - RUSTFS_OBS_LOGGER_LEVEL=${RUSTFS_OBS_LOGGER_LEVEL:-info}
+      - RUSTFS_UNSAFE_BYPASS_DISK_CHECK=${RUSTFS_UNSAFE_BYPASS_DISK_CHECK:-true}
+    extra_hosts:
+      - "host.docker.internal:host-gateway"
+    volumes:
+      - node4_data_0:/data/rustfs0
+      - node4_data_1:/data/rustfs1
+      - node4_data_2:/data/rustfs2
+      - node4_data_3:/data/rustfs3
+    ports:
+      - "9003:9000"
+    networks:
+      - rustfs-cluster-net
+
+volumes:
+  node1_data_0:
+  node1_data_1:
+  node1_data_2:
+  node1_data_3:
+  node2_data_0:
+  node2_data_1:
+  node2_data_2:
+  node2_data_3:
+  node3_data_0:
+  node3_data_1:
+  node3_data_2:
+  node3_data_3:
+  node4_data_0:
+  node4_data_1:
+  node4_data_2:
+  node4_data_3:
+
+networks:
+  rustfs-cluster-net:
+    driver: bridge

.docker/compose/docker-compose.observability.yaml

@@ -26,7 +26,7 @@ services:
     restart: "no"
 
   tempo:
-    image: grafana/tempo:latest
+    image: grafana/tempo:2.10.5
     user: "10001"
     command: [ "-config.file=/etc/tempo.yaml" ]
     volumes:
@@ -36,9 +36,19 @@ services:
       - "3200:3200" # tempo
       - "4317"      # otlp grpc
       - "4318"      # otlp http
+      - "7946"      # memberlist
     restart: unless-stopped
     networks:
       - rustfs-network
+    depends_on:
+      tempo-init:
+        condition: service_completed_successfully
+    healthcheck:
+      test: [ "CMD", "/tempo", "-version" ]
+      interval: 10s
+      timeout: 5s
+      retries: 3
+      start_period: 15s
 
   otel-collector:
     image: otel/opentelemetry-collector-contrib:latest
@@ -61,6 +71,12 @@ services:
       - jaeger
       - prometheus
       - loki
+    restart: unless-stopped
+    healthcheck:
+      test: [ "CMD", "/otelcol-contrib", "--version" ]
+      interval: 10s
+      timeout: 5s
+      retries: 3
 
   jaeger:
     image: jaegertracing/jaeger:latest
@@ -72,12 +88,23 @@ services:
       - BADGER_DIRECTORY_KEY=/badger/key
       - COLLECTOR_OTLP_ENABLED=true
     volumes:
+      - ../../.docker/observability/jaeger.yaml:/etc/jaeger/config.yml:ro
       - jaeger-data:/badger
     ports:
       - "16686:16686" # Web UI
       - "14269:14269" # Admin/Metrics
+      - "4317"        # otlp grpc
+      - "4318"        # otlp http
+    command: [ "--config", "/etc/jaeger/config.yml" ]
     networks:
       - rustfs-network
+    restart: unless-stopped
+    healthcheck:
+      test: [ "CMD", "wget", "--spider", "-q", "http://localhost:14269" ]
+      interval: 10s
+      timeout: 5s
+      retries: 3
+      start_period: 15s
 
   prometheus:
     image: prom/prometheus:latest
@@ -85,6 +112,7 @@ services:
       - TZ=Asia/Shanghai
     volumes:
       - ../../.docker/observability/prometheus.yml:/etc/prometheus/prometheus.yml:ro
+      - ../../.docker/observability/prometheus-rules:/etc/prometheus/rules:ro
       - prometheus-data:/prometheus
     ports:
       - "9090:9090"
@@ -94,23 +122,45 @@ services:
       - '--web.enable-remote-write-receiver'
       - '--enable-feature=promql-experimental-functions'
       - '--storage.tsdb.path=/prometheus'
-      - '--web.console.libraries=/usr/share/prometheus/console_libraries'
-      - '--web.console.templates=/usr/share/prometheus/consoles'
+      - '--storage.tsdb.retention.time=30d'
     networks:
       - rustfs-network
+    restart: unless-stopped
+    healthcheck:
+      test: [ "CMD", "wget", "--spider", "-q", "http://localhost:9090/-/healthy" ]
+      interval: 10s
+      timeout: 5s
+      retries: 3
 
   loki:
     image: grafana/loki:latest
     environment:
       - TZ=Asia/Shanghai
     volumes:
-      - ../../.docker/observability/loki.yaml:/etc/loki/local-config.yaml:ro
+      - ../../.docker/observability/loki.yaml:/etc/loki/loki.yaml:ro
       - loki-data:/loki
     ports:
       - "3100:3100"
-    command: -config.file=/etc/loki/local-config.yaml
+    command: -config.file=/etc/loki/loki.yaml
     networks:
       - rustfs-network
+    restart: unless-stopped
+    healthcheck:
+      test: [ "CMD", "wget", "--spider", "-q", "http://localhost:3100/ready" ]
+      interval: 15s
+      timeout: 10s
+      retries: 5
+      start_period: 60s
+
+  pyroscope:
+    image: grafana/pyroscope:latest
+    ports:
+      - "4040:4040"
+    command:
+      - -self-profiling.disable-push=true
+    networks:
+      - rustfs-network
+    restart: unless-stopped
 
   grafana:
     image: grafana/grafana:latest
@@ -127,10 +177,17 @@ services:
     volumes:
       - ../../.docker/observability/grafana/provisioning:/etc/grafana/provisioning:ro
       - ../../.docker/observability/grafana/dashboards:/var/lib/grafana/dashboards:ro
+      - grafana-data:/var/lib/grafana
     depends_on:
       - prometheus
       - tempo
       - loki
+    restart: unless-stopped
+    healthcheck:
+      test: [ "CMD", "wget", "--spider", "-q", "http://localhost:3000/api/health" ]
+      interval: 10s
+      timeout: 5s
+      retries: 3
 
   # --- RustFS Cluster ---
 
@@ -215,6 +272,7 @@ volumes:
   tempo-data:
   loki-data:
   jaeger-data:
+  grafana-data:
 
 networks:
   rustfs-network:

.docker/observability/README.md

@@ -13,6 +13,9 @@ The stack is composed of the following best-in-class open-source components:
 - **Jaeger** (v1.59.0): Distributed tracing system (configured as a secondary UI/storage).
 - **OpenTelemetry Collector** (v0.104.0): A vendor-agnostic implementation for receiving, processing, and exporting telemetry data.
 
+By default, this stack uses Tempo in single-binary mode and does not require Kafka/Redpanda.
+If you want the Kafka-backed HA Tempo path, use `docker-compose-example-for-rustfs.yml` together with `docker-compose-tempo-ha-override.yml`.
+
 ## Architecture
 
 1. **Telemetry Collection**: Applications send OTLP (OpenTelemetry Protocol) data (Metrics, Logs, Traces) to the **OpenTelemetry Collector**.
@@ -46,6 +49,15 @@ Run the following command to start the entire stack:
 docker compose up -d
 ```
 
+### High Availability Tempo
+
+The default `docker-compose.yml` is the single-node stack.
+If you need the Kafka-backed HA Tempo configuration, start it with:
+
+```bash
+docker compose -f docker-compose-example-for-rustfs.yml -f docker-compose-tempo-ha-override.yml up -d
+```
+
 ### Access Dashboards
 
 | Service        | URL                                              | Credentials       | Description                    |

.docker/observability/README_ZH.md

@@ -13,6 +13,9 @@
 - **Jaeger** (v1.59.0): 分布式追踪系统（配置为辅助 UI/存储）。
 - **OpenTelemetry Collector** (v0.104.0): 接收、处理和导出遥测数据的供应商无关实现。
 
+默认情况下，这套技术栈使用 Tempo 单二进制模式，不依赖 Kafka/Redpanda。
+如果需要基于 Kafka 的 HA Tempo 路径，请使用 `docker-compose-example-for-rustfs.yml` 配合 `docker-compose-tempo-ha-override.yml`。
+
 ## 架构
 
 1.  **遥测收集**: 应用程序将 OTLP (OpenTelemetry Protocol) 数据（指标、日志、追踪）发送到 **OpenTelemetry Collector**。
@@ -46,6 +49,15 @@
 docker compose up -d
 ```
 
+### Tempo 高可用模式
+
+默认的 `docker-compose.yml` 对应单机栈。
+如果需要基于 Kafka 的 HA Tempo 配置，请使用：
+
+```bash
+docker compose -f docker-compose-example-for-rustfs.yml -f docker-compose-tempo-ha-override.yml up -d
+```
+
 ### 访问仪表盘
 
 | 服务 | URL | 凭据 | 描述 |

.docker/observability/docker-compose-example-for-rustfs.yml

@@ -85,10 +85,8 @@ services:
     networks:
       - otel-network
     restart: unless-stopped
-    depends_on:
-      - redpanda
     healthcheck:
-      test: [ "CMD", "wget", "--spider", "-q", "http://localhost:3200/ready" ]
+      test: [ "CMD", "/tempo", "-version" ]
       interval: 10s
       timeout: 5s
       retries: 3
@@ -205,7 +203,7 @@ services:
       - prometheus
       - loki
     healthcheck:
-      test: [ "CMD", "wget", "--spider", "-q", "http://localhost:13133" ]
+      test: [ "CMD", "/otelcol-contrib", "--version" ]
       interval: 10s
       timeout: 5s
       retries: 3

.docker/observability/docker-compose.yml

@@ -15,9 +15,8 @@
 services:
 
   # --- Tracing ---
-
   tempo:
-    image: grafana/tempo:latest
+    image: grafana/tempo:2.10.5
     container_name: tempo
     command: [ "-config.file=/etc/tempo.yaml" ]
     volumes:
@@ -32,12 +31,24 @@ services:
       - otel-network
     restart: unless-stopped
     healthcheck:
-      test: [ "CMD", "wget", "--spider", "-q", "http://localhost:3200/ready" ]
+      test: [ "CMD", "/tempo", "-version" ]
       interval: 10s
       timeout: 5s
       retries: 3
       start_period: 15s
 
+  vulture:
+    image: grafana/tempo-vulture:latest
+    restart: always
+    command:
+      [
+        "-prometheus-listen-address=:8080",
+        "-tempo-query-url=http://tempo:3200",
+        "-tempo-push-url=http://tempo:4317",
+      ]
+    depends_on:
+      - tempo
+
   jaeger:
     image: jaegertracing/jaeger:latest
     container_name: jaeger
@@ -73,6 +84,7 @@ services:
     container_name: prometheus
     volumes:
       - ./prometheus.yml:/etc/prometheus/prometheus.yml:ro
+      - ./prometheus-rules:/etc/prometheus/rules:ro
       - prometheus-data:/prometheus
     ports:
       - "9090:9090"
@@ -135,7 +147,7 @@ services:
       - prometheus
       - loki
     healthcheck:
-      test: [ "CMD", "wget", "--spider", "-q", "http://localhost:13133" ]
+      test: [ "CMD", "/otelcol-contrib", "--version" ]
       interval: 10s
       timeout: 5s
       retries: 3

.docker/observability/grafana/provisioning/datasources.yaml

@@ -1,97 +0,0 @@
-# Copyright 2024 RustFS Team
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-apiVersion: 1
-
-datasources:
-  - name: Prometheus
-    type: prometheus
-    uid: prometheus
-    access: proxy
-    orgId: 1
-    url: http://prometheus:9090
-    isDefault: true
-    version: 1
-    editable: false
-    jsonData:
-      httpMethod: GET
-      exemplarTraceIdDestinations:
-        - name: trace_id
-          datasourceUid: tempo
-
-  - name: Tempo
-    type: tempo
-    uid: tempo
-    access: proxy
-    orgId: 1
-    url: http://tempo:3200
-    isDefault: false
-    version: 1
-    editable: false
-    jsonData:
-      httpMethod: GET
-      serviceMap:
-        datasourceUid: prometheus
-      tracesToLogs:
-        datasourceUid: loki
-        tags: [ 'job', 'instance', 'pod', 'namespace', 'service.name' ]
-        mappedTags: [ { key: 'service.name', value: 'app' } ]
-        spanStartTimeShift: '1s'
-        spanEndTimeShift: '-1s'
-        filterByTraceID: true
-        filterBySpanID: false
-      tracesToMetrics:
-        datasourceUid: prometheus
-        tags: [ { key: 'service.name' }, { key: 'job' } ]
-        queries:
-          - name: 'Service-Level Latency'
-            query: 'sum(rate(traces_spanmetrics_latency_bucket{$$__tags}[5m])) by (le)'
-          - name: 'Service-Level Calls'
-            query: 'sum(rate(traces_spanmetrics_calls_total{$$__tags}[5m]))'
-          - name: 'Service-Level Errors'
-            query: 'sum(rate(traces_spanmetrics_calls_total{status_code="ERROR", $$__tags}[5m]))'
-      nodeGraph:
-        enabled: true
-
-  - name: Loki
-    type: loki
-    uid: loki
-    orgId: 1
-    url: http://loki:3100
-    isDefault: false
-    version: 1
-    editable: false
-    jsonData:
-      derivedFields:
-        - datasourceUid: tempo
-          matcherRegex: 'trace_id=(\w+)'
-          name: 'TraceID'
-          url: '$${__value.raw}'
-
-  - name: Jaeger
-    type: jaeger
-    uid: jaeger
-    url: http://jaeger:16686
-    access: proxy
-    isDefault: false
-    editable: false
-    jsonData:
-      tracesToLogs:
-        datasourceUid: loki
-        tags: [ 'job', 'instance', 'pod', 'namespace', 'service.name' ]
-        mappedTags: [ { key: 'service.name', value: 'app' } ]
-        spanStartTimeShift: '1s'
-        spanEndTimeShift: '-1s'
-        filterByTraceID: true
-        filterBySpanID: false

.docker/observability/prometheus-rules/rustfs-dashboard.yml

@@ -0,0 +1,40 @@
+groups:
+  - name: rustfs-dashboard
+    interval: 30s
+    rules:
+      - record: rustfs:http_server_requests:rate5m
+        expr: sum by (job) (rate(rustfs_http_server_requests_total[5m]))
+
+      - record: rustfs:http_server_request_duration_seconds:p50_5m
+        expr: histogram_quantile(0.50, sum by (le, job) (rate(rustfs_http_server_request_duration_seconds_bucket[5m])))
+      - record: rustfs:http_server_request_duration_seconds:p95_5m
+        expr: histogram_quantile(0.95, sum by (le, job) (rate(rustfs_http_server_request_duration_seconds_bucket[5m])))
+      - record: rustfs:http_server_request_duration_seconds:p99_5m
+        expr: histogram_quantile(0.99, sum by (le, job) (rate(rustfs_http_server_request_duration_seconds_bucket[5m])))
+
+      - record: rustfs:http_server_response_body_size_bytes:p50_5m
+        expr: histogram_quantile(0.50, sum by (le, job) (rate(rustfs_http_server_response_body_size_bytes_bucket[5m])))
+      - record: rustfs:http_server_response_body_size_bytes:p95_5m
+        expr: histogram_quantile(0.95, sum by (le, job) (rate(rustfs_http_server_response_body_size_bytes_bucket[5m])))
+      - record: rustfs:http_server_response_body_size_bytes:p99_5m
+        expr: histogram_quantile(0.99, sum by (le, job) (rate(rustfs_http_server_response_body_size_bytes_bucket[5m])))
+
+      - record: rustfs:log_cleaner_runs:rate15m
+        expr: sum by (job) (rate(rustfs_log_cleaner_runs_total[15m]))
+      - record: rustfs:log_cleaner_failure_ratio:rate5m
+        expr: sum by (job) (rate(rustfs_log_cleaner_run_failures_total[5m])) / clamp_min(sum by (job) (rate(rustfs_log_cleaner_runs_total[5m])), 1e-9)
+      - record: rustfs:log_cleaner_rotation_failure_ratio:rate5m
+        expr: sum by (job) (rate(rustfs_log_cleaner_rotation_failures_total[5m])) / clamp_min(sum by (job) (rate(rustfs_log_cleaner_rotation_total[5m])), 1e-9)
+      - record: rustfs:log_cleaner_rotation_duration_seconds:p95_5m
+        expr: histogram_quantile(0.95, sum by (le, job) (rate(rustfs_log_cleaner_rotation_duration_seconds_bucket[5m])))
+      - record: rustfs:log_cleaner_compress_duration_seconds:p95_5m
+        expr: histogram_quantile(0.95, sum by (le, job) (rate(rustfs_log_cleaner_compress_duration_seconds_bucket[5m])))
+
+      - record: rustfs:scanner_objects_scanned:rate5m
+        expr: sum by (job) (rate(rustfs_scanner_objects_scanned_total[5m]))
+      - record: rustfs:scanner_directories_scanned:rate5m
+        expr: sum by (job) (rate(rustfs_scanner_directories_scanned_total[5m]))
+      - record: rustfs:scanner_buckets_scanned:rate5m
+        expr: sum by (job) (rate(rustfs_scanner_buckets_scanned_total[5m]))
+      - record: rustfs:scanner_cycles_success:rate5m
+        expr: sum by (job) (rate(rustfs_scanner_cycles_total{result="success"}[5m]))

.docker/observability/prometheus.yml

@@ -19,6 +19,9 @@ global:
     cluster: 'rustfs-dev'  # Label to identify the cluster
     replica: '1'  # Replica identifier
 
+rule_files:
+  - /etc/prometheus/rules/*.yml
+
 scrape_configs:
   - job_name: 'otel-collector'
     static_configs:
@@ -50,6 +53,11 @@ scrape_configs:
     static_configs:
       - targets: [ 'localhost:9090' ]
 
+  - job_name: 'vulture'
+    static_configs:
+      - targets:
+          - 'vulture:8080'
+
 otlp:
   promote_resource_attributes:
     - service.instance.id

.docker/observability/redpanda-console.yaml

@@ -0,0 +1,3 @@
+kafka:
+  brokers:
+    - redpanda:9092

.docker/observability/tempo.yaml

@@ -12,7 +12,6 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-partition_ring_live_store: true
 stream_over_http_enabled: true
 
 server:
@@ -25,56 +24,25 @@ memberlist:
   join_members:
     - tempo:7946
 
-# Distributor configuration - receives traces and writes directly to ingesters
 distributor:
-  ingester_write_path_enabled: true
-  kafka_write_path_enabled: false
   receivers:
     otlp:
       protocols:
         grpc:
-          endpoint: "tempo:4317"
+          endpoint: "0.0.0.0:4317"
         http:
-          endpoint: "tempo:4318"
-  ring:
-    kvstore:
-      store: memberlist
+          endpoint: "0.0.0.0:4318"
 
-# Ingester configuration - consumes from Kafka and stores traces
 ingester:
-  lifecycler:
-    ring:
-      kvstore:
-        store: memberlist
-      replication_factor: 1
-    tokens_file_path: /var/tempo/tokens.json
-  trace_idle_period: 10s
-  max_block_bytes: 1_000_000
   max_block_duration: 5m
 
-backend_scheduler:
-  provider:
-    compaction:
-      compaction:
-        block_retention: 1h
-
-backend_worker:
-  backend_scheduler_addr: localhost:3200
-  compaction:
-    block_retention: 1h
-  ring:
-    kvstore:
-      store: memberlist
-
-querier:
-  frontend_worker:
-    frontend_address: tempo:3200
-
 metrics_generator:
   registry:
     external_labels:
       source: tempo
       cluster: docker-compose
+  traces_storage:
+    path: /var/tempo/generator/traces
   storage:
     path: /var/tempo/generator/wal
     remote_write:
@@ -90,9 +58,9 @@ storage:
   trace:
     backend: local
     wal:
-      path: /var/tempo/wal
+      path: /var/tempo/wal # where to store the wal locally
     local:
-      path: /var/tempo/blocks
+      path: /var/tempo/blocks # where to store the traces locally
 
 overrides:
   defaults:
@@ -100,25 +68,5 @@ overrides:
       processors: [ "span-metrics", "service-graphs", "local-blocks" ]
       generate_native_histograms: both
 
-ingest:
-  enabled: false
-  # Disabled because using direct ingester write path
-  # If you want Kafka path, enable this and set:
-  # kafka:
-  #   brokers: [redpanda:9092]
-  #   topic: tempo-ingest
-  #   encoding: protobuf
-  #   consumer_group: tempo-ingest-consumer
-
-block_builder:
-  consume_cycle_duration: 30s
-
-compactor:
-  compaction:
-    block_retention: 168h  # 7 days
-  ring:
-    kvstore:
-      store: memberlist
-
 usage_report:
   reporting_enabled: false

.github/AGENTS.md

@@ -25,6 +25,12 @@ Current `test-and-lint` gate includes:
 
 - `cargo nextest run --all --exclude e2e_test`
 - `cargo test --all --doc`
+- `cargo test -p rustfs get_object_chunk_fast_path`
+- `cargo test -p rustfs materialize_chunk_stream_before_commit`
+- `touch rustfs/build.rs`
+- `cargo build -p rustfs --bins --jobs 2`
+- `cargo test -p e2e_test archive_multipart_roundtrip_preserves_bytes`
+- `cargo test -p e2e_test presigned_get_and_reverse_proxy_preserve_multipart_bytes_with_fast_path`
 - `cargo fmt --all --check`
 - `cargo clippy --all-targets --all-features -- -D warnings`
 - `./scripts/check_layer_dependencies.sh`

.github/actions/setup/action.yml

@@ -62,7 +62,7 @@ runs:
         repo-token: ${{ inputs.github-token }}
 
     - name: Install flatc
-      uses: Nugine/setup-flatc@v1
+      uses: Nugine/setup-flatc@v1.2.4
       with:
         version: "25.9.23"
 

.github/cla.yml

@@ -0,0 +1,17 @@
+enabled: true
+
+document:
+  version: v1
+  url: https://github.com/rustfs/cla/blob/main/cla/v1.md
+
+signing:
+  mode: comment
+  comment_pattern: I have read and agree to the CLA.
+
+registry:
+  type: json-repo
+  repository: rustfs/cla
+  path_prefix: signatures
+
+status:
+  check_name: CLA Check

.github/dependabot.yml

@@ -26,11 +26,20 @@ updates:
       day: "monday"
       timezone: "Asia/Shanghai"
       time: "08:00"
+    assignees:
+      - "houseme"
+    reviewers:
+      - "overtrue"
+      - "majinghe"
     ignore:
       - dependency-name: "object_store"
         versions: [ "0.13.x" ]
       - dependency-name: "libunftp"
         versions: [ "0.23.x" ]
+      - dependency-name: "ratelimit"
+        versions: [ "1.x" ]
+      - dependency-name: "ratelimit"
+        versions: [ "2.x" ]
     groups:
       s3s:
         update-types:

.github/pull_request_template.md

@@ -34,4 +34,4 @@ Pull Request Template for RustFS
 
 ---
 
-Thank you for your contribution! Please ensure your PR follows the community standards ([CODE_OF_CONDUCT.md](CODE_OF_CONDUCT.md)) and sign the CLA if this is your first contribution.
+Thank you for your contribution! Please ensure your PR follows the community standards ([CODE_OF_CONDUCT.md](CODE_OF_CONDUCT.md)). If this is your first contribution, review the [CLA document](https://github.com/rustfs/cla/blob/main/cla/v1.md) and sign it by commenting `I have read and agree to the CLA.` on the PR.

.github/workflows/build.yml

@@ -204,7 +204,12 @@ jobs:
     runs-on: ${{ matrix.os }}
     timeout-minutes: 60
     env:
-      RUSTFLAGS: ${{ matrix.rustflags }}
+      FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
+      # Always enable Tokio unstable features (required by dial9-tokio-telemetry).
+      # The RUSTFLAGS env var takes precedence over .cargo/config.toml [build] rustflags,
+      # so we must include --cfg tokio_unstable here explicitly; otherwise an empty
+      # RUSTFLAGS value would shadow the config-file flag and silently break tracing.
+      RUSTFLAGS: "--cfg tokio_unstable ${{ matrix.rustflags }}"
     strategy:
       fail-fast: false
       matrix: ${{ fromJson(needs.prepare-platform-matrix.outputs.matrix) }}
@@ -316,19 +321,26 @@ jobs:
               ;;
           esac
 
+          # Normalize version used for package filenames
+          PACKAGE_VERSION="${VERSION}"
+          if [[ "$PACKAGE_VERSION" == v* ]]; then
+            PACKAGE_VERSION="${PACKAGE_VERSION#v}"
+          fi
+
           # Generate package name based on build type
           if [[ -n "$VARIANT" ]]; then
             ARCH_WITH_VARIANT="${ARCH}-${VARIANT}"
           else
             ARCH_WITH_VARIANT="${ARCH}"
           fi
+          PACKAGE_BASENAME="rustfs-${PLATFORM}-${ARCH_WITH_VARIANT}"
 
           if [[ "$BUILD_TYPE" == "development" ]]; then
             # Development build: rustfs-${platform}-${arch}-${variant}-dev-${short_sha}.zip
             PACKAGE_NAME="rustfs-${PLATFORM}-${ARCH_WITH_VARIANT}-dev-${SHORT_SHA}"
           else
             # Release/Prerelease build: rustfs-${platform}-${arch}-${variant}-v${version}.zip
-            PACKAGE_NAME="rustfs-${PLATFORM}-${ARCH_WITH_VARIANT}-v${VERSION}"
+            PACKAGE_NAME="${PACKAGE_BASENAME}-v${PACKAGE_VERSION}"
           fi
 
           # Create zip packages for all platforms
@@ -398,7 +410,7 @@ jobs:
           if [[ "$BUILD_TYPE" == "release" ]] || [[ "$BUILD_TYPE" == "prerelease" ]]; then
             # Create latest version filename
             # Convert from rustfs-linux-x86_64-musl-v1.0.0 to rustfs-linux-x86_64-musl-latest
-            LATEST_FILE="${PACKAGE_NAME%-v*}-latest.zip"
+            LATEST_FILE="${PACKAGE_BASENAME}-latest.zip"
 
             echo "🔄 Creating latest version: ${PACKAGE_NAME}.zip -> $LATEST_FILE"
             cp "${PACKAGE_NAME}.zip" "$LATEST_FILE"
@@ -411,7 +423,7 @@ jobs:
             # Development builds (only main branch triggers development builds)
             # Create main-latest version filename
             # Convert from rustfs-linux-x86_64-dev-abc123 to rustfs-linux-x86_64-main-latest
-            MAIN_LATEST_FILE="${PACKAGE_NAME%-dev-*}-main-latest.zip"
+            MAIN_LATEST_FILE="${PACKAGE_BASENAME}-main-latest.zip"
 
             echo "🔄 Creating main-latest version: ${PACKAGE_NAME}.zip -> $MAIN_LATEST_FILE"
             cp "${PACKAGE_NAME}.zip" "$MAIN_LATEST_FILE"

.github/workflows/ci.yml

@@ -53,6 +53,8 @@ on:
       - ".github/workflows/docker.yml"
       - ".github/workflows/audit.yml"
       - ".github/workflows/performance.yml"
+  merge_group:
+    types: [checks_requested]
   schedule:
     - cron: "0 0 * * 0" # Weekly on Sunday at midnight UTC
   workflow_dispatch:
@@ -106,6 +108,8 @@ jobs:
     if: needs.skip-check.outputs.should_skip != 'true'
     runs-on: ubicloud-standard-4
     timeout-minutes: 60
+    env:
+      FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
     steps:
       - name: Checkout repository
         uses: actions/checkout@v6
@@ -138,6 +142,8 @@ jobs:
     if: needs.skip-check.outputs.should_skip != 'true'
     runs-on: ubicloud-standard-4
     timeout-minutes: 30
+    env:
+      FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
     steps:
       - name: Checkout repository
         uses: actions/checkout@v6

.github/workflows/cla.yml

@@ -0,0 +1,70 @@
+# Copyright 2026 RustFS Team
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+name: CLA Check
+
+on:
+  pull_request_target:
+    types: [opened, synchronize, reopened]
+  merge_group:
+    types: [checks_requested]
+  issue_comment:
+    types: [created, edited]
+
+permissions:
+  contents: write
+  pull-requests: write
+  issues: write
+  checks: write
+
+jobs:
+  cla:
+    if: ${{ github.event_name != 'issue_comment' || github.event.issue.pull_request }}
+    runs-on: ubuntu-latest
+    steps:
+      - name: Report CLA result for merge queue
+        if: github.event_name == 'merge_group'
+        uses: actions/github-script@v8
+        with:
+          script: |
+            await github.rest.checks.create({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              name: 'CLA Check',
+              head_sha: context.sha,
+              status: 'completed',
+              conclusion: 'success',
+              output: {
+                title: 'CLA requirements satisfied for merge queue',
+                summary: 'Queued pull requests must satisfy the required CLA check before they enter the merge queue. This reports the existing CLA result on the merge-group SHA.'
+              }
+            });
+
+      - name: Create token for rustfs/cla
+        if: github.event_name != 'merge_group'
+        id: registry-token
+        uses: actions/create-github-app-token@v3
+        with:
+          app-id: ${{ vars.CLA_BOT_APP_ID }}
+          private-key: ${{ secrets.CLA_BOT_APP_PRIVATE_KEY }}
+          owner: ${{ github.repository_owner }}
+          repositories: cla
+          permission-contents: write
+
+      - name: Run CLA Bot
+        if: github.event_name != 'merge_group'
+        uses: overtrue/cla-bot@v0.0.9
+        with:
+          github-token: ${{ github.token }}
+          registry-token: ${{ steps.registry-token.outputs.token }}

.github/workflows/nix-flake-update.yml

@@ -35,7 +35,7 @@ jobs:
     steps:
       - name: Checkout repository
         uses: actions/checkout@v6
-        
+
       - name: Install Nix
         uses: DeterminateSystems/determinate-nix-action@v3
 
@@ -46,10 +46,10 @@ jobs:
         id: update
         uses: DeterminateSystems/update-flake-lock@main
         with:
-          git-author-name: heihutu
-          git-author-email: heihutu@gmail.com
-          git-committer-name: heihutu
-          git-committer-email: heihutu@gmail.com
+          git-author-name: houseme
+          git-author-email: housemecn@gmail.com
+          git-committer-name: houseme
+          git-committer-email: housemecn@gmail.com
           pr-title: "chore(deps): update flake.lock"
           pr-labels: |
             dependencies

.github/workflows/performance.yml

@@ -42,6 +42,8 @@ jobs:
     name: Performance Profiling
     runs-on: ubicloud-standard-2
     timeout-minutes: 30
+    env:
+      FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
     steps:
       - name: Checkout repository
         uses: actions/checkout@v6
@@ -117,6 +119,8 @@ jobs:
     name: Benchmark Tests
     runs-on: ubicloud-standard-2
     timeout-minutes: 45
+    env:
+      FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
     steps:
       - name: Checkout repository
         uses: actions/checkout@v6

.gitignore

@@ -20,7 +20,8 @@ deploy/data/*
 *jsonl
 .env
 .rustfs.sys
-.cargo
+.cargo/
+!.cargo/config.toml
 profile.json
 .docker/openobserve-otel/data
 *.zst
@@ -39,7 +40,6 @@ artifacts/
 *.audit
 *.snappy
 PR_DESCRIPTION.md
-IMPLEMENTATION_PLAN.md
 scripts/s3-tests/selected_tests.txt
 docs
 
@@ -48,4 +48,8 @@ result*
 *.gz
 rustfs-webdav.code-workspace
 
-.aiexclude
+.aiexclude
+*.bak
+# Local test/benchmark artifacts
+benchmarks.logs
+tmp/

.vscode/launch.json

@@ -197,19 +197,27 @@
                 // "__RUSTFS_SSE_SIMPLE_CMK": "2dfNXGHlsEflGVCxb+5DIdGEl1sIvtwX+QfmYasi5QM=",
 
                 // 2. kms local backend test key
-                "RUSTFS_KMS_ENABLE": "true",
-                "RUSTFS_KMS_BACKEND": "local",
-                "RUSTFS_KMS_KEY_DIR": "./target/kms-key-dir",
-                "RUSTFS_KMS_LOCAL_MASTER_KEY": "my-secret-key", // Some Password
-                "RUSTFS_KMS_DEFAULT_KEY_ID": "rustfs-master-key",
+                // "RUSTFS_KMS_ENABLE": "true",
+                // "RUSTFS_KMS_BACKEND": "local",
+                // "RUSTFS_KMS_KEY_DIR": "./target/kms-key-dir",
+                // "RUSTFS_KMS_LOCAL_MASTER_KEY": "my-secret-key", // Some Password
+                // "RUSTFS_KMS_DEFAULT_KEY_ID": "rustfs-master-key",
 
                 // 3. kms vault backend test key
                 // "RUSTFS_KMS_ENABLE": "true",
-                // "RUSTFS_KMS_BACKEND": "vault",
+                // "RUSTFS_KMS_BACKEND": "vault-kv2",
                 // "RUSTFS_KMS_VAULT_ADDRESS": "http://127.0.0.1:8200",
                 // "RUSTFS_KMS_VAULT_TOKEN": "Dev Token",
                 // "RUSTFS_KMS_DEFAULT_KEY_ID": "rustfs-master-key",
 
+                // 4. kms vault transit backend test key
+                "RUSTFS_KMS_ENABLE": "true",
+                "RUSTFS_KMS_BACKEND": "vault-transit",
+                "RUSTFS_KMS_VAULT_ADDRESS": "http://127.0.0.1:8200",
+                "RUSTFS_KMS_VAULT_TOKEN": "Dev Token",
+                "RUSTFS_KMS_VAULT_MOUNT_PATH": "transit",
+                "RUSTFS_KMS_DEFAULT_KEY_ID": "rustfs-master-key",
+
             },
             "sourceLanguages": [
                 "rust"

AGENTS.md

@@ -6,15 +6,46 @@ Use the nearest subdirectory `AGENTS.md` for path-specific guidance.
 ## Rule Precedence
 
 1. System/developer instructions.
-2. This file (global defaults).
-3. The nearest `AGENTS.md` in the current path (more specific scope wins).
+2. Current user/task instructions.
+3. The nearest `AGENTS.md` in the current path.
+4. This file (global defaults).
 
 If repo-level instructions conflict, follow the nearest file and keep behavior aligned with CI.
 
+## Execution Discipline
+
+- Read the relevant existing code, tests, and local guidance before changing behavior.
+- State assumptions when they affect the implementation or verification path.
+- If a task has multiple plausible interpretations, list the options briefly and choose the narrowest reasonable path; ask when the ambiguity would make the change risky.
+- For multi-step work, keep the plan minimal and tied to verifiable outcomes.
+- Avoid redundant file reads, repeated commands, and unnecessary exploratory work once enough context is available.
+- A good result is a minimal diff with clear assumptions, no over-engineering, and independent verification.
+
 ## Communication and Language
 
 - Respond in the same language used by the requester.
 - Keep source code, comments, commit messages, and PR title/body in English.
+- Be concise. Avoid sycophantic openers, closing fluff, and verbose status reporting.
+
+## Change Style for Existing Logic
+
+- Prefer direct, local code over extracting one-off helpers.
+- Extract a helper only when logic is reused or the extraction materially clarifies a non-trivial flow.
+- Solve only the requested problem; do not add speculative features, configurability, or adjacent improvements.
+- Prefer editing existing code over rewriting files or reshaping unrelated logic.
+- Modify only what is required and remove only artifacts introduced by your own changes.
+- Preserve the existing control-flow and logic shape when fixing bugs or addressing review comments, especially in init, distributed coordination, locking, metadata, and concurrency paths.
+- Do not refactor existing code only to make it easier to unit test.
+- Keep fixes narrowly aligned with the requested behavior; avoid semantic-adjacent rewrites while touching sensitive paths.
+- Keep code elegant, concise, and direct. Prefer minimal, readable implementations over over-engineering and excessive abstraction. Use comments to clarify non-obvious intent and invariants, not to compensate for unclear code.
+- Mention unrelated issues when useful, but do not fix them as part of a narrow task.
+
+## Constant and String Usage
+
+- Before introducing new string literals, search for existing constants/enums that already represent the same semantic value.
+- Reuse existing constants for protocol labels, error identifiers, header keys, event names, metric names, command tags, and similar fixed tokens.
+- If a new string is truly unique, define a local constant near related logic and avoid scattering the literal across multiple sites.
+- When changing existing behavior, keep naming and format consistency by aligning with established project constants.
 
 ## Sources of Truth
 
@@ -26,16 +57,25 @@ If repo-level instructions conflict, follow the nearest file and keep behavior a
 Avoid duplicating long crate lists or command matrices in instruction files.
 Reference the source files above instead.
 
-## Mandatory Before Commit
+## Verification Before PR
 
-Run and pass:
+Convert changes into independently verifiable outcomes. Prefer focused tests for behavior changes and run the relevant checks before declaring completion.
+
+For code changes, run and pass the following before opening a PR:
 
 ```bash
 make pre-commit
 ```
 
+Before pushing code changes, make sure formatting is clean:
+
+- Run `cargo fmt --all`.
+- Run `cargo fmt --all --check` and ensure no files are modified unexpectedly.
+
 If `make` is unavailable, run the equivalent checks defined under `.config/make/`.
-Do not commit when required checks fail.
+Documentation-only or instruction-only changes are exempt from the verification commands above (including the `.config/make/` equivalents), though any installed git pre-commit hooks (for example, from `make setup-hooks`) may still run on commit unless explicitly skipped.
+After build-based verification completes, clean generated build artifacts before wrapping up to avoid unnecessary disk usage.
+Do not open a PR with code changes when the required checks fail.
 
 ## Git and PR Baseline
 
@@ -46,6 +86,12 @@ Do not commit when required checks fail.
 - Use `N/A` for non-applicable template sections.
 - Include verification commands in the PR description.
 - When using `gh pr create`/`gh pr edit`, use `--body-file` instead of inline `--body` for multiline markdown.
+- Do not include the literal sequence `\n` in any GitHub issue, pull request, or discussion comment.
+- After fixing code review comments or CI findings, always mark corresponding review
+  comments/threads as resolved before returning to the user.
+- In handling review comments, confirm the underlying issue before changing code.
+  If a suggested change is not appropriate for behavior or risk, reply with a
+  concise rationale instead of blindly applying it.
 
 ## Security Baseline
 

ARCHITECTURE.md

@@ -0,0 +1,376 @@
+# ARCHITECTURE.md
+
+> Last updated: 2026-04-13 · Revision: 1 (draft)
+>
+> This document describes the high-level architecture of RustFS.
+> If you want to familiarize yourself with the code base, you are in the right place!
+>
+> See also [CONTRIBUTING.md](CONTRIBUTING.md) for development workflow.
+
+## Bird's Eye View
+
+RustFS is a high-performance, S3-compatible distributed object storage system written
+in Rust. It uses erasure coding for data durability, supports multi-tenancy through
+IAM/STS, and provides a web-based admin console.
+
+A running RustFS node exposes:
+
+- **S3 API** (port 9000) — the primary data path for object CRUD
+- **Admin API** (port 9000, `/minio/` prefix) — cluster management, IAM, metrics
+- **Console** (port 9001) — web UI backed by the Admin API
+- **Inter-node RPC** (gRPC/tonic) — cluster communication for distributed mode
+
+The core data flow for a PUT request looks like:
+
+```
+HTTP request
+  → server (TLS, auth, routing, compression)
+    → app/object_usecase (validation, policy, lifecycle)
+      → storage/ecfs (erasure coding, encryption, checksums)
+        → ecstore (disk pool selection, data distribution)
+          → rio (reader pipeline: encrypt → compress → hash → write)
+            → io-core (zero-copy I/O, buffer pool, direct I/O)
+              → local disk / remote disk via RPC
+```
+
+## Code Map
+
+The repository is a Cargo workspace with a flat `crates/` layout:
+
+```
+rustfs/                      # Workspace root (virtual manifest)
+├── rustfs/                  # Main binary + library crate (75K lines)
+│   └── src/
+│       ├── main.rs          # Entry point, startup sequence
+│       ├── lib.rs           # Module tree root
+│       ├── server/          # HTTP server, TLS, routing, middleware
+│       ├── admin/           # Admin API handlers and console
+│       ├── app/             # Use-case layer (object, bucket, multipart)
+│       ├── storage/         # Storage engine interface and implementation
+│       ├── auth.rs          # S3 request authentication
+│       ├── config/          # CLI args, config parsing, workload profiles
+│       └── ...
+├── crates/                  # 39 library crates
+│   ├── ecstore/             # Erasure-coded storage engine (⚠️ 87K lines)
+│   ├── rio/                 # Reader I/O pipeline (encrypt, compress, hash)
+│   ├── io-core/             # Zero-copy I/O, scheduling, buffer pool
+│   ├── io-metrics/          # I/O metrics collection
+│   ├── common/              # Shared runtime state, globals, data usage types
+│   ├── config/              # Configuration types and parsing
+│   ├── utils/               # Pure utility functions
+│   ├── ...                  # (see "Crate Reference" below)
+│   └── e2e_test/            # End-to-end integration tests
+└── docs/                    # Design documents and analysis
+```
+
+### Main Crate Layers (`rustfs/src/`)
+
+The main crate is organized in layers, top to bottom:
+
+| Layer | Directory | Responsibility |
+|-------|-----------|----------------|
+| **Server** | `server/` | HTTP listener, TLS, CORS, compression, middleware, graceful shutdown |
+| **Admin** | `admin/` | Admin API routing, 30+ handler modules, web console |
+| **App** | `app/` | Use-case orchestration: object_usecase, bucket_usecase, multipart_usecase |
+| **Storage** | `storage/` | S3 API translation, erasure-coded FS, SSE encryption, RPC, concurrency |
+| **Auth** | `auth.rs` | S3 signature verification, credential validation |
+| **Config** | `config/` | CLI parsing, config struct, workload profiles |
+
+A request flows **downward** through the layers. No layer should reach upward
+(e.g., storage must not import from admin).
+
+### Crate Reference
+
+Crates are organized in a dependency DAG with 9 depth levels (0 = leaf, 8 = top):
+
+```
+Depth 0 — LEAF (no internal deps):
+  appauth, checksums, config, credentials, crypto, io-metrics,
+  madmin, s3-common, workers, zip
+
+Depth 1:
+  io-core (→ io-metrics)
+  policy (→ config, credentials, crypto)
+  utils (→ config)                        ⚠️ inverted: utils should be leaf
+
+Depth 2:
+  concurrency, filemeta, keystone, kms, lock, obs,
+  signer, targets, trusted-proxies
+
+Depth 3:
+  common (→ filemeta, madmin)             ⚠️ inverted: common should be leaf
+
+Depth 4:
+  object-capacity, protos, rio
+
+Depth 5 — CORE:
+  ecstore (16 internal deps, 11 dependents — the architectural heart)
+
+Depth 6:
+  audit, heal, iam, metrics, notify, s3select-api, scanner
+
+Depth 7:
+  object-io, protocols, s3select-query
+
+Depth 8 — TOP:
+  rustfs (35 internal deps — the binary, depends on almost everything)
+```
+
+#### By Domain
+
+**Core Infrastructure:**
+
+| Crate | Lines | Purpose |
+|-------|-------|---------|
+| `config` | 3.3K | Configuration types and environment parsing |
+| `utils` | 8.7K | Pure utilities (paths, compression, network, retry) |
+| `common` | 4.4K | Shared runtime state, globals, data usage types, metrics |
+| `madmin` | 5.5K | Admin API request/response types |
+
+**I/O Pipeline:**
+
+| Crate | Lines | Purpose |
+|-------|-------|---------|
+| `io-core` | 6.5K | Zero-copy I/O, buffer pool, direct I/O, scheduling, backpressure |
+| `io-metrics` | 4.5K | I/O operation metrics and counters |
+| `rio` | 6.9K | Composable reader chain (encrypt → compress → hash → limit) |
+| `object-io` | 2.4K | High-level object read/write using rio + ecstore |
+| `concurrency` | 1.8K | Concurrency control wrappers over io-core |
+
+**Storage Engine:**
+
+| Crate | Lines | Purpose |
+|-------|-------|---------|
+| `ecstore` | 87K | ⚠️ Erasure-coded storage: disks, pools, buckets, replication, lifecycle |
+| `filemeta` | 10K | File/object metadata types and versioning |
+| `checksums` | 732 | Checksum computation |
+| `lock` | 7.1K | Distributed lock manager |
+| `heal` | 5.9K | Data healing / bitrot repair |
+| `scanner` | 5.4K | Background data usage scanner |
+| `object-capacity` | 2.5K | Capacity tracking and management |
+
+**Security & Auth:**
+
+| Crate | Lines | Purpose |
+|-------|-------|---------|
+| `crypto` | 1.6K | Encryption primitives |
+| `credentials` | 713 | Credential types (access key / secret key) |
+| `signer` | 1.4K | S3 v4 request signing |
+| `iam` | 9.0K | Identity and access management |
+| `policy` | 8.8K | Policy engine (S3 bucket/IAM policies) |
+| `kms` | 8.1K | Key management service integration |
+| `keystone` | 1.9K | OpenStack Keystone auth |
+| `appauth` | 143 | Application-level auth tokens |
+
+**Protocol & API:**
+
+| Crate | Lines | Purpose |
+|-------|-------|---------|
+| `protos` | 5.7K | Protobuf/gRPC definitions for inter-node RPC |
+| `protocols` | 18K | FTP/FTPS, WebDAV, Swift API support |
+| `s3-common` | 738 | Shared S3 types |
+| `s3select-api` | 1.9K | S3 Select interface |
+| `s3select-query` | 3.6K | S3 Select query engine |
+
+**Observability:**
+
+| Crate | Lines | Purpose |
+|-------|-------|---------|
+| `metrics` | 8.4K | Prometheus metric collectors |
+| `io-metrics` | 4.5K | I/O-specific metrics |
+| `obs` | 5.6K | OpenTelemetry tracing and telemetry |
+| `audit` | 2.4K | Audit logging |
+
+**Events:**
+
+| Crate | Lines | Purpose |
+|-------|-------|---------|
+| `notify` | 5.5K | Event notification system |
+| `targets` | 3.2K | Notification targets (Kafka, AMQP, webhook, etc.) |
+
+**Other:**
+
+| Crate | Lines | Purpose |
+|-------|-------|---------|
+| `trusted-proxies` | 4.0K | Trusted proxy / IP forwarding |
+| `zip` | 986 | ZIP archive support for bulk downloads |
+| `workers` | 136 | Simple worker abstraction |
+
+## Architecture Invariants
+
+> These are rules that the codebase should follow. Some are currently violated
+> (marked with ⚠️). Documenting them here makes the violations explicit and
+> trackable.
+
+1. **Layers flow downward.** Server → Admin/App → Storage → ecstore → rio/io-core.
+   No upward imports.
+
+2. **Leaf crates have zero internal dependencies.** `config`, `credentials`, `crypto`,
+   `io-metrics`, `madmin`, `s3-common` should depend only on external crates.
+   - ⚠️ VIOLATED: `utils` depends on `config`, `common` depends on `filemeta` and `madmin`.
+
+3. **Each type has exactly one definition.** Types shared across crates must be defined
+   in one crate and re-exported or imported by others.
+   - ⚠️ VIOLATED: `ReplicationStats` (4 copies), `LastMinuteLatency` (3 copies),
+     `BackpressureConfig` (3 copies), `DataUsageInfo` (2 copies).
+
+4. **ecstore does not know about HTTP or S3 protocol details.** It operates on
+   storage-level abstractions (objects, buckets, disks, pools).
+
+5. **The `rustfs` binary crate is the only place that wires everything together.**
+   Individual crates should be testable in isolation.
+
+6. **Error types use `thiserror` with descriptive names** (e.g., `StorageError`,
+   not bare `Error`).
+   - ⚠️ VIOLATED: 6 crates use `pub enum Error`; 2 crates use `snafu`;
+      `heal` use `anyhow` in library code.
+
+## Known Structural Issues
+
+> This section documents known problems in the current architecture.
+> It exists so the team can track and address them deliberately.
+
+### Critical
+
+- **common/scanner code duplication (~3K lines).** `scanner` depends on `common`
+  but maintains its own copies of `DataUsageInfo`, `LastMinuteLatency`, and related
+  types instead of importing them.
+
+- **ecstore is a monolith (87K lines, 163 files).** It contains disk management,
+  bucket management, erasure coding, replication, lifecycle, RPC, and configuration
+  — all in one crate. It should be decomposed along its existing subdirectories.
+
+### High
+
+- **Dependency inversions.** `utils → config` and `common → filemeta/madmin` break
+  the layering model. These need to be untangled.
+
+- **Three-layer BackpressureConfig/DeadlockConfig duplication** across io-core,
+  concurrency, and rustfs/storage. Should be defined once with builder/composition.
+
+### Medium
+
+- **Inconsistent error handling.** Three strategies (thiserror/snafu/anyhow) and
+  mixed naming (bare `Error` vs descriptive names).
+
+- **Ambiguous common vs utils boundary.** Both described as "utilities and data
+  structures." Need clear ownership rules.
+
+## Cross-Cutting Concerns
+
+### Error Handling
+
+The project convention is `thiserror` for typed errors with descriptive names.
+See `AGENTS.md`: "Prefer thiserror for library-facing error types."
+
+```rust
+// GOOD
+#[derive(Debug, thiserror::Error)]
+pub enum StorageError {
+    #[error("disk not found: {0}")]
+    DiskNotFound(String),
+}
+
+// AVOID
+pub enum Error { ... }        // too generic
+anyhow::Result<T>             // in library code (OK in tests/CLI)
+```
+
+### Logging & Tracing
+
+- Use `tracing` crate (`info!`, `warn!`, `error!`, `debug!`, `trace!`)
+- Structured fields: `tracing::info!(bucket = %name, "created bucket")`
+- Spans for request-scoped context
+
+### Metrics
+
+- Prometheus-style metrics via `rustfs-obs` runtime and schema
+- I/O-specific counters via `rustfs-io-metrics`
+- Registration happens at crate level, collection/reporting in `rustfs-obs`
+
+### Testing
+
+- Unit tests: `#[cfg(test)] mod tests` in the same file
+- Integration tests: inside respective crates (not top-level `tests/`)
+- E2E tests: `crates/e2e_test/` — tests against a running server
+- Run all: `make test` or `cargo nextest run`
+
+## Startup Sequence
+
+The binary (`main.rs`) boots in this order:
+
+1. Environment variable compatibility (`MINIO_*` → `RUSTFS_*`)
+2. Tokio runtime construction
+3. CLI argument parsing
+4. License, observability, TLS, trusted proxies initialization
+5. Config parsing, server address resolution
+6. Credentials, endpoints, local disks, lock client initialization
+7. Capacity management initialization
+8. HTTP server start (S3 API + optional console)
+9. ECStore initialization (erasure coding storage engine)
+10. Global config, background replication, KMS
+11. Optional: FTP/FTPS/WebDAV servers
+12. Event notifier, audit system, deadlock detector
+13. Bucket metadata, IAM, Keystone, OIDC
+14. Scanner and heal manager
+15. Metrics system, mark `FullReady`
+16. Wait for shutdown signal → graceful shutdown
+
+## Dependency Diagram (Simplified)
+
+```
+                            ┌─────────┐
+                            │  rustfs │  (binary + lib, 75K lines)
+                            │  main   │
+                            └────┬────┘
+                                 │
+                 ┌───────────────┼───────────────┐
+                 │               │               │
+            ┌────▼────┐    ┌────▼────┐    ┌─────▼─────┐
+            │ server  │    │  admin  │    │    app     │
+            │ (HTTP)  │    │(console)│    │(use-cases) │
+            └────┬────┘    └────┬────┘    └─────┬─────┘
+                 │               │               │
+                 └───────────────┼───────────────┘
+                                 │
+                          ┌──────▼──────┐
+                          │   storage   │
+                          │ (ecfs, SSE, │
+                          │  RPC, ACL)  │
+                          └──────┬──────┘
+                                 │
+              ┌──────────────────┼──────────────────┐
+              │                  │                  │
+        ┌─────▼─────┐    ┌──────▼──────┐    ┌──────▼──────┐
+        │  ecstore   │    │     rio     │    │   io-core   │
+        │ (87K,core) │    │  (readers)  │    │ (zero-copy) │
+        └─────┬──────┘    └─────────────┘    └─────────────┘
+              │
+    ┌─────┬──┼──┬─────┬──────┐
+    │     │  │  │     │      │
+ common utils config policy filemeta ...
+```
+
+## How to Navigate
+
+- **"Where does S3 PutObject go?"**
+  `server/` routes → `app/object_usecase` validates → `storage/ecfs` encodes →
+  `ecstore` distributes → `rio` encrypts/compresses → `io-core` writes
+
+- **"Where are bucket policies enforced?"**
+  `app/bucket_usecase` calls into `crates/policy/`
+
+- **"Where is replication configured?"**
+  `admin/handlers/replication.rs` and `admin/handlers/site_replication.rs` for API,
+  `ecstore/src/bucket/replication/` for engine
+
+- **"Where do I add a new admin endpoint?"**
+  Add handler in `admin/handlers/`, register in `admin/router.rs`
+
+- **"Where do I add a new metric?"**
+  Define descriptor/collector in `crates/obs/src/metrics/`, expose via `/minio/v2/metrics`
+
+---
+
+*Inspired by [matklad's ARCHITECTURE.md](https://matklad.github.io/2021/02/06/ARCHITECTURE.md.html)
+and [rust-analyzer's architecture.md](https://github.com/rust-analyzer/rust-analyzer/blob/master/docs/book/src/contributing/architecture.md).*

Cargo.toml

@@ -31,14 +31,14 @@ members = [
     "crates/kms", # Key Management Service
     "crates/lock", # Distributed locking implementation
     "crates/madmin", # Management dashboard and admin API interface
-    "crates/mcp", # MCP server for S3 operations
-    "crates/metrics", # Metrics collection and reporting
     "crates/notify", # Notification system for events
     "crates/obs", # Observability utilities
+    "crates/object-capacity", # Capacity scan and refresh core
     "crates/policy", # Policy management
     "crates/protocols", # Protocol implementations (FTPS, SFTP, etc.)
     "crates/protos", # Protocol buffer definitions
     "crates/rio", # Rust I/O utilities and abstractions
+    "crates/concurrency", # Concurrency management for RustFS - timeout, locking, backpressure, and I/O scheduling
     "crates/s3-common", # Common utilities and data structures for S3 compatibility
     "crates/s3select-api", # S3 Select API interface
     "crates/s3select-query", # S3 Select query engine
@@ -48,6 +48,8 @@ members = [
     "crates/trusted-proxies", # Trusted proxies management
     "crates/utils", # Utility functions and helpers
     "crates/workers", # Worker thread pools and task scheduling
+    "crates/io-metrics", # Zero-copy metrics collection for performance analysis
+    "crates/io-core", # Zero-copy core reader and writer implementations
     "crates/zip", # ZIP file handling and compression
 ]
 resolver = "3"
@@ -57,7 +59,7 @@ edition = "2024"
 license = "Apache-2.0"
 repository = "https://github.com/rustfs/rustfs"
 rust-version = "1.93.0"
-version = "0.0.5"
+version = "1.0.0-beta.1"
 homepage = "https://rustfs.com"
 description = "RustFS is a high-performance distributed object storage software built using Rust, one of the most popular languages worldwide. "
 keywords = ["RustFS", "Minio", "object-storage", "filesystem", "s3"]
@@ -69,63 +71,70 @@ unsafe_code = "deny"
 
 [workspace.lints.clippy]
 all = "warn"
+needless_collect = "warn"
+redundant_clone = "warn"
 
 [workspace.dependencies]
 # RustFS Internal Crates
-rustfs = { path = "./rustfs", version = "0.0.5" }
-rustfs-heal = { path = "crates/heal", version = "0.0.5" }
-rustfs-appauth = { path = "crates/appauth", version = "0.0.5" }
-rustfs-audit = { path = "crates/audit", version = "0.0.5" }
-rustfs-checksums = { path = "crates/checksums", version = "0.0.5" }
-rustfs-common = { path = "crates/common", version = "0.0.5" }
-rustfs-config = { path = "./crates/config", version = "0.0.5" }
-rustfs-credentials = { path = "crates/credentials", version = "0.0.5" }
-rustfs-crypto = { path = "crates/crypto", version = "0.0.5" }
-rustfs-ecstore = { path = "crates/ecstore", version = "0.0.5" }
-rustfs-filemeta = { path = "crates/filemeta", version = "0.0.5" }
-rustfs-iam = { path = "crates/iam", version = "0.0.5" }
-rustfs-keystone = { path = "crates/keystone", version = "0.0.5" }
-rustfs-kms = { path = "crates/kms", version = "0.0.5" }
-rustfs-lock = { path = "crates/lock", version = "0.0.5" }
-rustfs-madmin = { path = "crates/madmin", version = "0.0.5" }
-rustfs-mcp = { path = "crates/mcp", version = "0.0.5" }
-rustfs-metrics = { path = "crates/metrics", version = "0.0.5" }
-rustfs-notify = { path = "crates/notify", version = "0.0.5" }
-rustfs-obs = { path = "crates/obs", version = "0.0.5" }
-rustfs-policy = { path = "crates/policy", version = "0.0.5" }
-rustfs-protos = { path = "crates/protos", version = "0.0.5" }
-rustfs-rio = { path = "crates/rio", version = "0.0.5" }
-rustfs-s3-common = { path = "crates/s3-common", version = "0.0.5" }
-rustfs-s3select-api = { path = "crates/s3select-api", version = "0.0.5" }
-rustfs-s3select-query = { path = "crates/s3select-query", version = "0.0.5" }
-rustfs-scanner = { path = "crates/scanner", version = "0.0.5" }
-rustfs-signer = { path = "crates/signer", version = "0.0.5" }
-rustfs-trusted-proxies = { path = "crates/trusted-proxies", version = "0.0.5" }
-rustfs-targets = { path = "crates/targets", version = "0.0.5" }
-rustfs-utils = { path = "crates/utils", version = "0.0.5" }
-rustfs-workers = { path = "crates/workers", version = "0.0.5" }
-rustfs-zip = { path = "./crates/zip", version = "0.0.5" }
-rustfs-protocols = { path = "crates/protocols", version = "0.0.5" }
+rustfs = { path = "./rustfs", version = "1.0.0-beta.1" }
+rustfs-heal = { path = "crates/heal", version = "1.0.0-beta.1" }
+rustfs-appauth = { path = "crates/appauth", version = "1.0.0-beta.1" }
+rustfs-audit = { path = "crates/audit", version = "1.0.0-beta.1" }
+rustfs-checksums = { path = "crates/checksums", version = "1.0.0-beta.1" }
+rustfs-common = { path = "crates/common", version = "1.0.0-beta.1" }
+rustfs-config = { path = "./crates/config", version = "1.0.0-beta.1" }
+rustfs-concurrency = { path = "./crates/concurrency", version = "1.0.0-beta.1" }
+rustfs-credentials = { path = "crates/credentials", version = "1.0.0-beta.1" }
+rustfs-crypto = { path = "crates/crypto", version = "1.0.0-beta.1" }
+rustfs-ecstore = { path = "crates/ecstore", version = "1.0.0-beta.1" }
+rustfs-filemeta = { path = "crates/filemeta", version = "1.0.0-beta.1" }
+rustfs-iam = { path = "crates/iam", version = "1.0.0-beta.1" }
+rustfs-keystone = { path = "crates/keystone", version = "1.0.0-beta.1" }
+rustfs-kms = { path = "crates/kms", version = "1.0.0-beta.1" }
+rustfs-lock = { path = "crates/lock", version = "1.0.0-beta.1" }
+rustfs-madmin = { path = "crates/madmin", version = "1.0.0-beta.1" }
+rustfs-notify = { path = "crates/notify", version = "1.0.0-beta.1" }
+rustfs-io-metrics = { path = "crates/io-metrics", version = "1.0.0-beta.1" }
+rustfs-io-core = { path = "crates/io-core", version = "1.0.0-beta.1" }
+rustfs-object-capacity = { path = "crates/object-capacity", version = "1.0.0-beta.1" }
+rustfs-obs = { path = "crates/obs", version = "1.0.0-beta.1" }
+rustfs-policy = { path = "crates/policy", version = "1.0.0-beta.1" }
+rustfs-protos = { path = "crates/protos", version = "1.0.0-beta.1" }
+rustfs-protocols = { path = "crates/protocols", version = "1.0.0-beta.1" }
+rustfs-rio = { path = "crates/rio", version = "1.0.0-beta.1" }
+rustfs-s3-common = { path = "crates/s3-common", version = "1.0.0-beta.1" }
+rustfs-s3select-api = { path = "crates/s3select-api", version = "1.0.0-beta.1" }
+rustfs-s3select-query = { path = "crates/s3select-query", version = "1.0.0-beta.1" }
+rustfs-scanner = { path = "crates/scanner", version = "1.0.0-beta.1" }
+rustfs-signer = { path = "crates/signer", version = "1.0.0-beta.1" }
+rustfs-trusted-proxies = { path = "crates/trusted-proxies", version = "1.0.0-beta.1" }
+rustfs-targets = { path = "crates/targets", version = "1.0.0-beta.1" }
+rustfs-utils = { path = "crates/utils", version = "1.0.0-beta.1" }
+rustfs-workers = { path = "crates/workers", version = "1.0.0-beta.1" }
+rustfs-zip = { path = "./crates/zip", version = "1.0.0-beta.1" }
 
 # Async Runtime and Networking
 async-channel = "2.5.0"
-async-compression = { version = "0.4.41" }
+async-compression = { version = "0.4.42" }
 async-recursion = "1.1.1"
 async-trait = "0.1.89"
-axum = "0.8.8"
+async-nats = "0.47.0"
+axum = "0.8.9"
 futures = "0.3.32"
 futures-core = "0.3.32"
 futures-util = "0.3.32"
 pollster = "0.4.0"
-hyper = { version = "1.8.1", features = ["http2", "http1", "server"] }
-hyper-rustls = { version = "0.27.7", default-features = false, features = ["native-tokio", "http1", "tls12", "logging", "http2", "aws-lc-rs", "webpki-roots"] }
+pulsar = { version = "6.7.2", default-features = false, features = ["tokio-rustls-runtime"] }
+hyper = { version = "1.9.0", features = ["http2", "http1", "server"] }
+hyper-rustls = { version = "0.27.9", default-features = false, features = ["native-tokio", "http1", "tls12", "logging", "http2", "aws-lc-rs", "webpki-roots"] }
 hyper-util = { version = "0.1.20", features = ["tokio", "server-auto", "server-graceful", "tracing"] }
 http = "1.4.0"
 http-body = "1.0.1"
 http-body-util = "0.1.3"
-reqwest = { version = "0.13.2", default-features = false, features = ["rustls", "charset", "http2", "system-proxy", "stream", "json", "blocking", "query", "form"] }
+reqwest = { version = "0.13.3", default-features = false, features = ["rustls", "charset", "http2", "system-proxy", "stream", "json", "blocking", "query", "form"] }
+rustfs-kafka-async = { version = "1.2.0" }
 socket2 = { version = "0.6.3", features = ["all"] }
-tokio = { version = "1.50.0", features = ["fs", "rt-multi-thread"] }
+tokio = { version = "1.52.1", features = ["fs", "rt-multi-thread"] }
 tokio-rustls = { version = "0.26.4", default-features = false, features = ["logging", "tls12", "aws-lc-rs"] }
 tokio-stream = { version = "0.1.18" }
 tokio-test = "0.4.5"
@@ -144,57 +153,53 @@ flatbuffers = "25.12.19"
 form_urlencoded = "1.2.2"
 prost = "0.14.3"
 quick-xml = "0.39.2"
-rmcp = { version = "1.2.0" }
 rmp = { version = "0.8.15" }
 rmp-serde = { version = "1.3.1" }
 serde = { version = "1.0.228", features = ["derive"] }
-serde_bytes = "0.11"
 serde_json = { version = "1.0.149", features = ["raw_value"] }
 serde_urlencoded = "0.7.1"
-schemars = "1.2.1"
 
 # Cryptography and Security
 aes-gcm = { version = "0.11.0-rc.3", features = ["rand_core"] }
-argon2 = { version = "0.6.0-rc.7" }
-blake2 = "0.11.0-rc.5"
-blake3 = { version = "1.8.3", features = ["rayon", "mmap"] }
+argon2 = { version = "0.6.0-rc.8" }
+blake2 = "0.11.0-rc.6"
 chacha20poly1305 = { version = "0.11.0-rc.3" }
 crc-fast = "1.9.0"
-hmac = { version = "0.13.0-rc.5" }
+hmac = { version = "0.13.0" }
 jsonwebtoken = { version = "10.3.0", features = ["aws_lc_rs"] }
 openidconnect = { version = "4.0", default-features = false }
-pbkdf2 = "0.13.0-rc.9"
-rsa = { version = "0.10.0-rc.17" }
-rustls = { version = "0.23.37", default-features = false, features = ["aws-lc-rs", "logging", "tls12", "prefer-post-quantum", "std"] }
-rustls-pki-types = "1.14.0"
-sha1 = "0.11.0-rc.5"
-sha2 = "0.11.0-rc.5"
+pbkdf2 = "0.13.0"
+rsa = { version = "0.10.0-rc.18" }
+rustls = { version = "0.23.39", default-features = false, features = ["aws-lc-rs", "logging", "tls12", "prefer-post-quantum", "std"] }
+rustls-pki-types = "1.14.1"
+sha1 = "0.11.0"
+sha2 = "0.11.0"
 subtle = "2.6"
 zeroize = { version = "1.8.2", features = ["derive"] }
 
 # Time and Date
 chrono = { version = "0.4.44", features = ["serde"] }
 humantime = "2.3.0"
-jiff = { version = "0.2.23", features = ["serde"] }
+jiff = { version = "0.2.24", features = ["serde"] }
 time = { version = "0.3.47", features = ["std", "parsing", "formatting", "macros", "serde"] }
 
 # Utilities and Tools
 anyhow = "1.0.102"
-arc-swap = "1.8.2"
-astral-tokio-tar = "0.6.0"
+arc-swap = "1.9.1"
+astral-tokio-tar = "0.6.1"
 atoi = "2.0.0"
 atomic_enum = "0.3.0"
-aws-config = { version = "1.8.15" }
+aws-config = { version = "1.8.16" }
 aws-credential-types = { version = "1.2.14" }
-aws-sdk-s3 = { version = "1.126.0", default-features = false, features = ["sigv4a", "default-https-client", "rt-tokio"] }
+aws-sdk-s3 = { version = "1.131.0", default-features = false, features = ["sigv4a", "default-https-client", "rt-tokio"] }
 aws-smithy-http-client = { version = "1.1.12", default-features = false, features = ["default-client", "rustls-aws-lc"] }
-aws-smithy-types = { version = "1.4.6" }
+aws-smithy-types = { version = "1.4.7" }
 backtrace = "0.3.76"
 base64 = "0.22.1"
 base64-simd = "0.8.0"
 brotli = "8.0.2"
 cfg-if = "1.0.4"
-clap = { version = "4.6.0", features = ["derive", "env"] }
+clap = { version = "4.6.1", features = ["derive", "env"] }
 const-str = { version = "1.1.0", features = ["std", "proc"] }
 convert_case = "0.11.0"
 criterion = { version = "0.8", features = ["html_reports"] }
@@ -202,52 +207,53 @@ crossbeam-queue = "0.3.12"
 crossbeam-channel = "0.5.15"
 crossbeam-deque = "0.8.6"
 crossbeam-utils = "0.8.21"
-datafusion = "52.3.0"
+datafusion = "53.1.0"
 derive_builder = "0.20.2"
 enumset = "1.1.10"
 faster-hex = "0.10.0"
 flate2 = "1.1.9"
 glob = "0.3.3"
-google-cloud-storage = "1.9.0"
-google-cloud-auth = "1.7.0"
-hashbrown = { version = "0.16.1", features = ["serde", "rayon"] }
+google-cloud-storage = "1.11.0"
+google-cloud-auth = "1.9.0"
+hashbrown = { version = "0.17.0", features = ["serde", "rayon"] }
 hex = "0.4.3"
 hex-simd = "0.8.0"
 highway = { version = "1.3.0" }
 ipnetwork = { version = "0.21.1", features = ["serde"] }
 lazy_static = "1.5.0"
-libc = "0.2.183"
+libc = "0.2.186"
 libsystemd = "0.7.2"
-local-ip-address = "0.6.10"
+local-ip-address = "0.6.12"
+memmap2 = "0.9.10"
 lz4 = "1.28.1"
-matchit = "0.9.1"
-md-5 = "0.11.0-rc.5"
+matchit = "0.9.2"
+md-5 = "0.11.0"
 md5 = "0.8.0"
 mime_guess = "2.0.5"
-moka = { version = "0.12.14", features = ["future"] }
+moka = { version = "0.12.15", features = ["future"] }
 netif = "0.1.6"
 num_cpus = { version = "1.17.0" }
-nvml-wrapper = "0.12.0"
-object_store = "0.12.5"
+nvml-wrapper = "0.12.1"
+object_store = "0.13.2"
 parking_lot = "0.12.5"
 path-absolutize = "3.1.1"
 path-clean = "1.0.1"
 percent-encoding = "2.3.2"
 pin-project-lite = "0.2.17"
 pretty_assertions = "1.4.1"
-rand = { version = "0.10.0", features = ["serde"] }
-ratelimit = "0.10.0"
-rayon = "1.11.0"
+rand = { version = "0.10.1", features = ["serde"] }
+ratelimit = "0.10.1"
+rayon = "1.12.0"
 reed-solomon-erasure = { version = "6.0", default-features = false, features = ["std", "simd-accel"] }
 reed-solomon-simd = "3.1.0"
 regex = { version = "1.12.3" }
-rumqttc = { version = "0.25.1" }
+rumqttc = { package = "rumqttc-next", version = "0.30.0", features = ["websocket"] }
 rustix = { version = "1.1.4", features = ["fs"] }
 rust-embed = { version = "8.11.0" }
-rustc-hash = { version = "2.1.1" }
-s3s = { git = "https://github.com/rustfs/s3s", rev = "d9556e3c0036bd3f2b330966009cbaa5aebf19a3", features = ["minio"] }
+rustc-hash = { version = "2.1.2" }
+s3s = { git = "https://github.com/rustfs/s3s", rev = "a3b16608df35aaeed8fff08b4988d03f4ca9445b", features = ["minio"] }
 serial_test = "3.4.0"
-shadow-rs = { version = "1.7.1", default-features = false }
+shadow-rs = { version = "2.0.0", default-features = false }
 siphasher = "1.0.2"
 smallvec = { version = "1.15.1", features = ["serde"] }
 smartstring = "1.0.1"
@@ -261,36 +267,37 @@ tempfile = "3.27.0"
 test-case = "3.3.1"
 thiserror = "2.0.18"
 tracing = { version = "0.1.44" }
-tracing-appender = "0.2.4"
+tracing-appender = "0.2.5"
 tracing-error = "0.2.1"
 tracing-opentelemetry = "0.32.1"
 tracing-subscriber = { version = "0.3.23", features = ["env-filter", "time"] }
 transform-stream = "0.3.1"
 url = "2.5.8"
 urlencoding = "2.1.3"
-uuid = { version = "1.22.0", features = ["v4", "fast-rng", "macro-diagnostics"] }
-vaultrs = { version = "0.7.4" }
+uuid = { version = "1.23.1", features = ["v4", "fast-rng", "macro-diagnostics"] }
+vaultrs = { version = "0.8.0" }
 walkdir = "2.5.0"
 wildmatch = { version = "2.6.1", features = ["serde"] }
 windows = { version = "0.62.2" }
 xxhash-rust = { version = "0.8.15", features = ["xxh64", "xxh3"] }
-zip = "8.2.0"
+zip = "8.6.0"
 zstd = "0.13.3"
 
 # Observability and Metrics
 metrics = "0.24.3"
+dial9-tokio-telemetry = "0.3"
 opentelemetry = { version = "0.31.0" }
 opentelemetry-appender-tracing = { version = "0.31.1", features = ["experimental_use_tracing_span_context", "experimental_metadata_attributes", "spec_unstable_logs_enabled"] }
-opentelemetry-otlp = { version = "0.31.0", features = ["gzip-http", "reqwest-rustls"] }
+opentelemetry-otlp = { version = "0.31.1", features = ["gzip-http", "reqwest-rustls"] }
 opentelemetry_sdk = { version = "0.31.0" }
 opentelemetry-semantic-conventions = { version = "0.31.0", features = ["semconv_experimental"] }
 opentelemetry-stdout = { version = "0.31.0" }
-pyroscope = { version = "2.0.0", features = ["backend-pprof-rs"] }
+pyroscope = { version = "2.0.3", features = ["backend-pprof-rs"] }
 
 # FTP and SFTP
 libunftp = { version = "0.23.0", features = ["experimental"] }
 unftp-core = "0.1.0"
-suppaftp = { version = "8.0.2", features = ["tokio", "tokio-rustls-aws-lc-rs"] }
+suppaftp = { version = "8.0.3", features = ["tokio", "tokio-rustls-aws-lc-rs"] }
 rcgen = "0.14.7"
 
 # WebDAV
@@ -310,7 +317,7 @@ jemalloc_pprof = { version = "0.8.2", features = ["symbolize", "flamegraph"] }
 pprof = { package = "pprof-pyroscope-fork", version = "0.1500.3", features = ["flamegraph", "protobuf-codec"] }
 
 [workspace.metadata.cargo-shear]
-ignored = ["rustfs", "rustfs-mcp"]
+ignored = ["rustfs"]
 
 [profile.release]
 opt-level = 3

Dockerfile

@@ -12,7 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-FROM alpine:3.23 AS build
+FROM alpine:3.23.4 AS build
 
 ARG TARGETARCH
 ARG RELEASE=latest
@@ -54,7 +54,7 @@ RUN set -eux; \
     rm -rf rustfs.zip /build/.tmp || true
 
 
-FROM alpine:3.23
+FROM alpine:3.23.4
 
 ARG RELEASE=latest
 ARG BUILD_DATE
@@ -72,7 +72,8 @@ LABEL name="RustFS" \
       url="https://rustfs.com" \
       license="Apache-2.0"
 
-RUN apk add --no-cache ca-certificates coreutils curl
+RUN apk update && \
+    apk add --no-cache ca-certificates coreutils curl
 
 COPY --from=build /etc/ssl/certs/ca-certificates.crt /etc/ssl/certs/
 COPY --from=build /build/rustfs /usr/bin/rustfs

Dockerfile.decommission-local

@@ -12,7 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-FROM rust:1.91-trixie
+FROM rust:1.93-trixie
 
 RUN set -eux; \
     export DEBIAN_FRONTEND=noninteractive; \

Dockerfile.source

@@ -30,7 +30,7 @@ ARG BUILDPLATFORM
 # -----------------------------
 # Build stage
 # -----------------------------
-FROM rust:1.91-trixie AS builder
+FROM rust:1.93-trixie AS builder
 
 # Re-declare args after FROM
 ARG TARGETPLATFORM
@@ -60,7 +60,15 @@ RUN set -eux; \
 
 # Optional: cross toolchain for aarch64 (only when targeting linux/arm64)
 RUN set -eux; \
-    if [ "${TARGETPLATFORM:-linux/amd64}" = "linux/arm64" ]; then \
+    target_platform="${TARGETPLATFORM:-}"; \
+    if [ -z "${target_platform}" ]; then \
+    case "$(uname -m)" in \
+    x86_64) target_platform="linux/amd64" ;; \
+    aarch64|arm64) target_platform="linux/arm64" ;; \
+    *) target_platform="linux/amd64" ;; \
+    esac; \
+    fi; \
+    if [ "${target_platform}" = "linux/arm64" ]; then \
     export DEBIAN_FRONTEND=noninteractive; \
     apt-get update; \
     apt-get install -y --no-install-recommends gcc-aarch64-linux-gnu; \
@@ -120,7 +128,16 @@ RUN --mount=type=cache,target=/usr/local/cargo/registry \
     --mount=type=cache,target=/usr/local/cargo/git \
     --mount=type=cache,target=/usr/src/rustfs/target \
     set -eux; \
-    case "${TARGETPLATFORM:-linux/amd64}" in \
+    rustup target add x86_64-unknown-linux-gnu aarch64-unknown-linux-gnu; \
+    target_platform="${TARGETPLATFORM:-}"; \
+    if [ -z "${target_platform}" ]; then \
+    case "$(uname -m)" in \
+    x86_64) target_platform="linux/amd64" ;; \
+    aarch64|arm64) target_platform="linux/arm64" ;; \
+    *) target_platform="linux/amd64" ;; \
+    esac; \
+    fi; \
+    case "${target_platform}" in \
     linux/amd64) \
     echo "Building for x86_64-unknown-linux-gnu"; \
     cargo build --release --locked --target x86_64-unknown-linux-gnu --bin rustfs -j "$(nproc)"; \
@@ -132,7 +149,7 @@ RUN --mount=type=cache,target=/usr/local/cargo/registry \
     install -m 0755 target/aarch64-unknown-linux-gnu/release/rustfs /usr/local/bin/rustfs \
     ;; \
     *) \
-    echo "Unsupported TARGETPLATFORM=${TARGETPLATFORM}" >&2; exit 1 \
+    echo "Unsupported target platform=${target_platform}" >&2; exit 1 \
     ;; \
     esac
 
@@ -178,7 +195,7 @@ CMD ["cargo", "run", "--bin", "rustfs", "--"]
 # -----------------------------
 # Runtime stage (Ubuntu minimal)
 # -----------------------------
-FROM ubuntu:22.04
+FROM ubuntu:24.04
 
 ARG BUILD_DATE
 ARG VCS_REF

README.md

@@ -100,28 +100,28 @@ To get started with RustFS, follow these steps:
 curl -O https://rustfs.com/install_rustfs.sh && bash install_rustfs.sh
 ```
 
-### 2\. Docker Quick Start (Option 2)
+### 2. Docker Quick Start (Option 2)
 
 The RustFS container runs as a non-root user `rustfs` (UID `10001`). If you run Docker with `-v` to mount a host directory, please ensure the host directory owner is set to `10001`, otherwise you will encounter permission denied errors.
 
 ```bash
- # Create data and logs directories
- mkdir -p data logs
+# Create data and logs directories
+mkdir -p data logs
 
- # Change the owner of these directories
- chown -R 10001:10001 data logs
+# Change the owner of these directories
+chown -R 10001:10001 data logs
 
- # Using latest version
- docker run -d -p 9000:9000 -p 9001:9001 -v $(pwd)/data:/data -v $(pwd)/logs:/logs rustfs/rustfs:latest
+# Using latest version
+docker run -d -p 9000:9000 -p 9001:9001 -v $(pwd)/data:/data -v $(pwd)/logs:/logs rustfs/rustfs:latest
 
- # Using specific version
- docker run -d -p 9000:9000 -p 9001:9001 -v $(pwd)/data:/data -v $(pwd)/logs:/logs rustfs/rustfs:1.0.0-alpha.76
+# Using specific version
+docker run -d -p 9000:9000 -p 9001:9001 -v $(pwd)/data:/data -v $(pwd)/logs:/logs rustfs/rustfs:v1.0.0-beta.1
 ```
 
 If you use [podman](https://github.com/containers/podman) instead of docker, you can install the RustFS with the below command
 
 ```bash
- podman run -d -p 9000:9000 -p 9001:9001 -v $(pwd)/data:/data -v $(pwd)/logs:/logs rustfs/rustfs:latest
+podman run -d -p 9000:9000 -p 9001:9001 -v $(pwd)/data:/data -v $(pwd)/logs:/logs rustfs/rustfs:latest
 ```
 
 You can also use Docker Compose. Using the `docker-compose.yml` file in the root directory:
@@ -136,7 +136,7 @@ Similarly, you can run the command with podman
 podman compose --profile observability up -d
 ```
 
-**NOTE**: We recommend reviewing the `docker-compose.yaml` file before running. It defines several services including Grafana, Prometheus, and Jaeger, which are helpful for RustFS observability. If you wish to start Redis or Nginx containers, you can specify the corresponding profiles.
+**NOTE**: We recommend reviewing the `docker-compose.yml` file before running. It defines several services including Grafana, Prometheus, and Jaeger, which are helpful for RustFS observability. If you wish to start Redis or Nginx containers, you can specify the corresponding profiles.
 
 ### 3\. Build from Source (Option 3) - Advanced Users
 
@@ -195,6 +195,19 @@ nix build
 nix run
 ```
 
+### 6\. X-CMD (Option 6)
+
+If you are an [x-cmd](https://www.x-cmd.com/install/rustfs) user:
+
+```bash
+# Run directly without installing
+x rustfs
+
+# Download the binary and install it to the global environment
+x env use rustfs
+rustfs --help
+```
+
 ---
 
 ### Accessing RustFS
@@ -206,6 +219,48 @@ nix run
 
 **NOTE**: To access the RustFS instance via `https`, please refer to the [TLS Configuration Docs](https://docs.rustfs.com/integration/tls-configured.html).
 
+### OIDC Roles Claim (Microsoft Entra ID)
+
+RustFS supports mapping an OIDC claim containing role values into the existing
+authorization pipeline. The `roles_claim` setting is **optional**: when unset or
+empty, only the `groups` claim contributes to authorization (same as older
+RustFS releases). For Microsoft Entra ID app roles, set `roles_claim=roles` so
+both console admin checks and bucket IAM policies can evaluate those roles.
+
+Example environment configuration (opt-in roles claim):
+
+```bash
+RUSTFS_IDENTITY_OPENID_ENABLE=on
+RUSTFS_IDENTITY_OPENID_CONFIG_URL="https://login.microsoftonline.com/<tenant-id>/v2.0/.well-known/openid-configuration"
+RUSTFS_IDENTITY_OPENID_CLIENT_ID="<client-id>"
+RUSTFS_IDENTITY_OPENID_CLIENT_SECRET="<client-secret>"
+RUSTFS_IDENTITY_OPENID_SCOPES="openid,profile,email"
+RUSTFS_IDENTITY_OPENID_GROUPS_CLAIM="groups"
+RUSTFS_IDENTITY_OPENID_ROLES_CLAIM="roles"
+```
+
+Policy condition example (evaluate app roles directly with `jwt:roles`; when
+`roles_claim` is configured, RustFS also merges those values into `jwt:groups`
+for backward compatibility with older policies):
+
+```json
+{
+  "Version": "2012-10-17",
+  "Statement": [
+    {
+      "Effect": "Allow",
+      "Action": ["admin:*"],
+      "Resource": ["arn:aws:s3:::*"],
+      "Condition": {
+        "ForAnyValue:StringEquals": {
+          "jwt:roles": ["RustFS.ConsoleAdmin"]
+        }
+      }
+    }
+  ]
+}
+```
+
 ## Documentation
 
 For detailed documentation, including configuration options, API references, and advanced usage, please visit our [Documentation](https://docs.rustfs.com).

README_ZH.md

@@ -113,7 +113,7 @@ RustFS 容器以非 root 用户 `rustfs` (UID `10001`) 运行。如果您使用
  docker run -d -p 9000:9000 -p 9001:9001 -v $(pwd)/data:/data -v $(pwd)/logs:/logs rustfs/rustfs:latest
 
  # 使用指定版本运行
- docker run -d -p 9000:9000 -p 9001:9001 -v $(pwd)/data:/data -v $(pwd)/logs:/logs rustfs/rustfs:1.0.0.alpha.68
+ docker run -d -p 9000:9000 -p 9001:9001 -v $(pwd)/data:/data -v $(pwd)/logs:/logs rustfs/rustfs:v1.0.0-beta.1
 ```
 
 您也可以使用 Docker Compose。使用根目录下的 `docker-compose.yml` 文件：
@@ -122,7 +122,7 @@ RustFS 容器以非 root 用户 `rustfs` (UID `10001`) 运行。如果您使用
 docker compose --profile observability up -d
 ```
 
-**注意**: 我们建议您在运行前查看 `docker-compose.yaml` 文件。该文件定义了包括 Grafana、Prometheus 和 Jaeger 在内的多个服务，有助于 RustFS 的可观测性监控。如果您还想启动 Redis 或 Nginx 容器，可以指定相应的 profile。
+**注意**: 我们建议您在运行前查看 `docker-compose.yml` 文件。该文件定义了包括 Grafana、Prometheus 和 Jaeger 在内的多个服务，有助于 RustFS 的可观测性监控。如果您还想启动 Redis 或 Nginx 容器，可以指定相应的 profile。
 
 ### 3\. 源码编译 (选项 3) - 进阶用户
 
@@ -163,11 +163,41 @@ make help-docker                      # 显示所有 Docker 相关命令
 
 请按照 [Helm Chart README](https://charts.rustfs.com) 上的说明在 Kubernetes 集群上安装 RustFS。
 
+### 5\. Nix Flake (Option 5)
+
+如果你已经 启用了 [Nix Flakes 功能](https://nixos.wiki/wiki/Flakes#Enable_flakes):
+
+```bash
+# 直接运行，无需安装
+nix run github:rustfs/rustfs
+
+# 编译二进制文件
+nix build github:rustfs/rustfs
+./result/bin/rustfs --help
+
+# 或者从本地检出的代码库运行/编译
+nix build
+nix run
+```
+
+### 6\. X-CMD (Option 6)
+
+如果你是 [x-cmd](https://www.x-cmd.com/install/rustfs) 用户：
+
+```bash
+# 直接运行，无需安装
+x rustfs
+
+# 下载二进制文件并安装到全局环境中
+x env use rustfs
+rustfs --help
+```
+
 ---
 
 ### 访问 RustFS
 
-1. **访问控制台**: 打开浏览器并访问 `http://localhost:9000` 进入 RustFS 控制台。
+1. **访问控制台**: 打开浏览器并访问 `http://localhost:9001` 进入 RustFS 控制台。
    - 默认账号/密码: `rustfsadmin` / `rustfsadmin`
 2. **创建存储桶**: 使用控制台为您​​的对象创建一个新的存储桶 (Bucket)。
 3. **上传对象**: 您可以直接通过控制台上传文件，或使用 S3 兼容的 API/客户端与您的 RustFS 实例进行交互。
@@ -198,7 +228,7 @@ make help-docker                      # 显示所有 Docker 相关命令
 - **商务合作**: [hello@rustfs.com](mailto:hello@rustfs.com)
 - **工作机会**: [jobs@rustfs.com](mailto:jobs@rustfs.com)
 - **一般讨论**: [GitHub Discussions](https://github.com/rustfs/rustfs/discussions)
-- **贡献指南**: [CONTRIBUTING.md](https://www.google.com/search?q=CONTRIBUTING.md)
+- **贡献指南**: [CONTRIBUTING.md](CONTRIBUTING.md)
 
 ## 贡献者
 

_typos.toml

@@ -39,6 +39,7 @@ abd = "abd"
 mak = "mak"
 gae = "gae"
 GAE = "GAE"
+thr = "thr"
 # s3-tests original test names (cannot be changed)
 nonexisted = "nonexisted"
 consts = "consts"

build-rustfs.sh

@@ -213,7 +213,7 @@ setup_rust_environment() {
     # Set up environment variables for musl targets
     if [[ "$PLATFORM" == *"musl"* ]]; then
         print_message $YELLOW "Setting up environment for musl target..."
-        export RUSTFLAGS="-C target-feature=-crt-static"
+        export RUSTFLAGS="'--cfg tokio_unstable -C target-feature=-crt-static'"
 
         # For cargo-zigbuild, set up additional environment variables
         if command -v cargo-zigbuild &> /dev/null; then
@@ -430,7 +430,7 @@ build_binary() {
         fi
     else
         # Native compilation
-        build_cmd="RUSTFLAGS=-Clink-arg=-lm cargo build"
+        build_cmd="RUSTFLAGS='--cfg tokio_unstable -Clink-arg=-lm' cargo build"
     fi
 
     if [ "$BUILD_TYPE" = "release" ]; then

crates/appauth/Cargo.toml

@@ -33,3 +33,6 @@ rand.workspace = true
 
 [lints]
 workspace = true
+
+[lib]
+doctest = false

crates/audit/Cargo.toml

@@ -41,9 +41,14 @@ serde_json = { workspace = true }
 thiserror = { workspace = true }
 tokio = { workspace = true, features = ["sync", "fs", "rt-multi-thread", "rt", "time", "macros"] }
 tracing = { workspace = true, features = ["std", "attributes"] }
-url = { workspace = true }
-rumqttc = { workspace = true }
 
+[dev-dependencies]
+temp-env = { workspace = true }
+url = { workspace = true }
 
 [lints]
 workspace = true
+
+[lib]
+test = false
+doctest = false

crates/audit/src/factory.rs

@@ -14,24 +14,19 @@
 
 use crate::AuditEntry;
 use async_trait::async_trait;
-use hashbrown::HashSet;
-use rumqttc::QoS;
-use rustfs_config::audit::{AUDIT_MQTT_KEYS, AUDIT_WEBHOOK_KEYS, ENV_AUDIT_MQTT_KEYS, ENV_AUDIT_WEBHOOK_KEYS};
-use rustfs_config::{
-    AUDIT_DEFAULT_DIR, DEFAULT_LIMIT, MQTT_BROKER, MQTT_KEEP_ALIVE_INTERVAL, MQTT_PASSWORD, MQTT_QOS, MQTT_QUEUE_DIR,
-    MQTT_QUEUE_LIMIT, MQTT_RECONNECT_INTERVAL, MQTT_TOPIC, MQTT_USERNAME, RUSTFS_WEBHOOK_SKIP_TLS_VERIFY_DEFAULT,
-    WEBHOOK_AUTH_TOKEN, WEBHOOK_CLIENT_CA, WEBHOOK_CLIENT_CERT, WEBHOOK_CLIENT_KEY, WEBHOOK_ENDPOINT, WEBHOOK_QUEUE_DIR,
-    WEBHOOK_QUEUE_LIMIT, WEBHOOK_SKIP_TLS_VERIFY,
-};
+use rustfs_config::AUDIT_DEFAULT_DIR;
+use rustfs_config::audit::{AUDIT_KAFKA_KEYS, AUDIT_MQTT_KEYS, AUDIT_NATS_KEYS, AUDIT_PULSAR_KEYS, AUDIT_WEBHOOK_KEYS};
 use rustfs_ecstore::config::KVS;
 use rustfs_targets::{
     Target,
+    config::{
+        build_kafka_args, build_mqtt_args, build_nats_args, build_pulsar_args, build_webhook_args, validate_kafka_config,
+        validate_mqtt_config, validate_nats_config, validate_pulsar_config, validate_webhook_config,
+    },
     error::TargetError,
-    target::{mqtt::MQTTArgs, webhook::WebhookArgs},
+    target::TargetType,
 };
-use std::time::Duration;
-use tracing::{debug, warn};
-use url::Url;
+use std::collections::HashSet;
 
 /// Trait for creating targets from configuration
 #[async_trait]
@@ -45,10 +40,6 @@ pub trait TargetFactory: Send + Sync {
     /// Returns a set of valid configuration field names for this target type.
     /// This is used to filter environment variables.
     fn get_valid_fields(&self) -> HashSet<String>;
-
-    /// Returns a set of valid configuration env field names for this target type.
-    /// This is used to filter environment variables.
-    fn get_valid_env_fields(&self) -> HashSet<String>;
 }
 
 /// Factory for creating Webhook targets
@@ -57,71 +48,18 @@ pub struct WebhookTargetFactory;
 #[async_trait]
 impl TargetFactory for WebhookTargetFactory {
     async fn create_target(&self, id: String, config: &KVS) -> Result<Box<dyn Target<AuditEntry> + Send + Sync>, TargetError> {
-        // All config values are now read directly from the merged `config` KVS.
-        let endpoint = config
-            .lookup(WEBHOOK_ENDPOINT)
-            .ok_or_else(|| TargetError::Configuration("Missing webhook endpoint".to_string()))?;
-        let parsed_endpoint = endpoint.trim();
-        let endpoint_url = Url::parse(parsed_endpoint)
-            .map_err(|e| TargetError::Configuration(format!("Invalid endpoint URL: {e} (value: '{parsed_endpoint}')")))?;
-
-        let args = WebhookArgs {
-            enable: true, // If we are here, it's already enabled.
-            endpoint: endpoint_url,
-            auth_token: config.lookup(WEBHOOK_AUTH_TOKEN).unwrap_or_default(),
-            queue_dir: config.lookup(WEBHOOK_QUEUE_DIR).unwrap_or(AUDIT_DEFAULT_DIR.to_string()),
-            queue_limit: config
-                .lookup(WEBHOOK_QUEUE_LIMIT)
-                .and_then(|v| v.parse::<u64>().ok())
-                .unwrap_or(DEFAULT_LIMIT),
-            client_cert: config.lookup(WEBHOOK_CLIENT_CERT).unwrap_or_default(),
-            client_key: config.lookup(WEBHOOK_CLIENT_KEY).unwrap_or_default(),
-            client_ca: config.lookup(WEBHOOK_CLIENT_CA).unwrap_or_default(),
-            skip_tls_verify: config
-                .lookup(WEBHOOK_SKIP_TLS_VERIFY)
-                .and_then(|v| v.parse::<bool>().ok())
-                .unwrap_or(RUSTFS_WEBHOOK_SKIP_TLS_VERIFY_DEFAULT),
-            target_type: rustfs_targets::target::TargetType::AuditLog,
-        };
-
+        let args = build_webhook_args(config, AUDIT_DEFAULT_DIR, TargetType::AuditLog)?;
         let target = rustfs_targets::target::webhook::WebhookTarget::new(id, args)?;
         Ok(Box::new(target))
     }
 
     fn validate_config(&self, _id: &str, config: &KVS) -> Result<(), TargetError> {
-        // Validation also uses the merged `config` KVS directly.
-        let endpoint = config
-            .lookup(WEBHOOK_ENDPOINT)
-            .ok_or_else(|| TargetError::Configuration("Missing webhook endpoint".to_string()))?;
-        debug!("endpoint: {}", endpoint);
-        let parsed_endpoint = endpoint.trim();
-        Url::parse(parsed_endpoint)
-            .map_err(|e| TargetError::Configuration(format!("Invalid endpoint URL: {e} (value: '{parsed_endpoint}')")))?;
-
-        let client_cert = config.lookup(WEBHOOK_CLIENT_CERT).unwrap_or_default();
-        let client_key = config.lookup(WEBHOOK_CLIENT_KEY).unwrap_or_default();
-
-        if client_cert.is_empty() != client_key.is_empty() {
-            return Err(TargetError::Configuration(
-                "Both client_cert and client_key must be specified together".to_string(),
-            ));
-        }
-
-        let queue_dir = config.lookup(WEBHOOK_QUEUE_DIR).unwrap_or(AUDIT_DEFAULT_DIR.to_string());
-        if !queue_dir.is_empty() && !std::path::Path::new(&queue_dir).is_absolute() {
-            return Err(TargetError::Configuration("Webhook queue directory must be an absolute path".to_string()));
-        }
-
-        Ok(())
+        validate_webhook_config(config, AUDIT_DEFAULT_DIR)
     }
 
     fn get_valid_fields(&self) -> HashSet<String> {
         AUDIT_WEBHOOK_KEYS.iter().map(|s| s.to_string()).collect()
     }
-
-    fn get_valid_env_fields(&self) -> HashSet<String> {
-        ENV_AUDIT_WEBHOOK_KEYS.iter().map(|s| s.to_string()).collect()
-    }
 }
 
 /// Factory for creating MQTT targets
@@ -130,101 +68,73 @@ pub struct MQTTTargetFactory;
 #[async_trait]
 impl TargetFactory for MQTTTargetFactory {
     async fn create_target(&self, id: String, config: &KVS) -> Result<Box<dyn Target<AuditEntry> + Send + Sync>, TargetError> {
-        let broker = config
-            .lookup(MQTT_BROKER)
-            .ok_or_else(|| TargetError::Configuration("Missing MQTT broker".to_string()))?;
-        let broker_url = Url::parse(&broker)
-            .map_err(|e| TargetError::Configuration(format!("Invalid broker URL: {e} (value: '{broker}')")))?;
-
-        let topic = config
-            .lookup(MQTT_TOPIC)
-            .ok_or_else(|| TargetError::Configuration("Missing MQTT topic".to_string()))?;
-
-        let args = MQTTArgs {
-            enable: true, // Assumed enabled.
-            broker: broker_url,
-            topic,
-            qos: config
-                .lookup(MQTT_QOS)
-                .and_then(|v| v.parse::<u8>().ok())
-                .map(|q| match q {
-                    0 => QoS::AtMostOnce,
-                    1 => QoS::AtLeastOnce,
-                    2 => QoS::ExactlyOnce,
-                    _ => QoS::AtLeastOnce,
-                })
-                .unwrap_or(QoS::AtLeastOnce),
-            username: config.lookup(MQTT_USERNAME).unwrap_or_default(),
-            password: config.lookup(MQTT_PASSWORD).unwrap_or_default(),
-            max_reconnect_interval: config
-                .lookup(MQTT_RECONNECT_INTERVAL)
-                .and_then(|v| v.parse::<u64>().ok())
-                .map(Duration::from_secs)
-                .unwrap_or_else(|| Duration::from_secs(5)),
-            keep_alive: config
-                .lookup(MQTT_KEEP_ALIVE_INTERVAL)
-                .and_then(|v| v.parse::<u64>().ok())
-                .map(Duration::from_secs)
-                .unwrap_or_else(|| Duration::from_secs(30)),
-            queue_dir: config.lookup(MQTT_QUEUE_DIR).unwrap_or(AUDIT_DEFAULT_DIR.to_string()),
-            queue_limit: config
-                .lookup(MQTT_QUEUE_LIMIT)
-                .and_then(|v| v.parse::<u64>().ok())
-                .unwrap_or(DEFAULT_LIMIT),
-            target_type: rustfs_targets::target::TargetType::AuditLog,
-        };
-
+        let args = build_mqtt_args(config, AUDIT_DEFAULT_DIR, TargetType::AuditLog)?;
         let target = rustfs_targets::target::mqtt::MQTTTarget::new(id, args)?;
         Ok(Box::new(target))
     }
 
     fn validate_config(&self, _id: &str, config: &KVS) -> Result<(), TargetError> {
-        let broker = config
-            .lookup(MQTT_BROKER)
-            .ok_or_else(|| TargetError::Configuration("Missing MQTT broker".to_string()))?;
-        let url = Url::parse(&broker)
-            .map_err(|e| TargetError::Configuration(format!("Invalid broker URL: {e} (value: '{broker}')")))?;
-
-        match url.scheme() {
-            "tcp" | "ssl" | "ws" | "wss" | "mqtt" | "mqtts" => {}
-            _ => {
-                return Err(TargetError::Configuration("Unsupported broker URL scheme".to_string()));
-            }
-        }
-
-        if config.lookup(MQTT_TOPIC).is_none() {
-            return Err(TargetError::Configuration("Missing MQTT topic".to_string()));
-        }
-
-        if let Some(qos_str) = config.lookup(MQTT_QOS) {
-            let qos = qos_str
-                .parse::<u8>()
-                .map_err(|_| TargetError::Configuration("Invalid QoS value".to_string()))?;
-            if qos > 2 {
-                return Err(TargetError::Configuration("QoS must be 0, 1, or 2".to_string()));
-            }
-        }
-
-        let queue_dir = config.lookup(MQTT_QUEUE_DIR).unwrap_or_default();
-        if !queue_dir.is_empty() {
-            if !std::path::Path::new(&queue_dir).is_absolute() {
-                return Err(TargetError::Configuration("MQTT queue directory must be an absolute path".to_string()));
-            }
-            if let Some(qos_str) = config.lookup(MQTT_QOS)
-                && qos_str == "0"
-            {
-                warn!("Using queue_dir with QoS 0 may result in event loss");
-            }
-        }
-
-        Ok(())
+        validate_mqtt_config(config)
     }
 
     fn get_valid_fields(&self) -> HashSet<String> {
         AUDIT_MQTT_KEYS.iter().map(|s| s.to_string()).collect()
     }
+}
 
-    fn get_valid_env_fields(&self) -> HashSet<String> {
-        ENV_AUDIT_MQTT_KEYS.iter().map(|s| s.to_string()).collect()
+pub struct NATSTargetFactory;
+
+#[async_trait]
+impl TargetFactory for NATSTargetFactory {
+    async fn create_target(&self, id: String, config: &KVS) -> Result<Box<dyn Target<AuditEntry> + Send + Sync>, TargetError> {
+        let args = build_nats_args(config, AUDIT_DEFAULT_DIR, TargetType::AuditLog)?;
+        let target = rustfs_targets::target::nats::NATSTarget::new(id, args)?;
+        Ok(Box::new(target))
+    }
+
+    fn validate_config(&self, _id: &str, config: &KVS) -> Result<(), TargetError> {
+        validate_nats_config(config, AUDIT_DEFAULT_DIR)
+    }
+
+    fn get_valid_fields(&self) -> HashSet<String> {
+        AUDIT_NATS_KEYS.iter().map(|s| s.to_string()).collect()
+    }
+}
+
+pub struct PulsarTargetFactory;
+
+#[async_trait]
+impl TargetFactory for PulsarTargetFactory {
+    async fn create_target(&self, id: String, config: &KVS) -> Result<Box<dyn Target<AuditEntry> + Send + Sync>, TargetError> {
+        let args = build_pulsar_args(config, AUDIT_DEFAULT_DIR, TargetType::AuditLog)?;
+        let target = rustfs_targets::target::pulsar::PulsarTarget::new(id, args)?;
+        Ok(Box::new(target))
+    }
+
+    fn validate_config(&self, _id: &str, config: &KVS) -> Result<(), TargetError> {
+        validate_pulsar_config(config, AUDIT_DEFAULT_DIR)
+    }
+
+    fn get_valid_fields(&self) -> HashSet<String> {
+        AUDIT_PULSAR_KEYS.iter().map(|s| s.to_string()).collect()
+    }
+}
+
+pub struct KafkaTargetFactory;
+
+#[async_trait]
+impl TargetFactory for KafkaTargetFactory {
+    async fn create_target(&self, id: String, config: &KVS) -> Result<Box<dyn Target<AuditEntry> + Send + Sync>, TargetError> {
+        let args = build_kafka_args(config, AUDIT_DEFAULT_DIR, TargetType::AuditLog)?;
+        let target = rustfs_targets::target::kafka::KafkaTarget::new(id, args)?;
+        Ok(Box::new(target))
+    }
+
+    fn validate_config(&self, _id: &str, config: &KVS) -> Result<(), TargetError> {
+        validate_kafka_config(config, AUDIT_DEFAULT_DIR)
+    }
+
+    fn get_valid_fields(&self) -> HashSet<String> {
+        AUDIT_KAFKA_KEYS.iter().map(|s| s.to_string()).collect()
     }
 }

crates/audit/src/global.rs

@@ -12,7 +12,7 @@
 //  See the License for the specific language governing permissions and
 //  limitations under the License.
 
-use crate::{AuditEntry, AuditResult, AuditSystem};
+use crate::{AuditEntry, AuditResult, AuditSystem, system::AuditTargetMetricSnapshot};
 use rustfs_ecstore::config::Config;
 use std::sync::{Arc, OnceLock};
 use tracing::{debug, error, trace, warn};
@@ -89,6 +89,15 @@ pub async fn reload_audit_config(config: Config) -> AuditResult<()> {
     with_audit_system!(|system: Arc<AuditSystem>| async move { system.reload_config(config).await })
 }
 
+/// Returns per-target audit delivery metrics for Prometheus collection.
+pub async fn audit_target_metrics() -> Vec<AuditTargetMetricSnapshot> {
+    if let Some(system) = audit_system() {
+        system.snapshot_target_metrics().await
+    } else {
+        Vec::new()
+    }
+}
+
 /// Check if the global audit system is running
 pub async fn is_audit_system_running() -> bool {
     if let Some(system) = audit_system() {

crates/audit/src/lib.rs

@@ -31,4 +31,4 @@ pub use error::{AuditError, AuditResult};
 pub use global::*;
 pub use observability::{AuditMetrics, AuditMetricsReport, PerformanceValidation};
 pub use registry::AuditRegistry;
-pub use system::AuditSystem;
+pub use system::{AuditSystem, AuditTargetMetricSnapshot};

crates/audit/src/registry.rs

@@ -14,18 +14,19 @@
 
 use crate::{
     AuditEntry, AuditError, AuditResult,
-    factory::{MQTTTargetFactory, TargetFactory, WebhookTargetFactory},
+    factory::{
+        KafkaTargetFactory, MQTTTargetFactory, NATSTargetFactory, PulsarTargetFactory, TargetFactory, WebhookTargetFactory,
+    },
 };
 use futures::StreamExt;
 use futures::stream::FuturesUnordered;
-use hashbrown::{HashMap, HashSet};
-use rustfs_config::{DEFAULT_DELIMITER, ENABLE_KEY, ENV_PREFIX, EnableState, audit::AUDIT_ROUTE_PREFIX};
+use hashbrown::HashMap;
+use rustfs_config::audit::AUDIT_ROUTE_PREFIX;
 use rustfs_ecstore::config::{Config, KVS};
 use rustfs_targets::arn::TargetID;
-use rustfs_targets::{Target, TargetError, target::ChannelTargetType};
-use std::str::FromStr;
+use rustfs_targets::{Target, TargetError, config::collect_target_configs, target::ChannelTargetType};
 use std::sync::Arc;
-use tracing::{debug, error, info, warn};
+use tracing::{error, info};
 
 /// Registry for managing audit targets
 pub struct AuditRegistry {
@@ -52,6 +53,9 @@ impl AuditRegistry {
         // Register built-in factories
         registry.register(ChannelTargetType::Webhook.as_str(), Box::new(WebhookTargetFactory));
         registry.register(ChannelTargetType::Mqtt.as_str(), Box::new(MQTTTargetFactory));
+        registry.register(ChannelTargetType::Nats.as_str(), Box::new(NATSTargetFactory));
+        registry.register(ChannelTargetType::Pulsar.as_str(), Box::new(PulsarTargetFactory));
+        registry.register(ChannelTargetType::Kafka.as_str(), Box::new(KafkaTargetFactory));
 
         registry
     }
@@ -105,227 +109,31 @@ impl AuditRegistry {
         &self,
         config: &Config,
     ) -> AuditResult<Vec<Box<dyn Target<AuditEntry> + Send + Sync>>> {
-        // Collect only environment variables with the relevant prefix to reduce memory usage
-        let all_env: Vec<(String, String)> = std::env::vars().filter(|(key, _)| key.starts_with(ENV_PREFIX)).collect();
-        // A collection of asynchronous tasks for concurrently executing target creation
         let mut tasks = FuturesUnordered::new();
-        // let final_config = config.clone(); // Clone a configuration for aggregating the final result
-        // Record the defaults for each segment so that the segment can eventually be rebuilt
-        let mut section_defaults: HashMap<String, KVS> = HashMap::new();
-        // 1. Traverse all registered plants and process them by target type
         for (target_type, factory) in &self.factories {
             tracing::Span::current().record("target_type", target_type.as_str());
             info!("Start working on target types...");
-
-            // 2. Prepare the configuration source
-            // 2.1. Get the configuration segment in the file, e.g. 'audit_webhook'
-            let section_name = format!("{AUDIT_ROUTE_PREFIX}{target_type}").to_lowercase();
-            let file_configs = config.0.get(&section_name).cloned().unwrap_or_default();
-            // 2.2. Get the default configuration for that type
-            let default_cfg = file_configs.get(DEFAULT_DELIMITER).cloned().unwrap_or_default();
-            debug!(?default_cfg, "Get the default configuration");
-
-            // Save defaults for eventual write back
-            section_defaults.insert(section_name.clone(), default_cfg.clone());
-
-            // *** Optimization point 1: Get all legitimate fields of the current target type ***
             let valid_fields = factory.get_valid_fields();
-            debug!(?valid_fields, "Get the legitimate configuration fields");
-
-            // 3. Resolve instance IDs and configuration overrides from environment variables
-            let mut instance_ids_from_env = HashSet::new();
-            // 3.1. Instance discovery: Based on the '..._ENABLE_INSTANCEID' format
-            let enable_prefix =
-                format!("{ENV_PREFIX}{AUDIT_ROUTE_PREFIX}{target_type}{DEFAULT_DELIMITER}{ENABLE_KEY}{DEFAULT_DELIMITER}")
-                    .to_uppercase();
-            for (key, value) in &all_env {
-                if EnableState::from_str(value).ok().map(|s| s.is_enabled()).unwrap_or(false)
-                    && let Some(id) = key.strip_prefix(&enable_prefix)
-                    && !id.is_empty()
-                {
-                    instance_ids_from_env.insert(id.to_lowercase());
-                }
-            }
-
-            // 3.2. Parse all relevant environment variable configurations
-            // 3.2.1. Build environment variable prefixes such as 'RUSTFS_AUDIT_WEBHOOK_'
-            let env_prefix = format!("{ENV_PREFIX}{AUDIT_ROUTE_PREFIX}{target_type}{DEFAULT_DELIMITER}").to_uppercase();
-            // 3.2.2. 'env_overrides' is used to store configurations parsed from environment variables in the format: {instance id -> {field -> value}}
-            let mut env_overrides: HashMap<String, HashMap<String, String>> = HashMap::new();
-            for (key, value) in &all_env {
-                if let Some(rest) = key.strip_prefix(&env_prefix) {
-                    // Use rsplitn to split from the right side to properly extract the INSTANCE_ID at the end
-                    // Format: <FIELD_NAME>_<INSTANCE_ID> or <FIELD_NAME>
-                    let mut parts = rest.rsplitn(2, DEFAULT_DELIMITER);
-
-                    // The first part from the right is INSTANCE_ID
-                    let instance_id_part = parts.next().unwrap_or(DEFAULT_DELIMITER);
-                    // The remaining part is FIELD_NAME
-                    let field_name_part = parts.next();
-
-                    let (field_name, instance_id) = match field_name_part {
-                        // Case 1: The format is <FIELD_NAME>_<INSTANCE_ID>
-                        // e.g., rest = "ENDPOINT_PRIMARY" -> field_name="ENDPOINT", instance_id="PRIMARY"
-                        Some(field) => (field.to_lowercase(), instance_id_part.to_lowercase()),
-                        // Case 2: The format is <FIELD_NAME> (without INSTANCE_ID)
-                        // e.g., rest = "ENABLE" -> field_name="ENABLE", instance_id="" (Universal configuration `_ DEFAULT_DELIMITER`)
-                        None => (instance_id_part.to_lowercase(), DEFAULT_DELIMITER.to_string()),
-                    };
-
-                    // *** Optimization point 2: Verify whether the parsed field_name is legal ***
-                    if !field_name.is_empty() && valid_fields.contains(&field_name) {
-                        debug!(
-                            instance_id = %if instance_id.is_empty() { DEFAULT_DELIMITER } else { &instance_id },
-                            %field_name,
-                            %value,
-                            "Parsing to environment variables"
-                        );
-                        env_overrides
-                            .entry(instance_id)
-                            .or_default()
-                            .insert(field_name, value.clone());
-                    } else {
-                        // Ignore illegal field names
-                        warn!(
-                            field_name = %field_name,
-                            "Ignore environment variable fields, not found in the list of valid fields for target type {}",
-                            target_type
-                        );
-                    }
-                }
-            }
-            debug!(?env_overrides, "Complete the environment variable analysis");
-
-            // 4. Determine all instance IDs that need to be processed
-            let mut all_instance_ids: HashSet<String> =
-                file_configs.keys().filter(|k| *k != DEFAULT_DELIMITER).cloned().collect();
-            all_instance_ids.extend(instance_ids_from_env);
-            debug!(?all_instance_ids, "Determine all instance IDs");
-
-            // 5. Merge configurations and create tasks for each instance
-            for id in all_instance_ids {
-                // 5.1. Merge configuration, priority: Environment variables > File instance configuration > File default configuration
-                let mut merged_config = default_cfg.clone();
-                // Instance-specific configuration in application files
-                if let Some(file_instance_cfg) = file_configs.get(&id) {
-                    merged_config.extend(file_instance_cfg.clone());
-                }
-                // Application instance-specific environment variable configuration
-                if let Some(env_instance_cfg) = env_overrides.get(&id) {
-                    // Convert HashMap<String, String> to KVS
-                    let mut kvs_from_env = KVS::new();
-                    for (k, v) in env_instance_cfg {
-                        kvs_from_env.insert(k.clone(), v.clone());
-                    }
-                    merged_config.extend(kvs_from_env);
-                }
-                debug!(instance_id = %id, ?merged_config, "Complete configuration merge");
-
-                // 5.2. Check if the instance is enabled
-                let enabled = merged_config
-                    .lookup(ENABLE_KEY)
-                    .map(|v| {
-                        EnableState::from_str(v.as_str())
-                            .ok()
-                            .map(|s| s.is_enabled())
-                            .unwrap_or(false)
-                    })
-                    .unwrap_or(false);
-
-                if enabled {
-                    info!(instance_id = %id, "Target is enabled, ready to create a task");
-                    // 5.3. Create asynchronous tasks for enabled instances
-                    let target_type_clone = target_type.clone();
-                    let tid = id.clone();
-                    let merged_config_arc = Arc::new(merged_config);
-                    tasks.push(async move {
-                        let result = factory.create_target(tid.clone(), &merged_config_arc).await;
-                        (target_type_clone, tid, result, Arc::clone(&merged_config_arc))
-                    });
-                } else {
-                    info!(instance_id = %id, "Skip the disabled target and will be removed from the final configuration");
-                    // Remove disabled target from final configuration
-                    // final_config.0.entry(section_name.clone()).or_default().remove(&id);
-                }
+            for (id, merged_config) in collect_target_configs(config, AUDIT_ROUTE_PREFIX, target_type, &valid_fields) {
+                info!(instance_id = %id, "Target is enabled, ready to create a task");
+                let tid = id.clone();
+                let merged_config_arc = Arc::new(merged_config);
+                tasks.push(async move {
+                    let result = factory.create_target(tid.clone(), &merged_config_arc).await;
+                    (tid, result)
+                });
             }
         }
 
-        // 6. Concurrently execute all creation tasks and collect results
         let mut successful_targets = Vec::new();
-        let mut successful_configs = Vec::new();
-        while let Some((target_type, id, result, final_config)) = tasks.next().await {
+        while let Some((id, result)) = tasks.next().await {
             match result {
                 Ok(target) => {
-                    info!(target_type = %target_type, instance_id = %id, "Create a target successfully");
+                    info!(target_type = %target.id().name, instance_id = %id, "Create a target successfully");
                     successful_targets.push(target);
-                    successful_configs.push((target_type, id, final_config));
                 }
                 Err(e) => {
-                    error!(target_type = %target_type, instance_id = %id, error = %e, "Failed to create a target");
-                }
-            }
-        }
-
-        // 7. Aggregate new configuration and write back to system configuration
-        if !successful_configs.is_empty() || !section_defaults.is_empty() {
-            info!(
-                "Prepare to update {} successfully created target configurations to the system configuration...",
-                successful_configs.len()
-            );
-
-            let mut successes_by_section: HashMap<String, HashMap<String, KVS>> = HashMap::new();
-
-            for (target_type, id, kvs) in successful_configs {
-                let section_name = format!("{AUDIT_ROUTE_PREFIX}{target_type}").to_lowercase();
-                successes_by_section
-                    .entry(section_name)
-                    .or_default()
-                    .insert(id.to_lowercase(), (*kvs).clone());
-            }
-
-            let mut new_config = config.clone();
-            // Collection of segments that need to be processed: Collect all segments where default items exist or where successful instances exist
-            let mut sections: HashSet<String> = HashSet::new();
-            sections.extend(section_defaults.keys().cloned());
-            sections.extend(successes_by_section.keys().cloned());
-
-            for section in sections {
-                let mut section_map: std::collections::HashMap<String, KVS> = std::collections::HashMap::new();
-                // Add default item
-                if let Some(default_kvs) = section_defaults.get(&section)
-                    && !default_kvs.is_empty()
-                {
-                    section_map.insert(DEFAULT_DELIMITER.to_string(), default_kvs.clone());
-                }
-
-                // Add successful instance item
-                if let Some(instances) = successes_by_section.get(&section) {
-                    for (id, kvs) in instances {
-                        section_map.insert(id.clone(), kvs.clone());
-                    }
-                }
-
-                // Empty breaks are removed and non-empty breaks are replaced entirely.
-                if section_map.is_empty() {
-                    new_config.0.remove(&section);
-                } else {
-                    new_config.0.insert(section, section_map);
-                }
-            }
-
-            let Some(store) = rustfs_ecstore::global::new_object_layer_fn() else {
-                return Err(AuditError::StorageNotAvailable(
-                    "Failed to save target configuration: server storage not initialized".to_string(),
-                ));
-            };
-
-            match rustfs_ecstore::config::com::save_server_config(store, &new_config).await {
-                Ok(_) => {
-                    info!("The new configuration was saved to the system successfully.")
-                }
-                Err(e) => {
-                    error!("Failed to save the new configuration: {}", e);
-                    return Err(AuditError::SaveConfig(Box::new(e)));
+                    error!(instance_id = %id, error = %e, "Failed to create a target");
                 }
             }
         }
@@ -365,6 +173,11 @@ impl AuditRegistry {
         self.targets.get(id).map(|t| t.as_ref())
     }
 
+    /// Lists cloned target values for runtime inspection without exposing mutable registry access.
+    pub fn list_target_values(&self) -> Vec<Box<dyn Target<AuditEntry> + Send + Sync>> {
+        self.targets.values().map(|target| target.clone_dyn()).collect()
+    }
+
     /// Lists all target IDs
     ///
     /// # Returns
@@ -387,8 +200,8 @@ impl AuditRegistry {
             }
         }
 
-        if !errors.is_empty() {
-            return Err(AuditError::Target(errors.into_iter().next().unwrap()));
+        if let Some(error) = errors.into_iter().next() {
+            return Err(AuditError::Target(error));
         }
 
         Ok(())

crates/audit/src/system.rs

@@ -13,16 +13,25 @@
 //  limitations under the License.
 
 use crate::{AuditEntry, AuditError, AuditRegistry, AuditResult, observability};
+use hashbrown::HashMap;
 use rustfs_ecstore::config::Config;
 use rustfs_targets::{
     StoreError, Target, TargetError,
     store::{Key, Store},
-    target::EntityTarget,
+    target::{EntityTarget, QueuedPayload},
 };
 use std::sync::Arc;
-use tokio::sync::{Mutex, RwLock};
+use tokio::sync::{Mutex, RwLock, mpsc};
 use tracing::{error, info, warn};
 
+#[derive(Debug, Clone, Default, PartialEq, Eq)]
+pub struct AuditTargetMetricSnapshot {
+    pub failed_messages: u64,
+    pub queue_length: u64,
+    pub target_id: String,
+    pub total_messages: u64,
+}
+
 /// State of the audit system
 #[derive(Debug, Clone, PartialEq, Eq)]
 pub enum AuditSystemState {
@@ -39,6 +48,8 @@ pub struct AuditSystem {
     registry: Arc<Mutex<AuditRegistry>>,
     state: Arc<RwLock<AuditSystemState>>,
     config: Arc<RwLock<Option<Config>>>,
+    /// Cancellation senders for active audit stream tasks (target_id -> cancel tx)
+    stream_cancellers: Arc<RwLock<HashMap<String, mpsc::Sender<()>>>>,
 }
 
 impl Default for AuditSystem {
@@ -54,6 +65,7 @@ impl AuditSystem {
             registry: Arc::new(Mutex::new(AuditRegistry::new())),
             state: Arc::new(RwLock::new(AuditSystemState::Stopped)),
             config: Arc::new(RwLock::new(None)),
+            stream_cancellers: Arc::new(RwLock::new(HashMap::new())),
         }
     }
 
@@ -110,27 +122,7 @@ impl AuditSystem {
 
                 // Initialize all targets
                 for target in targets {
-                    let target_id = target.id().to_string();
-                    if let Err(e) = target.init().await {
-                        error!(target_id = %target_id, error = %e, "Failed to initialize audit target");
-                    } else {
-                        // After successful initialization, if enabled and there is a store, start the send from storage task
-                        if target.is_enabled() {
-                            if let Some(store) = target.store() {
-                                info!(target_id = %target_id, "Start audit stream processing for target");
-                                let store_clone: Box<dyn Store<EntityTarget<AuditEntry>, Error = StoreError, Key = Key> + Send> =
-                                    store.boxed_clone();
-                                let target_arc: Arc<dyn Target<AuditEntry> + Send + Sync> = Arc::from(target.clone_dyn());
-                                self.start_audit_stream_with_batching(store_clone, target_arc);
-                                info!(target_id = %target_id, "Audit stream processing started");
-                            } else {
-                                info!(target_id = %target_id, "No store configured, skip audit stream processing");
-                            }
-                        } else {
-                            info!(target_id = %target_id, "Target disabled, skip audit stream processing");
-                        }
-                        registry.add_target(target_id, target);
-                    }
+                    self.init_and_register_target(target, &mut registry).await;
                 }
 
                 // Update state to running
@@ -214,6 +206,9 @@ impl AuditSystem {
 
         info!("Stopping audit system");
 
+        // Stop all stream tasks first
+        self.stop_all_streams().await;
+
         // Close all targets
         let mut registry = self.registry.lock().await;
         if let Err(e) = registry.close_all().await {
@@ -258,52 +253,49 @@ impl AuditSystem {
         let state = self.state.read().await;
 
         match *state {
-            AuditSystemState::Running => {
-                // Continue with dispatch
-                info!("Dispatching audit log entry");
-            }
+            AuditSystemState::Running => {}
             AuditSystemState::Paused => {
-                // Skip dispatch when paused
                 return Ok(());
             }
             _ => {
-                // Don't dispatch when not running
                 return Err(AuditError::NotInitialized("Audit system is not running".to_string()));
             }
         }
         drop(state);
 
-        let registry = self.registry.lock().await;
-        let target_keys = registry.list_targets();
+        // Collect cloned targets under lock, then dispatch without holding it
+        let targets: Vec<(String, Box<dyn Target<AuditEntry> + Send + Sync>)> = {
+            let registry = self.registry.lock().await;
+            let target_keys = registry.list_targets();
 
-        if target_keys.is_empty() {
-            warn!("No audit targets configured for dispatch");
-            return Ok(());
-        }
+            if target_keys.is_empty() {
+                warn!("No audit targets configured for dispatch");
+                return Ok(());
+            }
 
-        // Dispatch to all targets concurrently
+            target_keys
+                .into_iter()
+                .filter_map(|key| registry.get_target(&key).map(|t| (key, t.clone_dyn())))
+                .collect()
+        };
+
+        // Dispatch to all targets concurrently (no lock held)
         let mut tasks = Vec::new();
 
-        for target_key in target_keys {
-            if let Some(target) = registry.get_target(&target_key) {
-                let entry_clone = Arc::clone(&entry);
-                let target_key_clone = target_key.clone();
+        for (target_key, target) in targets {
+            let entity_target = EntityTarget {
+                object_name: entry.api.name.clone().unwrap_or_default(),
+                bucket_name: entry.api.bucket.clone().unwrap_or_default(),
+                event_name: entry.event,
+                data: (*entry).clone(),
+            };
 
-                // Create EntityTarget for the audit log entry
-                let entity_target = EntityTarget {
-                    object_name: entry.api.name.clone().unwrap_or_default(),
-                    bucket_name: entry.api.bucket.clone().unwrap_or_default(),
-                    event_name: entry.event, // Default, should be derived from entry
-                    data: (*entry_clone).clone(),
-                };
+            let task = async move {
+                let result = target.save(Arc::new(entity_target)).await;
+                (target_key, result)
+            };
 
-                let task = async move {
-                    let result = target.save(Arc::new(entity_target)).await;
-                    (target_key_clone, result)
-                };
-
-                tasks.push(task);
-            }
+            tasks.push(task);
         }
 
         // Execute all dispatch tasks
@@ -359,39 +351,45 @@ impl AuditSystem {
         }
         drop(state);
 
-        let registry = self.registry.lock().await;
-        let target_keys = registry.list_targets();
+        // Collect targets under lock, then dispatch without holding it
+        let targets: Vec<(String, Box<dyn Target<AuditEntry> + Send + Sync>)> = {
+            let registry = self.registry.lock().await;
+            let target_keys = registry.list_targets();
 
-        if target_keys.is_empty() {
-            warn!("No audit targets configured for batch dispatch");
-            return Ok(());
-        }
+            if target_keys.is_empty() {
+                warn!("No audit targets configured for batch dispatch");
+                return Ok(());
+            }
+
+            target_keys
+                .into_iter()
+                .filter_map(|key| registry.get_target(&key).map(|t| (key, t.clone_dyn())))
+                .collect()
+        };
 
         let mut tasks = Vec::new();
-        for target_key in target_keys {
-            if let Some(target) = registry.get_target(&target_key) {
-                let entries_clone: Vec<_> = entries.iter().map(Arc::clone).collect();
-                let target_key_clone = target_key.clone();
+        for (target_key, target) in targets {
+            let entries_clone: Vec<_> = entries.iter().map(Arc::clone).collect();
+            let target_key_clone = target_key.clone();
 
-                let task = async move {
-                    let mut success_count = 0;
-                    let mut errors = Vec::new();
-                    for entry in entries_clone {
-                        let entity_target = EntityTarget {
-                            object_name: entry.api.name.clone().unwrap_or_default(),
-                            bucket_name: entry.api.bucket.clone().unwrap_or_default(),
-                            event_name: entry.event,
-                            data: (*entry).clone(),
-                        };
-                        match target.save(Arc::new(entity_target)).await {
-                            Ok(_) => success_count += 1,
-                            Err(e) => errors.push(e),
-                        }
+            let task = async move {
+                let mut success_count = 0;
+                let mut errors = Vec::new();
+                for entry in entries_clone {
+                    let entity_target = EntityTarget {
+                        object_name: entry.api.name.clone().unwrap_or_default(),
+                        bucket_name: entry.api.bucket.clone().unwrap_or_default(),
+                        event_name: entry.event,
+                        data: (*entry).clone(),
+                    };
+                    match target.save(Arc::new(entity_target)).await {
+                        Ok(_) => success_count += 1,
+                        Err(e) => errors.push(e),
                     }
-                    (target_key_clone, success_count, errors)
-                };
-                tasks.push(task);
-            }
+                }
+                (target_key_clone, success_count, errors)
+            };
+            tasks.push(task);
         }
 
         let results = futures::future::join_all(tasks).await;
@@ -417,6 +415,60 @@ impl AuditSystem {
         Ok(())
     }
 
+    /// Stops all active audit stream tasks by sending cancellation signals.
+    async fn stop_all_streams(&self) {
+        let mut cancellers = self.stream_cancellers.write().await;
+        for (target_id, cancel_tx) in cancellers.drain() {
+            info!(target_id = %target_id, "Stopping audit stream");
+            let _ = cancel_tx.send(()).await;
+        }
+    }
+
+    /// Initializes a single target: runs init(), starts stream if store is present,
+    /// and adds it to the registry. For store-backed targets, registration and stream
+    /// startup proceed even if init() fails so queued entries can be drained later.
+    async fn init_and_register_target(
+        &self,
+        target: Box<dyn Target<AuditEntry> + Send + Sync>,
+        registry: &mut AuditRegistry,
+    ) -> Option<String> {
+        let target_id = target.id().to_string();
+        let has_store = target.store().is_some();
+
+        if let Err(e) = target.init().await {
+            error!(target_id = %target_id, error = %e, "Failed to initialize audit target");
+            // Non-store targets: init failure is fatal.
+            if !has_store {
+                return None;
+            }
+            // Store-backed targets: still register and start the stream so queued
+            // entries can be drained when connectivity recovers.
+            warn!(
+                target_id = %target_id,
+                "Proceeding with store-backed audit target despite init failure"
+            );
+        }
+
+        if target.is_enabled() {
+            if let Some(store) = target.store() {
+                info!(target_id = %target_id, "Start audit stream processing for target");
+                let store_clone: Box<dyn Store<QueuedPayload, Error = StoreError, Key = Key> + Send> = store.boxed_clone();
+                let target_arc: Arc<dyn Target<AuditEntry> + Send + Sync> = Arc::from(target.clone_dyn());
+                let cancel_tx = self.start_audit_stream_with_batching(store_clone, target_arc);
+
+                self.stream_cancellers.write().await.insert(target_id.clone(), cancel_tx);
+                info!(target_id = %target_id, "Audit stream processing started");
+            } else {
+                info!(target_id = %target_id, "No store configured, skip audit stream processing");
+            }
+        } else {
+            info!(target_id = %target_id, "Target disabled, skip audit stream processing");
+        }
+
+        registry.add_target(target_id.clone(), target);
+        Some(target_id)
+    }
+
     /// Starts the audit stream processing for a target with batching and retry logic
     ///
     /// # Arguments
@@ -427,9 +479,10 @@ impl AuditSystem {
     /// and attempts to send them to the specified target. It implements retry logic with exponential backoff
     fn start_audit_stream_with_batching(
         &self,
-        store: Box<dyn Store<EntityTarget<AuditEntry>, Error = StoreError, Key = Key> + Send>,
+        store: Box<dyn Store<QueuedPayload, Error = StoreError, Key = Key> + Send>,
         target: Arc<dyn Target<AuditEntry> + Send + Sync>,
-    ) {
+    ) -> mpsc::Sender<()> {
+        let (cancel_tx, mut cancel_rx) = mpsc::channel(1);
         let state = self.state.clone();
 
         tokio::spawn(async move {
@@ -442,6 +495,12 @@ impl AuditSystem {
             const BASE_RETRY_DELAY: Duration = Duration::from_secs(2);
 
             loop {
+                // Check for cancellation signal
+                if cancel_rx.try_recv().is_ok() {
+                    info!("Audit stream cancelled for target: {}", target.id());
+                    break;
+                }
+
                 match *state.read().await {
                     AuditSystemState::Running | AuditSystemState::Paused | AuditSystemState::Starting => {}
                     _ => {
@@ -452,11 +511,22 @@ impl AuditSystem {
 
                 let keys: Vec<Key> = store.list();
                 if keys.is_empty() {
-                    sleep(Duration::from_millis(500)).await;
+                    tokio::select! {
+                        _ = sleep(Duration::from_millis(500)) => {},
+                        _ = cancel_rx.recv() => {
+                            info!("Audit stream cancelled during idle for target: {}", target.id());
+                            return;
+                        }
+                    }
                     continue;
                 }
 
                 for key in keys {
+                    if cancel_rx.try_recv().is_ok() {
+                        info!("Audit stream cancelled during processing for target: {}", target.id());
+                        return;
+                    }
+
                     let mut retries = 0usize;
                     let mut success = false;
 
@@ -475,8 +545,14 @@ impl AuditSystem {
                                     TargetError::Timeout(_) => {
                                         warn!("Timeout sending to target {}, retrying...", target.id());
                                     }
+                                    TargetError::Dropped(reason) => {
+                                        warn!("Dropped queued payload for target {}: {}", target.id(), reason);
+                                        observability::record_target_failure();
+                                        break;
+                                    }
                                     _ => {
                                         error!("Permanent error for target {}: {}", target.id(), e);
+                                        target.record_final_failure();
                                         observability::record_target_failure();
                                         break;
                                     }
@@ -490,6 +566,7 @@ impl AuditSystem {
 
                     if retries >= MAX_RETRIES && !success {
                         warn!("Max retries exceeded for key {}, target: {}, skipping", key.to_string(), target.id());
+                        target.record_final_failure();
                         observability::record_target_failure();
                     }
                 }
@@ -497,6 +574,8 @@ impl AuditSystem {
                 sleep(Duration::from_millis(100)).await;
             }
         });
+
+        cancel_tx
     }
 
     /// Enables a specific target
@@ -594,6 +673,31 @@ impl AuditSystem {
         registry.list_targets()
     }
 
+    /// Returns cloned target values for read-only runtime inspection.
+    pub async fn get_target_values(&self) -> Vec<Box<dyn Target<AuditEntry> + Send + Sync>> {
+        let registry = self.registry.lock().await;
+        registry.list_target_values()
+    }
+
+    /// Returns per-target delivery metrics for Prometheus collection.
+    pub async fn snapshot_target_metrics(&self) -> Vec<AuditTargetMetricSnapshot> {
+        let targets = self.get_target_values().await;
+        let mut snapshots = Vec::with_capacity(targets.len());
+
+        for target in targets {
+            let delivery = target.delivery_snapshot();
+            snapshots.push(AuditTargetMetricSnapshot {
+                failed_messages: delivery.failed_messages,
+                queue_length: delivery.queue_length,
+                target_id: target.id().to_string(),
+                total_messages: delivery.total_messages,
+            });
+        }
+
+        snapshots.sort_by(|a, b| a.target_id.cmp(&b.target_id));
+        snapshots
+    }
+
     /// Gets information about a specific target
     ///
     /// # Arguments
@@ -616,9 +720,11 @@ impl AuditSystem {
     pub async fn reload_config(&self, new_config: Config) -> AuditResult<()> {
         info!("Reloading audit system configuration");
 
-        // Record config reload
         observability::record_config_reload();
 
+        // Stop all existing stream tasks first
+        self.stop_all_streams().await;
+
         // Store new configuration
         {
             let mut config_guard = self.config.write().await;
@@ -636,28 +742,9 @@ impl AuditSystem {
             Ok(targets) => {
                 info!(target_count = targets.len(), "Reloaded audit targets successfully");
 
-                // Initialize all new targets
                 for target in targets {
-                    let target_id = target.id().to_string();
-                    if let Err(e) = target.init().await {
-                        error!(target_id = %target_id, error = %e, "Failed to initialize reloaded audit target");
-                    } else {
-                        // Same starts the storage stream after a heavy load
-                        if target.is_enabled() {
-                            if let Some(store) = target.store() {
-                                info!(target_id = %target_id, "Start audit stream processing for target (reload)");
-                                let store_clone: Box<dyn Store<EntityTarget<AuditEntry>, Error = StoreError, Key = Key> + Send> =
-                                    store.boxed_clone();
-                                let target_arc: Arc<dyn Target<AuditEntry> + Send + Sync> = Arc::from(target.clone_dyn());
-                                self.start_audit_stream_with_batching(store_clone, target_arc);
-                                info!(target_id = %target_id, "Audit stream processing started (reload)");
-                            } else {
-                                info!(target_id = %target_id, "No store configured, skip audit stream processing (reload)");
-                            }
-                        } else {
-                            info!(target_id = %target_id, "Target disabled, skip audit stream processing (reload)");
-                        }
-                        registry.add_target(target.id().to_string(), target);
+                    if let Some(target_id) = self.init_and_register_target(target, &mut registry).await {
+                        info!(target_id = %target_id, "Target initialized (reload)");
                     }
                 }
 

crates/audit/tests/config_parsing_test.rs

@@ -53,6 +53,12 @@ fn test_mqtt_valid_fields() {
         "reconnect_interval",
         "queue_dir",
         "queue_limit",
+        "tls_policy",
+        "tls_ca",
+        "tls_client_cert",
+        "tls_client_key",
+        "tls_trust_leaf_as_ca",
+        "ws_path_allowlist",
     ];
 
     // This tests the MQTT configuration fields we support

crates/audit/tests/integration_test.rs

@@ -15,6 +15,7 @@
 use rustfs_audit::*;
 use rustfs_ecstore::config::{Config, KVS};
 use std::collections::HashMap;
+use temp_env::with_vars;
 
 #[tokio::test]
 async fn test_audit_system_creation() {
@@ -35,34 +36,42 @@ async fn test_config_parsing_webhook() {
     let mut config = Config(HashMap::new());
     let mut audit_webhook_section = HashMap::new();
 
-    // Create default configuration
     let mut default_kvs = KVS::new();
-    default_kvs.insert("enable".to_string(), "on".to_string());
-    default_kvs.insert("endpoint".to_string(), "http://localhost:3020/webhook".to_string());
-
+    default_kvs.insert("enable".to_string(), "off".to_string());
+    default_kvs.insert("endpoint".to_string(), "".to_string());
     audit_webhook_section.insert("_".to_string(), default_kvs);
+    let mut instance_kvs = KVS::new();
+    instance_kvs.insert("enable".to_string(), "on".to_string());
+    instance_kvs.insert("endpoint".to_string(), "http://localhost:3020/webhook".to_string());
+    audit_webhook_section.insert("primary".to_string(), instance_kvs);
     config.0.insert("audit_webhook".to_string(), audit_webhook_section);
 
     let registry = AuditRegistry::new();
 
-    // This should not fail even if server storage is not initialized
-    // as it's an integration test
     let result = registry.create_audit_targets_from_config(&config).await;
+    assert!(result.is_ok(), "audit target creation should not require server storage");
+}
 
-    // We expect this to fail due to server storage not being initialized
-    // but the parsing should work correctly
-    match result {
-        Err(AuditError::StorageNotAvailable(_)) => {
-            // This is expected in test environment
-        }
-        Err(e) => {
-            // Other errors might indicate parsing issues
-            println!("Unexpected error: {e}");
-        }
-        Ok(_) => {
-            // Unexpected success in test environment without server storage
-        }
-    }
+#[test]
+fn test_env_only_audit_target_does_not_require_server_storage() {
+    with_vars(
+        [
+            ("RUSTFS_AUDIT_WEBHOOK_ENABLE_PRIMARY", Some("on")),
+            ("RUSTFS_AUDIT_WEBHOOK_ENDPOINT_PRIMARY", Some("http://localhost:3020/webhook")),
+        ],
+        || {
+            let runtime = tokio::runtime::Builder::new_current_thread()
+                .enable_all()
+                .build()
+                .expect("failed to create tokio runtime");
+            runtime.block_on(async {
+                let config = Config(HashMap::new());
+                let registry = AuditRegistry::new();
+                let result = registry.create_audit_targets_from_config(&config).await;
+                assert!(result.is_ok(), "env-only audit target creation should not require server storage");
+            });
+        },
+    )
 }
 
 #[test]

crates/audit/tests/performance_test.rs

@@ -159,7 +159,8 @@ async fn test_audit_log_dispatch_performance() {
 
     let start = Instant::now();
 
-    // Dispatch audit log (should be fast since no targets are configured)
+    // Dispatch audit log against an unstarted system state. Empty config keeps
+    // the audit system stopped, so dispatch should fail fast without targets.
     let result = system.dispatch(Arc::new(audit_entry)).await;
     let elapsed = start.elapsed();
 
@@ -168,8 +169,10 @@ async fn test_audit_log_dispatch_performance() {
     // Should be very fast (sub-millisecond for no targets)
     assert!(elapsed < Duration::from_millis(100), "Dispatch took too long: {elapsed:?}");
 
-    // Should succeed even with no targets
-    assert!(result.is_ok(), "Dispatch should succeed with no targets");
+    assert!(
+        matches!(result, Err(AuditError::NotInitialized(_))),
+        "Dispatch on a stopped system should return NotInitialized, got: {result:?}"
+    );
 
     // Clean up
     let _ = system.close().await;
@@ -186,11 +189,11 @@ async fn test_system_state_transitions() {
     let config = rustfs_ecstore::config::Config(std::collections::HashMap::new());
     let start_result = system.start(config).await;
 
-    // Should be running (or failed due to server storage)
+    // Empty config keeps the audit system stopped even when start() succeeds.
     let state = system.get_state().await;
     match start_result {
         Ok(_) => {
-            assert_eq!(state, rustfs_audit::system::AuditSystemState::Running);
+            assert_eq!(state, rustfs_audit::system::AuditSystemState::Stopped);
         }
         Err(_) => {
             // Expected in test environment due to server storage not being initialized

crates/audit/tests/system_integration_test.rs

@@ -29,27 +29,21 @@ async fn test_complete_audit_system_lifecycle() {
     assert_eq!(system.get_state().await, system::AuditSystemState::Stopped);
     assert!(!system.is_running().await);
 
-    // 2. Start with empty config (will fail due to no server storage in test)
+    // 2. Start with empty config. The current implementation returns Ok(())
+    // but keeps the system stopped when no audit targets are enabled.
     let config = Config(HashMap::new());
     let start_result = system.start(config).await;
 
-    // Should fail in test environment but state handling should work
+    // State handling should remain consistent for both empty-config success and
+    // storage-unavailable failure paths.
     match start_result {
         Err(AuditError::StorageNotAvailable(_)) => {
             // Expected in test environment
             assert_eq!(system.get_state().await, system::AuditSystemState::Stopped);
         }
         Ok(_) => {
-            // If it somehow succeeds, verify running state
-            assert_eq!(system.get_state().await, system::AuditSystemState::Running);
-            assert!(system.is_running().await);
-
-            // Test pause/resume
-            system.pause().await.expect("Should pause successfully");
-            assert_eq!(system.get_state().await, system::AuditSystemState::Paused);
-
-            system.resume().await.expect("Should resume successfully");
-            assert_eq!(system.get_state().await, system::AuditSystemState::Running);
+            assert_eq!(system.get_state().await, system::AuditSystemState::Stopped);
+            assert!(!system.is_running().await);
         }
         Err(e) => {
             panic!("Unexpected error: {e}");

crates/checksums/Cargo.toml

@@ -36,3 +36,6 @@ sha2 = { workspace = true }
 
 [dev-dependencies]
 pretty_assertions = { workspace = true }
+
+[lib]
+doctest = false

crates/checksums/src/lib.rs

@@ -349,7 +349,7 @@ mod tests {
     fn base64_encoded_checksum_to_hex_string(header_value: &HeaderValue) -> String {
         let decoded_checksum = base64::decode(header_value.to_str().unwrap()).unwrap();
         let decoded_checksum = decoded_checksum.into_iter().fold(String::new(), |mut acc, byte| {
-            write!(acc, "{byte:02X?}").expect("string will always be writeable");
+            write!(acc, "{byte:02X?}").expect("string will always be writable");
             acc
         });
 

crates/common/Cargo.toml

@@ -41,3 +41,6 @@ rmp-serde = { workspace = true }
 async-trait = { workspace = true }
 s3s = { workspace = true }
 tracing = { workspace = true }
+
+[lib]
+doctest = false

crates/common/src/capacity_scope.rs

@@ -0,0 +1,308 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::collections::{HashMap, HashSet};
+use std::sync::{Mutex, OnceLock};
+use std::time::{Duration, Instant};
+use uuid::Uuid;
+
+const CAPACITY_SCOPE_REGISTRY_SOFT_LIMIT: usize = 2_048;
+const CAPACITY_SCOPE_REGISTRY_HARD_LIMIT: usize = 4_096;
+const CAPACITY_SCOPE_TTL: Duration = Duration::from_secs(300);
+
+#[derive(Debug, Clone, PartialEq, Eq, Hash)]
+pub struct CapacityScopeDisk {
+    pub endpoint: String,
+    pub drive_path: String,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Default)]
+pub struct CapacityScope {
+    pub disks: Vec<CapacityScopeDisk>,
+}
+
+#[derive(Debug, Clone)]
+struct CapacityScopeEntry {
+    scope: CapacityScope,
+    recorded_at: Instant,
+}
+
+fn capacity_scope_registry() -> &'static Mutex<HashMap<Uuid, CapacityScopeEntry>> {
+    static REGISTRY: OnceLock<Mutex<HashMap<Uuid, CapacityScopeEntry>>> = OnceLock::new();
+    REGISTRY.get_or_init(|| Mutex::new(HashMap::new()))
+}
+
+fn global_dirty_scope_registry() -> &'static Mutex<HashSet<CapacityScopeDisk>> {
+    static REGISTRY: OnceLock<Mutex<HashSet<CapacityScopeDisk>>> = OnceLock::new();
+    REGISTRY.get_or_init(|| Mutex::new(HashSet::new()))
+}
+
+fn prune_expired_entries(entries: &mut HashMap<Uuid, CapacityScopeEntry>, now: Instant) {
+    entries.retain(|_, entry| now.duration_since(entry.recorded_at) <= CAPACITY_SCOPE_TTL);
+}
+
+fn enforce_hard_limit(entries: &mut HashMap<Uuid, CapacityScopeEntry>, max_len: usize) {
+    if entries.len() < max_len {
+        return;
+    }
+
+    let evict_count = entries.len() - max_len + 1;
+    let mut eviction_order: Vec<_> = entries.iter().map(|(token, entry)| (*token, entry.recorded_at)).collect();
+    eviction_order.sort_unstable_by_key(|(_, recorded_at)| *recorded_at);
+
+    for (token, _) in eviction_order.into_iter().take(evict_count) {
+        entries.remove(&token);
+    }
+}
+
+fn merge_capacity_scopes(existing: &mut CapacityScope, incoming: CapacityScope) {
+    let mut seen: HashSet<CapacityScopeDisk> = existing.disks.iter().cloned().collect();
+    for disk in incoming.disks {
+        if seen.insert(disk.clone()) {
+            existing.disks.push(disk);
+        }
+    }
+}
+
+pub fn record_capacity_scope(token: Uuid, scope: CapacityScope) {
+    let now = Instant::now();
+    let mut entries = capacity_scope_registry().lock().unwrap_or_else(|p| p.into_inner());
+    if !entries.contains_key(&token) && entries.len() >= CAPACITY_SCOPE_REGISTRY_SOFT_LIMIT {
+        prune_expired_entries(&mut entries, now);
+        enforce_hard_limit(&mut entries, CAPACITY_SCOPE_REGISTRY_HARD_LIMIT);
+    }
+    if let Some(entry) = entries.get_mut(&token) {
+        merge_capacity_scopes(&mut entry.scope, scope);
+        entry.recorded_at = now;
+    } else {
+        entries.insert(token, CapacityScopeEntry { scope, recorded_at: now });
+    }
+}
+
+pub fn take_capacity_scope(token: Uuid) -> Option<CapacityScope> {
+    let now = Instant::now();
+    let mut entries = capacity_scope_registry().lock().unwrap_or_else(|p| p.into_inner());
+    let entry = entries.remove(&token)?;
+    if now.duration_since(entry.recorded_at) > CAPACITY_SCOPE_TTL {
+        return None;
+    }
+    Some(entry.scope)
+}
+
+pub fn record_global_dirty_scope(scope: CapacityScope) {
+    if scope.disks.is_empty() {
+        return;
+    }
+
+    let mut dirty_scopes = global_dirty_scope_registry().lock().unwrap_or_else(|p| p.into_inner());
+    dirty_scopes.extend(scope.disks);
+}
+
+pub fn drain_global_dirty_scopes() -> Vec<CapacityScopeDisk> {
+    let mut dirty_scopes = global_dirty_scope_registry().lock().unwrap_or_else(|p| p.into_inner());
+    dirty_scopes.drain().collect()
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use std::sync::Mutex;
+    use std::thread;
+
+    fn test_lock() -> &'static Mutex<()> {
+        static LOCK: OnceLock<Mutex<()>> = OnceLock::new();
+        LOCK.get_or_init(|| Mutex::new(()))
+    }
+
+    fn clear_capacity_scope_registry_for_test() {
+        capacity_scope_registry()
+            .lock()
+            .unwrap_or_else(|poisoned| poisoned.into_inner())
+            .clear();
+        global_dirty_scope_registry()
+            .lock()
+            .unwrap_or_else(|poisoned| poisoned.into_inner())
+            .clear();
+    }
+
+    fn poison_capacity_scope_registry_for_test() {
+        let _ = thread::spawn(|| {
+            let _guard = capacity_scope_registry()
+                .lock()
+                .expect("capacity scope registry lock should succeed");
+            panic!("poison capacity scope registry");
+        })
+        .join();
+    }
+
+    fn poison_global_dirty_scope_registry_for_test() {
+        let _ = thread::spawn(|| {
+            let _guard = global_dirty_scope_registry()
+                .lock()
+                .expect("global dirty scope registry lock should succeed");
+            panic!("poison global dirty scope registry");
+        })
+        .join();
+    }
+
+    #[test]
+    fn record_and_take_capacity_scope_round_trips() {
+        let _guard = test_lock().lock().expect("test lock poisoned");
+        clear_capacity_scope_registry_for_test();
+        let token = Uuid::new_v4();
+        let scope = CapacityScope {
+            disks: vec![CapacityScopeDisk {
+                endpoint: "node-a".to_string(),
+                drive_path: "/tmp/disk-a".to_string(),
+            }],
+        };
+
+        record_capacity_scope(token, scope.clone());
+
+        assert_eq!(take_capacity_scope(token), Some(scope));
+        assert_eq!(take_capacity_scope(token), None);
+        clear_capacity_scope_registry_for_test();
+    }
+
+    #[test]
+    fn record_capacity_scope_merges_disks_for_same_token() {
+        let _guard = test_lock().lock().expect("test lock poisoned");
+        clear_capacity_scope_registry_for_test();
+        let token = Uuid::new_v4();
+        record_capacity_scope(
+            token,
+            CapacityScope {
+                disks: vec![CapacityScopeDisk {
+                    endpoint: "node-a".to_string(),
+                    drive_path: "/tmp/disk-a".to_string(),
+                }],
+            },
+        );
+        record_capacity_scope(
+            token,
+            CapacityScope {
+                disks: vec![
+                    CapacityScopeDisk {
+                        endpoint: "node-b".to_string(),
+                        drive_path: "/tmp/disk-b".to_string(),
+                    },
+                    CapacityScopeDisk {
+                        endpoint: "node-a".to_string(),
+                        drive_path: "/tmp/disk-a".to_string(),
+                    },
+                ],
+            },
+        );
+
+        let scope = take_capacity_scope(token).expect("scope should exist");
+        assert_eq!(scope.disks.len(), 2);
+        assert!(scope.disks.iter().any(|disk| disk.endpoint == "node-a"));
+        assert!(scope.disks.iter().any(|disk| disk.endpoint == "node-b"));
+        clear_capacity_scope_registry_for_test();
+    }
+
+    #[test]
+    fn record_capacity_scope_enforces_hard_limit() {
+        let _guard = test_lock().lock().expect("test lock poisoned");
+        clear_capacity_scope_registry_for_test();
+
+        for _ in 0..(CAPACITY_SCOPE_REGISTRY_HARD_LIMIT + 32) {
+            record_capacity_scope(
+                Uuid::new_v4(),
+                CapacityScope {
+                    disks: vec![CapacityScopeDisk {
+                        endpoint: "node-a".to_string(),
+                        drive_path: "/tmp/disk-a".to_string(),
+                    }],
+                },
+            );
+        }
+
+        let entries = capacity_scope_registry()
+            .lock()
+            .unwrap_or_else(|poisoned| poisoned.into_inner());
+        assert!(entries.len() <= CAPACITY_SCOPE_REGISTRY_HARD_LIMIT);
+        drop(entries);
+        clear_capacity_scope_registry_for_test();
+    }
+
+    #[test]
+    fn record_capacity_scope_recovers_from_poisoned_registry() {
+        let _guard = test_lock().lock().expect("test lock poisoned");
+        clear_capacity_scope_registry_for_test();
+        poison_capacity_scope_registry_for_test();
+        let token = Uuid::new_v4();
+        let scope = CapacityScope {
+            disks: vec![CapacityScopeDisk {
+                endpoint: "node-a".to_string(),
+                drive_path: "/tmp/disk-a".to_string(),
+            }],
+        };
+
+        record_capacity_scope(token, scope.clone());
+
+        assert_eq!(take_capacity_scope(token), Some(scope));
+        clear_capacity_scope_registry_for_test();
+    }
+
+    #[test]
+    fn record_and_drain_global_dirty_scope_round_trips() {
+        let _guard = test_lock().lock().expect("test lock poisoned");
+        clear_capacity_scope_registry_for_test();
+        record_global_dirty_scope(CapacityScope {
+            disks: vec![CapacityScopeDisk {
+                endpoint: "node-a".to_string(),
+                drive_path: "/tmp/disk-a".to_string(),
+            }],
+        });
+        record_global_dirty_scope(CapacityScope {
+            disks: vec![
+                CapacityScopeDisk {
+                    endpoint: "node-b".to_string(),
+                    drive_path: "/tmp/disk-b".to_string(),
+                },
+                CapacityScopeDisk {
+                    endpoint: "node-a".to_string(),
+                    drive_path: "/tmp/disk-a".to_string(),
+                },
+            ],
+        });
+
+        let drained = drain_global_dirty_scopes();
+        assert_eq!(drained.len(), 2);
+        assert!(drained.iter().any(|disk| disk.endpoint == "node-a"));
+        assert!(drained.iter().any(|disk| disk.endpoint == "node-b"));
+        assert!(drain_global_dirty_scopes().is_empty());
+        clear_capacity_scope_registry_for_test();
+    }
+
+    #[test]
+    fn record_global_dirty_scope_recovers_from_poisoned_registry() {
+        let _guard = test_lock().lock().expect("test lock poisoned");
+        clear_capacity_scope_registry_for_test();
+        poison_global_dirty_scope_registry_for_test();
+
+        record_global_dirty_scope(CapacityScope {
+            disks: vec![CapacityScopeDisk {
+                endpoint: "node-a".to_string(),
+                drive_path: "/tmp/disk-a".to_string(),
+            }],
+        });
+
+        let drained = drain_global_dirty_scopes();
+        assert_eq!(drained.len(), 1);
+        assert_eq!(drained[0].endpoint, "node-a");
+        clear_capacity_scope_registry_for_test();
+    }
+}

crates/common/src/data_usage.rs

@@ -51,14 +51,14 @@ impl AllTierStats {
     pub fn add_sizes(&mut self, tiers: HashMap<String, TierStats>) {
         for (tier, st) in tiers {
             self.tiers
-                .insert(tier.clone(), self.tiers.get(&tier).unwrap_or(&TierStats::default()).add(&st));
+                .insert(tier.clone(), self.tiers.get(&tier).copied().unwrap_or_default().add(&st));
         }
     }
 
     pub fn merge(&mut self, other: AllTierStats) {
         for (tier, st) in other.tiers {
             self.tiers
-                .insert(tier.clone(), self.tiers.get(&tier).unwrap_or(&TierStats::default()).add(&st));
+                .insert(tier.clone(), self.tiers.get(&tier).copied().unwrap_or_default().add(&st));
         }
     }
 
@@ -673,7 +673,7 @@ impl DataUsageCache {
         let mut leaves = Vec::new();
         let mut remove = total - limit;
         add(self, path, &mut leaves);
-        leaves.sort_by(|a, b| a.objects.cmp(&b.objects));
+        leaves.sort_by_key(|a| a.objects);
 
         while remove > 0 && !leaves.is_empty() {
             let Some(e) = leaves.first() else {
@@ -1097,7 +1097,7 @@ impl DataUsageInfo {
 
     /// Add bucket usage info
     pub fn add_bucket_usage(&mut self, bucket: String, usage: BucketUsageInfo) {
-        self.buckets_usage.insert(bucket.clone(), usage);
+        self.buckets_usage.insert(bucket, usage);
         self.buckets_count = self.buckets_usage.len() as u64;
         self.last_update = Some(SystemTime::now());
     }

crates/common/src/heal_channel.rs

@@ -18,7 +18,7 @@ use std::{
     fmt::{self, Display},
     sync::OnceLock,
 };
-use tokio::sync::{broadcast, mpsc};
+use tokio::sync::{broadcast, mpsc, oneshot};
 use uuid::Uuid;
 
 pub const HEAL_DELETE_DANGLING: bool = true;
@@ -206,11 +206,59 @@ pub struct HealOpts {
     pub set: Option<usize>,
 }
 
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub enum HealAdmissionDropReason {
+    QueueFull,
+    PolicyDropped,
+}
+
+impl HealAdmissionDropReason {
+    pub fn as_str(self) -> &'static str {
+        match self {
+            Self::QueueFull => "queue_full",
+            Self::PolicyDropped => "policy_dropped",
+        }
+    }
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub enum HealAdmissionResult {
+    Accepted,
+    Merged,
+    Full,
+    Dropped(HealAdmissionDropReason),
+}
+
+impl HealAdmissionResult {
+    pub fn result_label(self) -> &'static str {
+        match self {
+            Self::Accepted => "accepted",
+            Self::Merged => "merged",
+            Self::Full => "full",
+            Self::Dropped(_) => "dropped",
+        }
+    }
+
+    pub fn reason_label(self) -> &'static str {
+        match self {
+            Self::Dropped(reason) => reason.as_str(),
+            _ => "none",
+        }
+    }
+
+    pub fn is_admitted(self) -> bool {
+        matches!(self, Self::Accepted | Self::Merged)
+    }
+}
+
 /// Heal channel command type
-#[derive(Debug, Clone)]
+#[derive(Debug)]
 pub enum HealChannelCommand {
     /// Start a new heal task
-    Start(HealChannelRequest),
+    Start {
+        request: HealChannelRequest,
+        response_tx: oneshot::Sender<Result<HealAdmissionResult, String>>,
+    },
     /// Query heal task status
     Query { heal_path: String, client_token: String },
     /// Cancel heal task
@@ -339,9 +387,22 @@ pub fn subscribe_heal_responses() -> broadcast::Receiver<HealChannelResponse> {
     heal_response_sender().subscribe()
 }
 
+/// Send heal start request and wait for structured admission feedback.
+pub async fn send_heal_request_with_admission(request: HealChannelRequest) -> Result<HealAdmissionResult, String> {
+    let (response_tx, response_rx) = oneshot::channel();
+    send_heal_command(HealChannelCommand::Start { request, response_tx }).await?;
+    response_rx
+        .await
+        .map_err(|e| format!("Failed to receive heal admission response: {e}"))?
+}
+
 /// Send heal start request
 pub async fn send_heal_request(request: HealChannelRequest) -> Result<(), String> {
-    send_heal_command(HealChannelCommand::Start(request)).await
+    match send_heal_request_with_admission(request).await? {
+        HealAdmissionResult::Accepted | HealAdmissionResult::Merged => Ok(()),
+        HealAdmissionResult::Full => Err("Heal request queue is full".to_string()),
+        HealAdmissionResult::Dropped(reason) => Err(format!("Heal request dropped: {}", reason.as_str())),
+    }
 }
 
 /// Send heal query request
@@ -450,7 +511,7 @@ pub fn lc_has_active_rules(config: &BucketLifecycleConfiguration, prefix: &str)
         }
 
         if let Some(e) = &rule.noncurrent_version_expiration {
-            if let Some(true) = e.noncurrent_days.map(|d| d > 0) {
+            if e.noncurrent_days.is_some() {
                 return true;
             }
             if let Some(true) = e.newer_noncurrent_versions.map(|d| d > 0) {
@@ -542,6 +603,19 @@ pub async fn send_heal_disk(set_disk_id: String, priority: Option<HealChannelPri
 mod tests {
     use super::*;
 
+    #[test]
+    fn heal_admission_result_labels_are_stable() {
+        assert_eq!(HealAdmissionResult::Accepted.result_label(), "accepted");
+        assert_eq!(HealAdmissionResult::Merged.result_label(), "merged");
+        assert_eq!(HealAdmissionResult::Full.result_label(), "full");
+        assert_eq!(
+            HealAdmissionResult::Dropped(HealAdmissionDropReason::QueueFull).reason_label(),
+            "queue_full"
+        );
+        assert!(HealAdmissionResult::Merged.is_admitted());
+        assert!(!HealAdmissionResult::Full.is_admitted());
+    }
+
     #[tokio::test]
     async fn heal_response_broadcast_reaches_subscriber() {
         let mut receiver = subscribe_heal_responses();

crates/common/src/internode_metrics.rs

@@ -0,0 +1,166 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use metrics::{counter, gauge};
+use std::sync::{
+    Arc, LazyLock,
+    atomic::{AtomicU64, Ordering},
+};
+use std::time::{Duration, SystemTime, UNIX_EPOCH};
+
+#[derive(Debug, Clone, Copy, Default, PartialEq, Eq)]
+pub struct InternodeMetricsSnapshot {
+    pub sent_bytes_total: u64,
+    pub recv_bytes_total: u64,
+    pub outgoing_requests_total: u64,
+    pub incoming_requests_total: u64,
+    pub errors_total: u64,
+    pub dial_errors_total: u64,
+    pub dial_avg_time_nanos: u64,
+    pub last_dial_unix_millis: u64,
+}
+
+#[derive(Debug, Default)]
+pub struct InternodeMetrics {
+    sent_bytes_total: AtomicU64,
+    recv_bytes_total: AtomicU64,
+    outgoing_requests_total: AtomicU64,
+    incoming_requests_total: AtomicU64,
+    errors_total: AtomicU64,
+    dial_errors_total: AtomicU64,
+    dial_total_time_nanos: AtomicU64,
+    dial_samples_total: AtomicU64,
+    last_dial_unix_millis: AtomicU64,
+}
+
+impl InternodeMetrics {
+    pub fn record_sent_bytes(&self, bytes: usize) {
+        let bytes = bytes as u64;
+        if bytes == 0 {
+            return;
+        }
+        self.sent_bytes_total.fetch_add(bytes, Ordering::Relaxed);
+        counter!("rustfs_system_network_internode_sent_bytes_total").increment(bytes);
+    }
+
+    pub fn record_recv_bytes(&self, bytes: usize) {
+        let bytes = bytes as u64;
+        if bytes == 0 {
+            return;
+        }
+        self.recv_bytes_total.fetch_add(bytes, Ordering::Relaxed);
+        counter!("rustfs_system_network_internode_recv_bytes_total").increment(bytes);
+    }
+
+    pub fn record_outgoing_request(&self) {
+        self.outgoing_requests_total.fetch_add(1, Ordering::Relaxed);
+        counter!("rustfs_system_network_internode_requests_outgoing_total").increment(1);
+    }
+
+    pub fn record_incoming_request(&self) {
+        self.incoming_requests_total.fetch_add(1, Ordering::Relaxed);
+        counter!("rustfs_system_network_internode_requests_incoming_total").increment(1);
+    }
+
+    pub fn record_error(&self) {
+        self.errors_total.fetch_add(1, Ordering::Relaxed);
+        counter!("rustfs_system_network_internode_errors_total").increment(1);
+    }
+
+    pub fn record_dial_result(&self, duration: Duration, success: bool) {
+        let elapsed_nanos = duration.as_nanos().min(u128::from(u64::MAX)) as u64;
+        self.dial_total_time_nanos.fetch_add(elapsed_nanos, Ordering::Relaxed);
+        let samples = self.dial_samples_total.fetch_add(1, Ordering::Relaxed) + 1;
+        let total = self.dial_total_time_nanos.load(Ordering::Relaxed);
+        gauge!("rustfs_system_network_internode_dial_avg_time_nanos").set(total as f64 / samples as f64);
+
+        if !success {
+            self.dial_errors_total.fetch_add(1, Ordering::Relaxed);
+            counter!("rustfs_system_network_internode_dial_errors_total").increment(1);
+        }
+
+        let now_ms = SystemTime::now()
+            .duration_since(UNIX_EPOCH)
+            .unwrap_or_default()
+            .as_millis()
+            .min(u128::from(u64::MAX)) as u64;
+        self.last_dial_unix_millis.store(now_ms, Ordering::Relaxed);
+    }
+
+    pub fn snapshot(&self) -> InternodeMetricsSnapshot {
+        let dial_samples_total = self.dial_samples_total.load(Ordering::Relaxed);
+        let dial_total_time_nanos = self.dial_total_time_nanos.load(Ordering::Relaxed);
+        let dial_avg_time_nanos = dial_total_time_nanos.checked_div(dial_samples_total).unwrap_or(0);
+
+        InternodeMetricsSnapshot {
+            sent_bytes_total: self.sent_bytes_total.load(Ordering::Relaxed),
+            recv_bytes_total: self.recv_bytes_total.load(Ordering::Relaxed),
+            outgoing_requests_total: self.outgoing_requests_total.load(Ordering::Relaxed),
+            incoming_requests_total: self.incoming_requests_total.load(Ordering::Relaxed),
+            errors_total: self.errors_total.load(Ordering::Relaxed),
+            dial_errors_total: self.dial_errors_total.load(Ordering::Relaxed),
+            dial_avg_time_nanos,
+            last_dial_unix_millis: self.last_dial_unix_millis.load(Ordering::Relaxed),
+        }
+    }
+
+    #[doc(hidden)]
+    pub fn reset_for_test(&self) {
+        self.sent_bytes_total.store(0, Ordering::Relaxed);
+        self.recv_bytes_total.store(0, Ordering::Relaxed);
+        self.outgoing_requests_total.store(0, Ordering::Relaxed);
+        self.incoming_requests_total.store(0, Ordering::Relaxed);
+        self.errors_total.store(0, Ordering::Relaxed);
+        self.dial_errors_total.store(0, Ordering::Relaxed);
+        self.dial_total_time_nanos.store(0, Ordering::Relaxed);
+        self.dial_samples_total.store(0, Ordering::Relaxed);
+        self.last_dial_unix_millis.store(0, Ordering::Relaxed);
+    }
+}
+
+pub fn global_internode_metrics() -> &'static Arc<InternodeMetrics> {
+    static GLOBAL_INTERNODE_METRICS: LazyLock<Arc<InternodeMetrics>> = LazyLock::new(|| Arc::new(InternodeMetrics::default()));
+    &GLOBAL_INTERNODE_METRICS
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn snapshot_reports_recorded_values() {
+        let metrics = global_internode_metrics();
+        metrics.reset_for_test();
+
+        metrics.record_sent_bytes(64);
+        metrics.record_recv_bytes(32);
+        metrics.record_outgoing_request();
+        metrics.record_incoming_request();
+        metrics.record_error();
+        metrics.record_dial_result(Duration::from_millis(9), true);
+        metrics.record_dial_result(Duration::from_millis(3), false);
+
+        let snapshot = metrics.snapshot();
+        assert_eq!(snapshot.sent_bytes_total, 64);
+        assert_eq!(snapshot.recv_bytes_total, 32);
+        assert_eq!(snapshot.outgoing_requests_total, 1);
+        assert_eq!(snapshot.incoming_requests_total, 1);
+        assert_eq!(snapshot.errors_total, 1);
+        assert_eq!(snapshot.dial_errors_total, 1);
+        assert_eq!(snapshot.dial_avg_time_nanos, 6_000_000);
+        assert!(snapshot.last_dial_unix_millis > 0);
+
+        metrics.reset_for_test();
+    }
+}

crates/common/src/lib.rs

@@ -13,10 +13,12 @@
 // limitations under the License.
 
 pub mod bucket_stats;
+pub mod capacity_scope;
 // pub mod error;
 pub mod data_usage;
 pub mod globals;
 pub mod heal_channel;
+pub mod internode_metrics;
 pub mod last_minute;
 pub mod metrics;
 mod readiness;

crates/common/src/metrics.rs

@@ -22,12 +22,12 @@ use std::{
     future::Future,
     pin::Pin,
     sync::{
-        Arc, OnceLock,
+        Arc, Mutex, OnceLock,
         atomic::{AtomicU64, Ordering},
     },
     time::{Duration, SystemTime},
 };
-use tokio::sync::{Mutex, RwLock};
+use tokio::sync::RwLock;
 
 #[derive(Debug, Clone, Copy, PartialEq, Eq)]
 pub enum IlmAction {
@@ -110,7 +110,7 @@ pub fn global_metrics() -> &'static Arc<Metrics> {
     GLOBAL_METRICS.get_or_init(|| Arc::new(Metrics::new()))
 }
 
-#[derive(Clone, Debug, PartialEq, PartialOrd)]
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
 pub enum Metric {
     // START Realtime metrics, that only records
     // last minute latencies and total operation count.
@@ -188,7 +188,6 @@ impl Metric {
         if index >= Self::Last as usize {
             return None;
         }
-        // Safe conversion using match instead of unsafe transmute
         match index {
             0 => Some(Self::ReadMetadata),
             1 => Some(Self::CheckMissing),
@@ -220,16 +219,48 @@ impl Metric {
     }
 }
 
-/// Thread-safe wrapper for LastMinuteLatency with atomic operations
-#[derive(Default)]
+// ---------------------------------------------------------------------------
+// LockedLastMinuteLatency
+// ---------------------------------------------------------------------------
+//
+// Uses std::sync::Mutex instead of tokio::sync::Mutex.
+//
+// Rationale: the critical section is a handful of integer additions inside
+// LastMinuteLatency::add_all / get_total — no I/O, no blocking syscalls, no
+// awaiting.  A std blocking mutex is cheaper (no task-wakeup overhead) and,
+// crucially, lets every caller be *synchronous*.  That eliminates the need to
+// spawn a background task just to record a duration, which was the only reason
+// tokio::spawn appeared in log/time/time_size/time_n/time_ilm.
+//
+// Note: vec![LockedLastMinuteLatency::default(); N] with the old Arc-based
+// Clone made every element share the *same* inner mutex — a latent bug where
+// all metrics slots wrote to one counter.  The new Clone creates a fresh
+// independent Mutex per element, matching the intent.
+
+/// Thread-safe wrapper for LastMinuteLatency backed by a std blocking mutex.
 pub struct LockedLastMinuteLatency {
+    // Arc so Clone is cheap *and* each cloned value stays independent (its own
+    // allocation).  We never hand out the Arc to two Metrics at once; the Arc
+    // is purely a convenience for the Clone impl below.
     latency: Arc<Mutex<LastMinuteLatency>>,
 }
 
+impl Default for LockedLastMinuteLatency {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
+// Produce a fresh, independent slot — *not* a shared alias.  This is what
+// vec![val; N] and #[derive(Clone)] on the parent struct both need.
 impl Clone for LockedLastMinuteLatency {
     fn clone(&self) -> Self {
+        let inner = match self.latency.lock() {
+            Ok(guard) => guard.clone(),
+            Err(poisoned) => poisoned.into_inner().clone(),
+        };
         Self {
-            latency: Arc::clone(&self.latency),
+            latency: Arc::new(Mutex::new(inner)),
         }
     }
 }
@@ -241,14 +272,13 @@ impl LockedLastMinuteLatency {
         }
     }
 
-    /// Add a duration measurement
-    pub async fn add(&self, duration: Duration) {
-        self.add_size(duration, 0).await;
+    /// Record a duration sample (no size).
+    pub fn add(&self, duration: Duration) {
+        self.add_size(duration, 0);
     }
 
-    /// Add a duration measurement with size
-    pub async fn add_size(&self, duration: Duration, size: u64) {
-        let mut latency = self.latency.lock().await;
+    /// Record a duration sample with an associated byte count.
+    pub fn add_size(&self, duration: Duration, size: u64) {
         let now = SystemTime::now()
             .duration_since(SystemTime::UNIX_EPOCH)
             .unwrap_or_default()
@@ -259,17 +289,27 @@ impl LockedLastMinuteLatency {
             total: duration.as_secs(),
             size,
         };
-        latency.add_all(now, &elem);
+
+        match self.latency.lock() {
+            Ok(mut guard) => guard.add_all(now, &elem),
+            Err(poisoned) => poisoned.into_inner().add_all(now, &elem),
+        }
     }
 
-    /// Get total accumulated metrics for the last minute
-    pub async fn total(&self) -> AccElem {
-        let mut latency = self.latency.lock().await;
-        latency.get_total()
+    /// Return accumulated totals for the last minute window.
+    pub fn total(&self) -> AccElem {
+        match self.latency.lock() {
+            Ok(mut latency) => latency.get_total(),
+            Err(poisoned) => poisoned.into_inner().get_total(),
+        }
     }
 }
 
-/// Current path tracker for monitoring active scan paths
+// ---------------------------------------------------------------------------
+// CurrentPathTracker — unchanged, still uses tokio::sync::RwLock because it
+// lives inside async path-update callbacks.
+// ---------------------------------------------------------------------------
+
 struct CurrentPathTracker {
     current_path: Arc<RwLock<String>>,
 }
@@ -290,17 +330,16 @@ impl CurrentPathTracker {
     }
 }
 
-/// Main scanner metrics structure
+// ---------------------------------------------------------------------------
+// Metrics
+// ---------------------------------------------------------------------------
+
 pub struct Metrics {
-    // All fields must be accessed atomically and aligned.
     operations: Vec<AtomicU64>,
     latency: Vec<LockedLastMinuteLatency>,
     actions: Vec<AtomicU64>,
     actions_latency: Vec<LockedLastMinuteLatency>,
-    // Current paths contains disk -> tracker mappings
     current_paths: Arc<RwLock<HashMap<String, Arc<CurrentPathTracker>>>>,
-
-    // Cycle information
     cycle_info: Arc<RwLock<Option<CurrentCycle>>>,
 }
 
@@ -331,8 +370,6 @@ const OTEL_SCANNER_CYCLES: &str = "rustfs_scanner_cycles_total";
 const OTEL_SCANNER_CYCLE_DURATION_SECONDS: &str = "rustfs_scanner_cycle_duration_seconds";
 const OTEL_SCANNER_BUCKET_DRIVE_DURATION_SECONDS: &str = "rustfs_scanner_bucket_drive_duration_seconds";
 
-/// Emit an OTEL counter increment for the given scanner metric.
-/// ScanCycle and ScanBucketDrive are handled by dedicated emit functions with labels.
 fn emit_otel_counter(metric: usize, count: u64) {
     match Metric::from_index(metric) {
         Some(Metric::ScanObject) => {
@@ -345,8 +382,6 @@ fn emit_otel_counter(metric: usize, count: u64) {
     }
 }
 
-/// Emit OTel metrics for a completed scan cycle.
-/// Counter with result label + gauge for last successful cycle duration.
 pub fn emit_scan_cycle_complete(success: bool, duration: Duration) {
     let result = if success { "success" } else { "error" };
     metrics::counter!(OTEL_SCANNER_CYCLES, "result" => result).increment(1);
@@ -355,8 +390,6 @@ pub fn emit_scan_cycle_complete(success: bool, duration: Duration) {
     }
 }
 
-/// Emit OTel metrics for a completed bucket-drive scan.
-/// Counter with result/bucket/disk labels + histogram for duration.
 pub fn emit_scan_bucket_drive_complete(success: bool, bucket: &str, disk: &str, duration: Duration) {
     let result = if success { "success" } else { "error" };
     metrics::counter!(
@@ -376,226 +409,202 @@ pub fn emit_scan_bucket_drive_complete(success: bool, bucket: &str, disk: &str,
 
 impl Metrics {
     pub fn new() -> Self {
-        let operations = (0..Metric::Last as usize).map(|_| AtomicU64::new(0)).collect();
-
-        let latency = (0..Metric::LastRealtime as usize)
-            .map(|_| LockedLastMinuteLatency::new())
-            .collect();
-
         Self {
-            operations,
-            latency,
+            operations: (0..Metric::Last as usize).map(|_| AtomicU64::new(0)).collect(),
+            // Each slot gets its own fresh LockedLastMinuteLatency so that
+            // different metrics never accidentally share state.
+            latency: (0..Metric::LastRealtime as usize)
+                .map(|_| LockedLastMinuteLatency::new())
+                .collect(),
             actions: (0..IlmAction::ActionCount as usize).map(|_| AtomicU64::new(0)).collect(),
-            actions_latency: vec![LockedLastMinuteLatency::default(); IlmAction::ActionCount as usize],
+            actions_latency: (0..IlmAction::ActionCount as usize)
+                .map(|_| LockedLastMinuteLatency::new())
+                .collect(),
             current_paths: Arc::new(RwLock::new(HashMap::new())),
             cycle_info: Arc::new(RwLock::new(None)),
         }
     }
 
-    /// Log scanner action with custom metadata - compatible with existing usage
+    // -----------------------------------------------------------------------
+    // Metric recording helpers
+    //
+    // All of these are now pure sync closures.  No tokio::spawn, no heap-
+    // allocated future, no scheduler overhead — just an atomic increment and a
+    // std-mutex lock for fewer than ~10 ns of work.
+    // -----------------------------------------------------------------------
+
+    /// Return a closure that records one observation of `metric` (with
+    /// optional caller-supplied metadata).  Call it once the operation ends.
     pub fn log(metric: Metric) -> impl Fn(&HashMap<String, String>) {
-        let metric = metric as usize;
-        let start_time = SystemTime::now();
+        let metric_idx = metric as usize;
+        let start = SystemTime::now();
         move |_custom: &HashMap<String, String>| {
-            let duration = SystemTime::now().duration_since(start_time).unwrap_or_default();
-
-            // Update operation count
-            global_metrics().operations[metric].fetch_add(1, Ordering::Relaxed);
-            emit_otel_counter(metric, 1);
-
-            // Update latency for realtime metrics (spawn async task for this)
-            if (metric) < Metric::LastRealtime as usize {
-                let metric_index = metric;
-                tokio::spawn(async move {
-                    global_metrics().latency[metric_index].add(duration).await;
-                });
-            }
-
-            // Log trace metrics
-            if metric as u8 > Metric::StartTrace as u8 {
-                //debug!(metric = metric.as_str(), duration_ms = duration.as_millis(), "Scanner trace metric");
+            let duration = SystemTime::now().duration_since(start).unwrap_or_default();
+            global_metrics().operations[metric_idx].fetch_add(1, Ordering::Relaxed);
+            emit_otel_counter(metric_idx, 1);
+            if metric_idx < Metric::LastRealtime as usize {
+                global_metrics().latency[metric_idx].add(duration);
             }
         }
     }
 
-    /// Time scanner action with size - returns function that takes size
+    /// Return a closure that records one observation of `metric` together with
+    /// a byte count.  Call `done(size_bytes)` when the operation ends.
     pub fn time_size(metric: Metric) -> impl Fn(u64) {
-        let metric = metric as usize;
-        let start_time = SystemTime::now();
+        let metric_idx = metric as usize;
+        let start = SystemTime::now();
         move |size: u64| {
-            let duration = SystemTime::now().duration_since(start_time).unwrap_or_default();
-
-            // Update operation count
-            global_metrics().operations[metric].fetch_add(1, Ordering::Relaxed);
-            emit_otel_counter(metric, 1);
-
-            // Update latency for realtime metrics with size (spawn async task)
-            if (metric) < Metric::LastRealtime as usize {
-                let metric_index = metric;
-                tokio::spawn(async move {
-                    global_metrics().latency[metric_index].add_size(duration, size).await;
-                });
+            let duration = SystemTime::now().duration_since(start).unwrap_or_default();
+            global_metrics().operations[metric_idx].fetch_add(1, Ordering::Relaxed);
+            emit_otel_counter(metric_idx, 1);
+            if metric_idx < Metric::LastRealtime as usize {
+                global_metrics().latency[metric_idx].add_size(duration, size);
             }
         }
     }
 
-    /// Time a scanner action - returns a closure to call when done
+    /// Return a closure that records one observation of `metric`.
+    /// Call `done()` when the operation ends.
     pub fn time(metric: Metric) -> impl Fn() {
-        let metric = metric as usize;
-        let start_time = SystemTime::now();
+        let metric_idx = metric as usize;
+        let start = SystemTime::now();
         move || {
-            let duration = SystemTime::now().duration_since(start_time).unwrap_or_default();
-
-            // Update operation count
-            global_metrics().operations[metric].fetch_add(1, Ordering::Relaxed);
-            emit_otel_counter(metric, 1);
-
-            // Update latency for realtime metrics (spawn async task)
-            if (metric) < Metric::LastRealtime as usize {
-                let metric_index = metric;
-                tokio::spawn(async move {
-                    global_metrics().latency[metric_index].add(duration).await;
-                });
+            let duration = SystemTime::now().duration_since(start).unwrap_or_default();
+            global_metrics().operations[metric_idx].fetch_add(1, Ordering::Relaxed);
+            emit_otel_counter(metric_idx, 1);
+            if metric_idx < Metric::LastRealtime as usize {
+                global_metrics().latency[metric_idx].add(duration);
             }
         }
     }
 
-    /// Time N scanner actions - returns function that takes count, then returns completion function
+    /// Return a two-stage closure: first call takes an item count, second call
+    /// (returned closure) fires when the batch of `count` operations ends.
     pub fn time_n(metric: Metric) -> Box<dyn Fn(usize) -> Box<dyn Fn() + Send + Sync> + Send + Sync> {
-        let metric = metric as usize;
-        let start_time = SystemTime::now();
+        let metric_idx = metric as usize;
+        let start = SystemTime::now();
         Box::new(move |count: usize| {
             Box::new(move || {
-                let duration = SystemTime::now().duration_since(start_time).unwrap_or_default();
-
-                // Update operation count
-                global_metrics().operations[metric].fetch_add(count as u64, Ordering::Relaxed);
-                emit_otel_counter(metric, count as u64);
-
-                // Update latency for realtime metrics (spawn async task)
-                if (metric) < Metric::LastRealtime as usize {
-                    let metric_index = metric;
-                    tokio::spawn(async move {
-                        global_metrics().latency[metric_index].add(duration).await;
-                    });
+                let duration = SystemTime::now().duration_since(start).unwrap_or_default();
+                global_metrics().operations[metric_idx].fetch_add(count as u64, Ordering::Relaxed);
+                emit_otel_counter(metric_idx, count as u64);
+                if metric_idx < Metric::LastRealtime as usize {
+                    global_metrics().latency[metric_idx].add(duration);
                 }
             })
         })
     }
 
-    /// Time ILM action with versions - returns function that takes versions, then returns completion function
+    /// Return a two-stage closure for ILM actions: first call takes a version
+    /// count, second call fires when the action completes.
     pub fn time_ilm(a: IlmAction) -> Box<dyn Fn(u64) -> Box<dyn Fn() + Send + Sync> + Send + Sync> {
-        let a_clone = a as usize;
-        if a_clone == IlmAction::NoneAction as usize || a_clone >= IlmAction::ActionCount as usize {
-            return Box::new(move |_: u64| Box::new(move || {}));
+        let a_idx = a as usize;
+        if a_idx == IlmAction::NoneAction as usize || a_idx >= IlmAction::ActionCount as usize {
+            return Box::new(|_| Box::new(|| {}));
         }
         let start = SystemTime::now();
         Box::new(move |versions: u64| {
             Box::new(move || {
-                let duration = SystemTime::now().duration_since(start).unwrap_or(Duration::from_secs(0));
-                tokio::spawn(async move {
-                    global_metrics().actions[a_clone].fetch_add(versions, Ordering::Relaxed);
-                    global_metrics().actions_latency[a_clone].add(duration).await;
-                });
+                let duration = SystemTime::now().duration_since(start).unwrap_or_default();
+                global_metrics().actions[a_idx].fetch_add(versions, Ordering::Relaxed);
+                global_metrics().actions_latency[a_idx].add(duration);
             })
         })
     }
 
-    /// Increment time with specific duration
-    pub async fn inc_time(metric: Metric, duration: Duration) {
-        let metric = metric as usize;
-        // Update operation count
-        global_metrics().operations[metric].fetch_add(1, Ordering::Relaxed);
-        emit_otel_counter(metric, 1);
-
-        // Update latency for realtime metrics
-        if (metric) < Metric::LastRealtime as usize {
-            global_metrics().latency[metric].add(duration).await;
+    /// Record a single observation of `metric` with a caller-supplied duration.
+    /// No longer async — nothing inside requires it.
+    pub fn inc_time(metric: Metric, duration: Duration) {
+        let metric_idx = metric as usize;
+        global_metrics().operations[metric_idx].fetch_add(1, Ordering::Relaxed);
+        emit_otel_counter(metric_idx, 1);
+        if metric_idx < Metric::LastRealtime as usize {
+            global_metrics().latency[metric_idx].add(duration);
         }
     }
 
-    /// Get lifetime operation count for a metric
+    // -----------------------------------------------------------------------
+    // Read-side helpers
+    // -----------------------------------------------------------------------
+
+    /// Lifetime operation count for `metric`.
     pub fn lifetime(&self, metric: Metric) -> u64 {
-        let metric = metric as usize;
-        if (metric) >= Metric::Last as usize {
+        let idx = metric as usize;
+        if idx >= Metric::Last as usize {
             return 0;
         }
-        self.operations[metric].load(Ordering::Relaxed)
+        self.operations[idx].load(Ordering::Relaxed)
     }
 
-    /// Get last minute statistics for a metric
-    pub async fn last_minute(&self, metric: Metric) -> AccElem {
-        let metric = metric as usize;
-        if (metric) >= Metric::LastRealtime as usize {
+    /// Last-minute accumulated stats for a realtime metric.
+    /// No longer async — LockedLastMinuteLatency::total() is now synchronous.
+    pub fn last_minute(&self, metric: Metric) -> AccElem {
+        let idx = metric as usize;
+        if idx >= Metric::LastRealtime as usize {
             return AccElem::default();
         }
-        self.latency[metric].total().await
+        self.latency[idx].total()
     }
 
-    /// Set current cycle information
+    /// Replace the current cycle record.
     pub async fn set_cycle(&self, cycle: Option<CurrentCycle>) {
         *self.cycle_info.write().await = cycle;
     }
 
-    /// Get current cycle information
+    /// Read the current cycle record.
     pub async fn get_cycle(&self) -> Option<CurrentCycle> {
         self.cycle_info.read().await.clone()
     }
 
-    /// Get current active paths
+    /// Snapshot of every path currently being scanned.
     pub async fn get_current_paths(&self) -> Vec<String> {
-        let mut result = Vec::new();
         let paths = self.current_paths.read().await;
-
+        let mut result = Vec::with_capacity(paths.len());
         for (disk, tracker) in paths.iter() {
-            let path = tracker.get_path().await;
-            result.push(format!("{disk}/{path}"));
+            result.push(format!("{disk}/{}", tracker.get_path().await));
         }
-
         result
     }
 
-    /// Get number of active drives
+    /// Number of drives with an active scan in progress.
     pub async fn active_drives(&self) -> usize {
         self.current_paths.read().await.len()
     }
 
-    /// Generate metrics report
+    /// Build a full metrics report snapshot.
     pub async fn report(&self) -> M_ScannerMetrics {
-        let mut metrics = M_ScannerMetrics::default();
+        let mut m = M_ScannerMetrics::default();
 
-        // Set cycle information
         if let Some(cycle) = self.get_cycle().await {
-            metrics.current_cycle = cycle.current;
-            metrics.cycles_completed_at = cycle.cycle_completed;
-            metrics.current_started = cycle.started;
+            m.current_cycle = cycle.current;
+            m.cycles_completed_at = cycle.cycle_completed;
+            m.current_started = cycle.started;
         }
 
-        // Replace default start time with global init time if it's the placeholder
         if let Some(init_time) = crate::get_global_init_time().await {
-            metrics.current_started = init_time;
+            m.current_started = init_time;
         }
 
-        metrics.collected_at = Utc::now();
-        metrics.active_paths = self.get_current_paths().await;
+        m.collected_at = Utc::now();
+        m.active_paths = self.get_current_paths().await;
 
-        // Lifetime operations
+        // Lifetime operation counts
         for i in 0..Metric::Last as usize {
             let count = self.operations[i].load(Ordering::Relaxed);
             if count > 0
                 && let Some(metric) = Metric::from_index(i)
             {
-                metrics.life_time_ops.insert(metric.as_str().to_string(), count);
+                m.life_time_ops.insert(metric.as_str().to_string(), count);
             }
         }
 
-        // Last minute statistics for realtime metrics
+        // Last-minute stats for realtime metrics — now plain sync calls
         for i in 0..Metric::LastRealtime as usize {
-            let last_min = self.latency[i].total().await;
+            let last_min = self.latency[i].total();
             if last_min.n > 0
                 && let Some(metric) = Metric::from_index(i)
             {
-                metrics.last_minute.actions.insert(
+                m.last_minute.actions.insert(
                     metric.as_str().to_string(),
                     TimedAction {
                         count: last_min.n,
@@ -606,23 +615,23 @@ impl Metrics {
             }
         }
 
-        // Lifetime ILM operations
+        // Lifetime ILM counts
         for i in 0..IlmAction::ActionCount as usize {
             let count = self.actions[i].load(Ordering::Relaxed);
             if count > 0
                 && let Some(action) = IlmAction::from_index(i)
             {
-                metrics.life_time_ilm.insert(action.as_str().to_string(), count);
+                m.life_time_ilm.insert(action.as_str().to_string(), count);
             }
         }
 
-        // Last minute ILM latency
+        // Last-minute ILM latency — plain sync calls
         for i in 0..IlmAction::ActionCount as usize {
-            let last_min = self.actions_latency[i].total().await;
+            let last_min = self.actions_latency[i].total();
             if last_min.n > 0
                 && let Some(action) = IlmAction::from_index(i)
             {
-                metrics.last_minute.ilm.insert(
+                m.last_minute.ilm.insert(
                     action.as_str().to_string(),
                     TimedAction {
                         count: last_min.n,
@@ -633,56 +642,65 @@ impl Metrics {
             }
         }
 
-        metrics
+        m
     }
 }
 
-// Type aliases for compatibility with existing code
-pub type UpdateCurrentPathFn = Arc<dyn Fn(&str) -> Pin<Box<dyn Future<Output = ()> + Send>> + Send + Sync>;
-pub type CloseDiskFn = Arc<dyn Fn() -> Pin<Box<dyn Future<Output = ()> + Send>> + Send + Sync>;
-
-/// Create a current path updater for tracking scan progress
-pub fn current_path_updater(disk: &str, initial: &str) -> (UpdateCurrentPathFn, CloseDiskFn) {
-    let tracker = Arc::new(CurrentPathTracker::new(initial.to_string()));
-    let disk_name = disk.to_string();
-
-    // Store the tracker in global metrics
-    let tracker_clone = Arc::clone(&tracker);
-    let disk_clone = disk_name.clone();
-    tokio::spawn(async move {
-        global_metrics().current_paths.write().await.insert(disk_clone, tracker_clone);
-    });
-
-    let update_fn = {
-        let tracker = Arc::clone(&tracker);
-        Arc::new(move |path: &str| -> Pin<Box<dyn Future<Output = ()> + Send>> {
-            let tracker = Arc::clone(&tracker);
-            let path = path.to_string();
-            Box::pin(async move {
-                tracker.update_path(path).await;
-            })
-        })
-    };
-
-    let done_fn = {
-        let disk_name = disk_name.clone();
-        Arc::new(move || -> Pin<Box<dyn Future<Output = ()> + Send>> {
-            let disk_name = disk_name.clone();
-            Box::pin(async move {
-                global_metrics().current_paths.write().await.remove(&disk_name);
-            })
-        })
-    };
-
-    (update_fn, done_fn)
-}
-
 impl Default for Metrics {
     fn default() -> Self {
         Self::new()
     }
 }
 
+// ---------------------------------------------------------------------------
+// Path tracking helpers
+// ---------------------------------------------------------------------------
+
+pub type UpdateCurrentPathFn = Arc<dyn Fn(&str) -> Pin<Box<dyn Future<Output = ()> + Send>> + Send + Sync>;
+pub type CloseDiskFn = Arc<dyn Fn() -> Pin<Box<dyn Future<Output = ()> + Send>> + Send + Sync>;
+
+/// Register a new disk in the global path tracker and return two callbacks:
+/// one to update the current path and one to deregister the disk when done.
+pub fn current_path_updater(disk: &str, initial: &str) -> (UpdateCurrentPathFn, CloseDiskFn) {
+    let tracker = Arc::new(CurrentPathTracker::new(initial.to_string()));
+    let disk_name = disk.to_string();
+
+    let tracker_clone = Arc::clone(&tracker);
+    let disk_insert = disk_name.clone();
+    tokio::spawn(async move {
+        global_metrics()
+            .current_paths
+            .write()
+            .await
+            .insert(disk_insert, tracker_clone);
+    });
+
+    let update_fn: UpdateCurrentPathFn = {
+        let tracker = Arc::clone(&tracker);
+        Arc::new(move |path: &str| {
+            let tracker = Arc::clone(&tracker);
+            let path = path.to_string();
+            Box::pin(async move { tracker.update_path(path).await })
+        })
+    };
+
+    let done_fn: CloseDiskFn = {
+        let disk = disk_name;
+        Arc::new(move || {
+            let disk = disk.clone();
+            Box::pin(async move {
+                global_metrics().current_paths.write().await.remove(&disk);
+            })
+        })
+    };
+
+    (update_fn, done_fn)
+}
+
+// ---------------------------------------------------------------------------
+// CloseDiskGuard
+// ---------------------------------------------------------------------------
+
 pub struct CloseDiskGuard(CloseDiskFn);
 
 impl CloseDiskGuard {
@@ -697,16 +715,10 @@ impl CloseDiskGuard {
 
 impl Drop for CloseDiskGuard {
     fn drop(&mut self) {
-        // Drop cannot be async, so we spawn the async cleanup task
-        // The task will run in the background and complete asynchronously
         if let Ok(handle) = tokio::runtime::Handle::try_current() {
             let close_fn = self.0.clone();
-            handle.spawn(async move {
-                close_fn().await;
-            });
-        } else {
-            // If we're not in a tokio runtime context, we can't spawn
-            // This is a best-effort cleanup, so we just skip it
+            handle.spawn(async move { close_fn().await });
         }
+        // If there is no runtime we are in a test or shutdown path; skip cleanup.
     }
 }

crates/common/src/readiness.rs

@@ -60,6 +60,23 @@ impl GlobalReadiness {
     pub fn is_ready(&self) -> bool {
         self.status.load(Ordering::SeqCst) == SystemStage::FullReady as u8
     }
+
+    /// Get the current system stage
+    /// # Returns
+    /// The current SystemStage of the service
+    pub fn current_stage(&self) -> SystemStage {
+        match self.status.load(Ordering::SeqCst) {
+            0 => SystemStage::Booting,
+            1 => SystemStage::StorageReady,
+            2 => SystemStage::IamReady,
+            3 => SystemStage::FullReady,
+            invalid => {
+                debug_assert!(false, "GlobalReadiness::current_stage: invalid status value {}", invalid);
+                // Fallback to the most conservative stage on invalid values
+                SystemStage::Booting
+            }
+        }
+    }
 }
 
 #[cfg(test)]

crates/concurrency/Cargo.toml

@@ -0,0 +1,43 @@
+[package]
+name = "rustfs-concurrency"
+version.workspace = true
+edition.workspace = true
+license.workspace = true
+repository.workspace = true
+rust-version.workspace = true
+homepage.workspace = true
+description = "Concurrency management for RustFS - timeout, locking, backpressure, and I/O scheduling"
+keywords = ["rustfs", "concurrency", "timeout", "backpressure", "scheduling"]
+categories = ["concurrency", "filesystem"]
+
+[dependencies]
+# Internal crates
+rustfs-io-core = { workspace = true }
+rustfs-io-metrics = { workspace = true }
+
+# Async runtime
+tokio = { workspace = true, features = ["sync", "time", "rt"] }
+tokio-util = { workspace = true }
+
+# Error handling
+thiserror = { workspace = true }
+
+# Logging
+tracing = { workspace = true }
+
+[dev-dependencies]
+tokio = { workspace = true, features = ["test-util","macros","rt-multi-thread"] }
+
+[features]
+default = ["timeout", "lock", "deadlock", "backpressure", "scheduler"]
+
+# Feature modules
+timeout = []
+lock = []
+deadlock = []
+backpressure = []
+scheduler = []
+
+[package.metadata.docs.rs]
+all-features = true
+rustdoc-args = ["--cfg", "docsrs"]

crates/concurrency/src/backpressure.rs

@@ -0,0 +1,224 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+//! Backpressure management
+
+use rustfs_io_core::{BackpressureMonitor as CoreBackpressureMonitor, BackpressureState};
+use rustfs_io_metrics::backpressure_metrics;
+use std::sync::Arc;
+use std::time::Instant;
+use tokio::io::{DuplexStream, duplex};
+
+/// Backpressure configuration
+#[derive(Debug, Clone)]
+pub struct BackpressureConfig {
+    /// Buffer size in bytes
+    pub buffer_size: usize,
+    /// High watermark percentage
+    pub high_watermark: u32,
+    /// Low watermark percentage
+    pub low_watermark: u32,
+}
+
+impl Default for BackpressureConfig {
+    fn default() -> Self {
+        Self {
+            buffer_size: 4 * 1024 * 1024, // 4MB
+            high_watermark: 80,
+            low_watermark: 50,
+        }
+    }
+}
+
+impl BackpressureConfig {
+    /// Calculate high watermark threshold in bytes
+    pub fn high_watermark_bytes(&self) -> usize {
+        (self.buffer_size as u64 * self.high_watermark as u64 / 100) as usize
+    }
+
+    /// Calculate low watermark threshold in bytes
+    pub fn low_watermark_bytes(&self) -> usize {
+        (self.buffer_size as u64 * self.low_watermark as u64 / 100) as usize
+    }
+}
+
+/// Backpressure manager
+pub struct BackpressureManager {
+    config: BackpressureConfig,
+    monitor: Arc<CoreBackpressureMonitor>,
+}
+
+impl BackpressureManager {
+    /// Create a new backpressure manager
+    pub fn new(buffer_size: usize, high_watermark: u32, low_watermark: u32) -> Self {
+        let config = BackpressureConfig {
+            buffer_size,
+            high_watermark,
+            low_watermark,
+        };
+
+        let core_config = rustfs_io_core::BackpressureConfig {
+            max_concurrent: 32,
+            high_water_mark: high_watermark as f64 / 100.0,
+            low_water_mark: low_watermark as f64 / 100.0,
+            cooldown: std::time::Duration::from_millis(100),
+            enabled: true,
+        };
+
+        Self {
+            config,
+            monitor: Arc::new(CoreBackpressureMonitor::new(core_config)),
+        }
+    }
+
+    /// Get the configuration
+    pub fn config(&self) -> &BackpressureConfig {
+        &self.config
+    }
+
+    /// Get the monitor
+    pub fn monitor(&self) -> Arc<CoreBackpressureMonitor> {
+        self.monitor.clone()
+    }
+
+    /// Create a backpressure pipe
+    pub fn create_pipe(&self) -> BackpressurePipe {
+        BackpressurePipe::new(self.config.clone(), self.monitor.clone())
+    }
+
+    /// Get current state
+    pub fn state(&self) -> BackpressureState {
+        self.monitor.state()
+    }
+
+    /// Check if backpressure is active
+    pub fn is_active(&self) -> bool {
+        self.monitor.is_active()
+    }
+}
+
+/// Backpressure pipe wrapping tokio's duplex
+pub struct BackpressurePipe {
+    reader: DuplexStream,
+    writer: DuplexStream,
+    config: BackpressureConfig,
+    monitor: Arc<CoreBackpressureMonitor>,
+    created_at: Instant,
+}
+
+impl BackpressurePipe {
+    fn new(config: BackpressureConfig, monitor: Arc<CoreBackpressureMonitor>) -> Self {
+        let (reader, writer) = duplex(config.buffer_size);
+
+        Self {
+            reader,
+            writer,
+            config,
+            monitor,
+            created_at: Instant::now(),
+        }
+    }
+
+    /// Get the reader end
+    pub fn reader(&mut self) -> &mut DuplexStream {
+        &mut self.reader
+    }
+
+    /// Get the writer end
+    pub fn writer(&mut self) -> &mut DuplexStream {
+        &mut self.writer
+    }
+
+    /// Split into reader and writer
+    pub fn into_split(self) -> (DuplexStream, DuplexStream) {
+        (self.reader, self.writer)
+    }
+
+    /// Get the configuration
+    pub fn config(&self) -> &BackpressureConfig {
+        &self.config
+    }
+
+    /// Get current state
+    pub fn state(&self) -> BackpressureState {
+        self.monitor.state()
+    }
+
+    /// Get the age of this pipe
+    pub fn age(&self) -> std::time::Duration {
+        self.created_at.elapsed()
+    }
+
+    /// Check if should apply backpressure
+    pub fn should_apply_backpressure(&self) -> bool {
+        let should = self.monitor.should_apply_backpressure();
+        if should {
+            backpressure_metrics::record_backpressure_activation();
+        }
+        should
+    }
+}
+
+/// Backpressure event
+#[allow(dead_code)]
+#[derive(Debug, Clone)]
+pub struct BackpressureEvent {
+    /// Event timestamp
+    pub timestamp: Instant,
+    /// Event type
+    pub event_type: BackpressureEventType,
+    /// Buffer usage
+    pub buffer_usage: usize,
+    /// Buffer capacity
+    pub buffer_capacity: usize,
+}
+
+/// Backpressure event type
+#[allow(dead_code)]
+#[derive(Debug, Clone, Copy)]
+pub enum BackpressureEventType {
+    /// High watermark reached
+    HighWatermarkReached,
+    /// High watermark exited
+    HighWatermarkExited,
+    /// Backpressure applied
+    BackpressureApplied,
+    /// Backpressure released
+    BackpressureReleased,
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_backpressure_config() {
+        let config = BackpressureConfig::default();
+        assert_eq!(config.buffer_size, 4 * 1024 * 1024);
+        assert!(config.high_watermark > config.low_watermark);
+    }
+
+    #[test]
+    fn test_backpressure_manager() {
+        let manager = BackpressureManager::new(1024, 80, 50);
+        assert_eq!(manager.state(), BackpressureState::Normal);
+    }
+
+    #[test]
+    fn test_backpressure_pipe() {
+        let manager = BackpressureManager::new(1024, 80, 50);
+        let pipe = manager.create_pipe();
+        assert_eq!(pipe.state(), BackpressureState::Normal);
+    }
+}

crates/concurrency/src/config.rs

@@ -0,0 +1,256 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+//! Configuration for concurrency management
+
+use std::time::Duration;
+
+/// Feature flags for concurrency modules
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub struct ConcurrencyFeatures {
+    /// Enable timeout control
+    pub timeout: bool,
+    /// Enable lock optimization
+    pub lock: bool,
+    /// Enable deadlock detection
+    pub deadlock: bool,
+    /// Enable backpressure management
+    pub backpressure: bool,
+    /// Enable I/O scheduling
+    pub scheduler: bool,
+}
+
+impl Default for ConcurrencyFeatures {
+    fn default() -> Self {
+        Self {
+            timeout: cfg!(feature = "timeout"),
+            lock: cfg!(feature = "lock"),
+            deadlock: cfg!(feature = "deadlock"),
+            backpressure: cfg!(feature = "backpressure"),
+            scheduler: cfg!(feature = "scheduler"),
+        }
+    }
+}
+
+impl ConcurrencyFeatures {
+    /// Create with all features enabled
+    pub fn all() -> Self {
+        Self {
+            timeout: true,
+            lock: true,
+            deadlock: true,
+            backpressure: true,
+            scheduler: true,
+        }
+    }
+
+    /// Create with no features enabled
+    pub fn none() -> Self {
+        Self {
+            timeout: false,
+            lock: false,
+            deadlock: false,
+            backpressure: false,
+            scheduler: false,
+        }
+    }
+
+    /// Check if any feature is enabled
+    pub fn any_enabled(&self) -> bool {
+        self.timeout || self.lock || self.deadlock || self.backpressure || self.scheduler
+    }
+}
+
+/// Main configuration for concurrency management
+#[derive(Debug, Clone)]
+pub struct ConcurrencyConfig {
+    /// Feature flags
+    pub features: ConcurrencyFeatures,
+
+    // Timeout configuration
+    /// Default timeout duration
+    pub default_timeout: Duration,
+    /// Maximum timeout duration
+    pub max_timeout: Duration,
+    /// Enable dynamic timeout
+    pub enable_dynamic_timeout: bool,
+
+    // Lock configuration
+    /// Enable lock optimization
+    pub enable_lock_optimization: bool,
+    /// Lock acquisition timeout
+    pub lock_acquire_timeout: Duration,
+
+    // Deadlock configuration
+    /// Enable deadlock detection
+    pub enable_deadlock_detection: bool,
+    /// Deadlock check interval
+    pub deadlock_check_interval: Duration,
+    /// Hang threshold
+    pub hang_threshold: Duration,
+
+    // Backpressure configuration
+    /// Buffer size for backpressure
+    pub backpressure_buffer_size: usize,
+    /// High watermark percentage
+    pub high_watermark: u32,
+    /// Low watermark percentage
+    pub low_watermark: u32,
+
+    // Scheduler configuration
+    /// Base buffer size for I/O
+    pub io_buffer_size: usize,
+    /// Maximum buffer size
+    pub max_buffer_size: usize,
+    /// High priority size threshold
+    pub high_priority_threshold: usize,
+    /// Low priority size threshold
+    pub low_priority_threshold: usize,
+}
+
+impl Default for ConcurrencyConfig {
+    fn default() -> Self {
+        Self {
+            features: ConcurrencyFeatures::default(),
+
+            // Timeout defaults
+            default_timeout: Duration::from_secs(30),
+            max_timeout: Duration::from_secs(300),
+            enable_dynamic_timeout: true,
+
+            // Lock defaults
+            enable_lock_optimization: true,
+            lock_acquire_timeout: Duration::from_secs(5),
+
+            // Deadlock defaults
+            enable_deadlock_detection: false,
+            deadlock_check_interval: Duration::from_secs(10),
+            hang_threshold: Duration::from_secs(60),
+
+            // Backpressure defaults
+            backpressure_buffer_size: 4 * 1024 * 1024, // 4MB
+            high_watermark: 80,
+            low_watermark: 50,
+
+            // Scheduler defaults
+            io_buffer_size: 64 * 1024,                // 64KB
+            max_buffer_size: 4 * 1024 * 1024,         // 4MB
+            high_priority_threshold: 1024 * 1024,     // 1MB
+            low_priority_threshold: 10 * 1024 * 1024, // 10MB
+        }
+    }
+}
+
+impl ConcurrencyConfig {
+    /// Create configuration from environment variables
+    pub fn from_env() -> Self {
+        let mut config = Self::default();
+
+        // Read from environment if available
+        if let Ok(val) = std::env::var("RUSTFS_TIMEOUT_DEFAULT")
+            && let Ok(secs) = val.parse::<u64>()
+        {
+            config.default_timeout = Duration::from_secs(secs);
+        }
+
+        if let Ok(val) = std::env::var("RUSTFS_TIMEOUT_MAX")
+            && let Ok(secs) = val.parse::<u64>()
+        {
+            config.max_timeout = Duration::from_secs(secs);
+        }
+
+        if let Ok(val) = std::env::var("RUSTFS_BACKPRESSURE_BUFFER_SIZE")
+            && let Ok(size) = val.parse::<usize>()
+        {
+            config.backpressure_buffer_size = size;
+        }
+
+        if let Ok(val) = std::env::var("RUSTFS_IO_BUFFER_SIZE")
+            && let Ok(size) = val.parse::<usize>()
+        {
+            config.io_buffer_size = size;
+        }
+
+        config
+    }
+
+    /// Validate configuration
+    pub fn validate(&self) -> Result<(), ConfigError> {
+        if self.default_timeout > self.max_timeout {
+            return Err(ConfigError::InvalidTimeout("default_timeout cannot exceed max_timeout".to_string()));
+        }
+
+        if self.high_watermark <= self.low_watermark || self.high_watermark > 100 {
+            return Err(ConfigError::InvalidBackpressure(
+                "high_watermark must be > low_watermark and <= 100".to_string(),
+            ));
+        }
+
+        if self.io_buffer_size > self.max_buffer_size {
+            return Err(ConfigError::InvalidScheduler("io_buffer_size cannot exceed max_buffer_size".to_string()));
+        }
+
+        Ok(())
+    }
+}
+
+/// Configuration error
+#[allow(clippy::enum_variant_names)]
+#[derive(Debug, Clone, thiserror::Error)]
+pub enum ConfigError {
+    /// Invalid timeout configuration
+    #[error("Invalid timeout config: {0}")]
+    InvalidTimeout(String),
+
+    /// Invalid backpressure configuration
+    #[error("Invalid backpressure config: {0}")]
+    InvalidBackpressure(String),
+
+    /// Invalid scheduler configuration
+    #[error("Invalid scheduler config: {0}")]
+    InvalidScheduler(String),
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_default_config() {
+        let config = ConcurrencyConfig::default();
+        assert!(config.validate().is_ok());
+    }
+
+    #[test]
+    fn test_invalid_timeout() {
+        let config = ConcurrencyConfig {
+            default_timeout: Duration::from_secs(100),
+            max_timeout: Duration::from_secs(50),
+            ..Default::default()
+        };
+        assert!(
+            config.validate().is_err(),
+            "validate() should return an error when default_timeout > max_timeout"
+        );
+    }
+
+    #[test]
+    fn test_features() {
+        let features = ConcurrencyFeatures::all();
+        assert!(features.any_enabled());
+
+        let features = ConcurrencyFeatures::none();
+        assert!(!features.any_enabled());
+    }
+}

crates/concurrency/src/deadlock.rs

@@ -0,0 +1,207 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+//! Deadlock detection management
+
+use rustfs_io_core::{DeadlockDetector as CoreDeadlockDetector, LockType};
+use rustfs_io_metrics::deadlock_metrics;
+use std::collections::HashMap;
+use std::sync::Arc;
+use std::time::{Duration, Instant};
+
+/// Deadlock configuration
+#[derive(Debug, Clone)]
+pub struct DeadlockConfig {
+    /// Enable deadlock detection
+    pub enabled: bool,
+    /// Check interval
+    pub check_interval: Duration,
+    /// Hang threshold
+    pub hang_threshold: Duration,
+}
+
+impl Default for DeadlockConfig {
+    fn default() -> Self {
+        Self {
+            enabled: false,
+            check_interval: Duration::from_secs(10),
+            hang_threshold: Duration::from_secs(60),
+        }
+    }
+}
+
+/// Deadlock manager
+pub struct DeadlockManager {
+    config: DeadlockConfig,
+    detector: Arc<CoreDeadlockDetector>,
+    running: Arc<tokio::sync::Mutex<bool>>,
+}
+
+impl DeadlockManager {
+    /// Create a new deadlock manager
+    pub fn new(enabled: bool, check_interval: Duration, hang_threshold: Duration) -> Self {
+        let config = DeadlockConfig {
+            enabled,
+            check_interval,
+            hang_threshold,
+        };
+
+        let core_config = rustfs_io_core::DeadlockDetectorConfig {
+            enabled,
+            detection_interval: check_interval,
+            max_hold_time: hang_threshold,
+        };
+
+        Self {
+            config,
+            detector: Arc::new(CoreDeadlockDetector::new(core_config)),
+            running: Arc::new(tokio::sync::Mutex::new(false)),
+        }
+    }
+
+    /// Get the configuration
+    pub fn config(&self) -> &DeadlockConfig {
+        &self.config
+    }
+
+    /// Get the core detector
+    pub fn detector(&self) -> Arc<CoreDeadlockDetector> {
+        self.detector.clone()
+    }
+
+    /// Start the deadlock detection background task
+    pub async fn start(&self) {
+        if !self.config.enabled {
+            return;
+        }
+
+        let mut running = self.running.lock().await;
+        if *running {
+            return;
+        }
+        *running = true;
+        drop(running);
+
+        tracing::info!("Deadlock detection started");
+    }
+
+    /// Stop the deadlock detection
+    pub async fn stop(&self) {
+        let mut running = self.running.lock().await;
+        *running = false;
+
+        tracing::info!("Deadlock detection stopped");
+    }
+
+    /// Create a request tracker
+    pub fn track_request(&self, request_id: String, description: String) -> RequestTracker {
+        RequestTracker::new(request_id, description, self.detector.clone())
+    }
+
+    /// Register a lock
+    pub fn register_lock(&self, lock_type: LockType) -> u64 {
+        self.detector.register_lock(lock_type)
+    }
+
+    /// Unregister a lock
+    pub fn unregister_lock(&self, lock_id: u64) {
+        self.detector.unregister_lock(lock_id);
+    }
+
+    /// Detect deadlock
+    pub fn detect_deadlock(&self) -> Option<Vec<u64>> {
+        let result = self.detector.detect_deadlock();
+        if let Some(ref cycle) = result {
+            deadlock_metrics::record_deadlock_detected(cycle.len());
+        }
+        result
+    }
+}
+
+/// Request tracker for tracking resources
+pub struct RequestTracker {
+    request_id: String,
+    description: String,
+    start_time: Instant,
+    resources: HashMap<String, Vec<String>>,
+    detector: Arc<CoreDeadlockDetector>,
+}
+
+impl RequestTracker {
+    fn new(request_id: String, description: String, detector: Arc<CoreDeadlockDetector>) -> Self {
+        let start_time = Instant::now();
+        detector.register_request(&request_id, 1); // Use placeholder thread ID
+
+        Self {
+            request_id,
+            description,
+            start_time,
+            resources: HashMap::new(),
+            detector,
+        }
+    }
+
+    /// Get the request ID
+    pub fn request_id(&self) -> &str {
+        &self.request_id
+    }
+
+    /// Get the description
+    pub fn description(&self) -> &str {
+        &self.description
+    }
+
+    /// Get the elapsed time
+    pub fn elapsed(&self) -> Duration {
+        self.start_time.elapsed()
+    }
+
+    /// Record a lock acquisition
+    pub fn record_lock_acquire(&mut self, lock_id: u64, resource: String) {
+        self.resources.entry("locks".to_string()).or_default().push(resource);
+        self.detector.record_acquire(lock_id, 1); // Use placeholder thread ID
+        deadlock_metrics::record_lock_acquisition("read");
+    }
+
+    /// Record a lock release
+    pub fn record_lock_release(&mut self, lock_id: u64) {
+        self.detector.record_release(lock_id);
+    }
+}
+
+impl Drop for RequestTracker {
+    fn drop(&mut self) {
+        self.detector.unregister_request(&self.request_id);
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_deadlock_manager_creation() {
+        let manager = DeadlockManager::new(false, Duration::from_secs(10), Duration::from_secs(60));
+        assert!(!manager.config().enabled);
+    }
+
+    #[tokio::test]
+    async fn test_request_tracker() {
+        let manager = DeadlockManager::new(true, Duration::from_secs(10), Duration::from_secs(60));
+        let tracker = manager.track_request("req-1".to_string(), "test request".to_string());
+
+        assert_eq!(tracker.request_id(), "req-1");
+        assert_eq!(tracker.description(), "test request");
+    }
+}

crates/concurrency/src/lib.rs

@@ -0,0 +1,171 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+//! # RustFS Concurrency Management
+//!
+//! This crate provides comprehensive concurrency management for RustFS,
+//! including timeout control, lock optimization, deadlock detection,
+//! backpressure management, and I/O scheduling.
+//!
+//! ## Features
+//!
+//! All features are controlled by feature flags and can be enabled/disabled at compile time:
+//!
+//! - **timeout**: Dynamic timeout calculation based on data size and transfer rate
+//! - **lock**: Early lock release to reduce contention
+//! - **deadlock**: Request tracking and cycle detection
+//! - **backpressure**: Buffer-based flow control
+//! - **scheduler**: Adaptive buffer sizing and priority queuing
+//!
+//! ## Architecture
+//!
+//! ```text
+//! rustfs-concurrency (Business Layer)
+//!     ├── timeout (Timeout Control)
+//!     ├── lock (Lock Optimization)
+//!     ├── deadlock (Deadlock Detection)
+//!     ├── backpressure (Backpressure Management)
+//!     └── scheduler (I/O Scheduling)
+//!          │
+//!          ├── rustfs-io-core (Core Algorithms)
+//!          └── rustfs-io-metrics (Metrics Collection)
+//! ```
+//!
+//! ## Usage
+//!
+//! ```rust,no_run
+//! use rustfs_concurrency::{ConcurrencyConfig, ConcurrencyManager};
+//!
+//! # #[tokio::main]
+//! # async fn main() {
+//! // Create manager with all features enabled
+//! let config = ConcurrencyConfig::default();
+//! let manager = ConcurrencyManager::new(config);
+//!
+//! // Start services
+//! manager.start().await;
+//!
+//! // Use timeout control (if enabled)
+//! if manager.is_timeout_enabled() {
+//!     let timeout_manager = manager.timeout();
+//!     let _ = timeout_manager;
+//! }
+//!
+//! // Use lock optimization (if enabled)
+//! if manager.is_lock_enabled() {
+//!     let lock_manager = manager.lock();
+//!     let _ = lock_manager;
+//! }
+//!
+//! // Stop services
+//! manager.stop().await;
+//! # }
+//! ```
+
+#![deny(missing_docs)]
+#![deny(unsafe_code)]
+#![cfg_attr(docsrs, feature(doc_cfg))]
+
+// Re-export core types from io-core
+pub use rustfs_io_core::{
+    // Backpressure types
+    BackpressureConfig as CoreBackpressureConfig,
+    BackpressureMonitor as CoreBackpressureMonitor,
+    BackpressureState,
+
+    // Deadlock types
+    DeadlockDetector as CoreDeadlockDetector,
+    IoLoadLevel,
+    IoLoadMetrics,
+    IoPriority,
+    // Scheduler types
+    IoScheduler,
+    IoSchedulingContext,
+    LockInfo,
+    LockOptimizer as CoreLockOptimizer,
+
+    // Lock types
+    LockStats as CoreLockStats,
+    LockType,
+    // Timeout types
+    OperationProgress,
+    TimeoutError,
+    TimeoutStats,
+    WaitGraphEdge,
+
+    calculate_adaptive_timeout,
+    estimate_bytes_per_second,
+};
+
+// Module declarations with feature gates
+#[cfg(feature = "timeout")]
+mod timeout;
+
+#[cfg(feature = "lock")]
+mod lock;
+
+#[cfg(feature = "deadlock")]
+mod deadlock;
+
+#[cfg(feature = "backpressure")]
+mod backpressure;
+
+#[cfg(feature = "scheduler")]
+mod scheduler;
+
+// Public module exports with feature gates
+#[cfg(feature = "timeout")]
+pub use timeout::{TimeoutConfig, TimeoutGuard, TimeoutManager};
+
+#[cfg(feature = "lock")]
+pub use lock::{LockConfig, LockManager, LockScopeGuard, OptimizedLockGuard};
+
+#[cfg(feature = "deadlock")]
+pub use deadlock::{DeadlockConfig, DeadlockManager, RequestTracker};
+
+#[cfg(feature = "backpressure")]
+pub use backpressure::{BackpressureConfig, BackpressureManager, BackpressurePipe};
+
+#[cfg(feature = "scheduler")]
+pub use scheduler::{IoStrategy, SchedulerConfig, SchedulerManager};
+
+// Configuration
+mod config;
+pub use config::{ConcurrencyConfig, ConcurrencyFeatures};
+
+// Manager
+mod manager;
+pub use manager::{ConcurrencyManager, GetObjectQueueSnapshot};
+
+// Prelude for convenient imports
+pub mod prelude {
+    //! Prelude module for convenient imports
+
+    #[cfg(feature = "timeout")]
+    pub use crate::timeout::{TimeoutConfig, TimeoutGuard, TimeoutManager};
+
+    #[cfg(feature = "lock")]
+    pub use crate::lock::{LockConfig, LockManager, LockScopeGuard, OptimizedLockGuard};
+
+    #[cfg(feature = "deadlock")]
+    pub use crate::deadlock::{DeadlockConfig, DeadlockManager, RequestTracker};
+
+    #[cfg(feature = "backpressure")]
+    pub use crate::backpressure::{BackpressureConfig, BackpressureManager, BackpressurePipe};
+
+    #[cfg(feature = "scheduler")]
+    pub use crate::scheduler::{IoStrategy, SchedulerConfig, SchedulerManager};
+
+    pub use crate::{ConcurrencyConfig, ConcurrencyFeatures, ConcurrencyManager};
+}

crates/concurrency/src/lock.rs

@@ -0,0 +1,219 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+//! Lock optimization management
+
+use rustfs_io_core::{LockOptimizer as CoreLockOptimizer, LockStats};
+use rustfs_io_metrics::lock_metrics;
+use std::sync::Arc;
+use std::time::{Duration, Instant};
+
+/// Lock configuration
+#[derive(Debug, Clone)]
+pub struct LockConfig {
+    /// Enable lock optimization
+    pub enabled: bool,
+    /// Lock acquisition timeout
+    pub acquire_timeout: Duration,
+}
+
+impl Default for LockConfig {
+    fn default() -> Self {
+        Self {
+            enabled: true,
+            acquire_timeout: Duration::from_secs(5),
+        }
+    }
+}
+
+/// Lock manager
+pub struct LockManager {
+    config: LockConfig,
+    optimizer: Arc<CoreLockOptimizer>,
+}
+
+impl LockManager {
+    /// Create a new lock manager
+    pub fn new(enabled: bool, acquire_timeout: Duration) -> Self {
+        let config = LockConfig {
+            enabled,
+            acquire_timeout,
+        };
+
+        let core_config = rustfs_io_core::LockOptimizeConfig {
+            enabled,
+            acquire_timeout,
+            max_hold_time_warning: Duration::from_millis(100),
+            adaptive_spin: true,
+            max_spin_iterations: 1000,
+        };
+
+        Self {
+            config,
+            optimizer: Arc::new(CoreLockOptimizer::new(core_config)),
+        }
+    }
+
+    /// Get the configuration
+    pub fn config(&self) -> &LockConfig {
+        &self.config
+    }
+
+    /// Get the core optimizer
+    pub fn optimizer(&self) -> Arc<CoreLockOptimizer> {
+        self.optimizer.clone()
+    }
+
+    /// Get lock statistics
+    pub fn stats(&self) -> &LockStats {
+        self.optimizer.stats()
+    }
+
+    /// Optimize a lock guard
+    pub fn optimize<G>(&self, guard: G, resource: impl Into<String>) -> OptimizedLockGuard<G> {
+        OptimizedLockGuard::new(guard, resource, self.optimizer.clone())
+    }
+
+    /// Check if optimization is enabled
+    pub fn is_enabled(&self) -> bool {
+        self.config.enabled
+    }
+}
+
+/// Optimized lock guard with early release support
+pub struct OptimizedLockGuard<G> {
+    guard: Option<G>,
+    acquire_time: Instant,
+    released: bool,
+    resource: String,
+    optimizer: Arc<CoreLockOptimizer>,
+}
+
+impl<G> OptimizedLockGuard<G> {
+    fn new(guard: G, resource: impl Into<String>, optimizer: Arc<CoreLockOptimizer>) -> Self {
+        optimizer.on_acquire();
+        lock_metrics::record_lock_optimization_enabled(optimizer.config().enabled);
+
+        Self {
+            guard: Some(guard),
+            acquire_time: Instant::now(),
+            released: false,
+            resource: resource.into(),
+            optimizer,
+        }
+    }
+
+    /// Get the lock hold time
+    pub fn hold_time(&self) -> Duration {
+        self.acquire_time.elapsed()
+    }
+
+    /// Check if the lock has been released
+    pub fn is_released(&self) -> bool {
+        self.released
+    }
+
+    /// Release the lock early
+    pub fn early_release(&mut self) {
+        if self.released {
+            return;
+        }
+
+        let hold_time = self.hold_time();
+        self.guard.take();
+        self.released = true;
+
+        self.optimizer.on_release(hold_time);
+        lock_metrics::record_lock_hold_time(hold_time);
+
+        tracing::debug!(
+            resource = %self.resource,
+            hold_time_ms = hold_time.as_millis(),
+            "Lock released early (optimization active)"
+        );
+    }
+
+    /// Get a reference to the underlying guard
+    pub fn as_ref(&self) -> Option<&G> {
+        if self.released { None } else { self.guard.as_ref() }
+    }
+}
+
+impl<G> Drop for OptimizedLockGuard<G> {
+    fn drop(&mut self) {
+        if !self.released {
+            let hold_time = self.hold_time();
+            self.guard.take();
+            self.released = true;
+
+            self.optimizer.on_release(hold_time);
+            lock_metrics::record_lock_hold_time(hold_time);
+
+            tracing::debug!(
+                resource = %self.resource,
+                hold_time_ms = hold_time.as_millis(),
+                "Lock released on drop (normal release)"
+            );
+        }
+    }
+}
+
+/// Lock scope guard for RAII semantics
+pub struct LockScopeGuard<G> {
+    guard: Option<G>,
+}
+
+impl<G> LockScopeGuard<G> {
+    /// Create a new scope guard
+    pub fn new(guard: G) -> Self {
+        Self { guard: Some(guard) }
+    }
+
+    /// Get a reference to the guard
+    pub fn as_ref(&self) -> Option<&G> {
+        self.guard.as_ref()
+    }
+}
+
+impl<G> Drop for LockScopeGuard<G> {
+    fn drop(&mut self) {
+        self.guard.take();
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use std::sync::Mutex;
+
+    #[test]
+    fn test_lock_manager_creation() {
+        let manager = LockManager::new(true, Duration::from_secs(5));
+        assert!(manager.is_enabled());
+    }
+
+    #[test]
+    fn test_optimized_lock_guard() {
+        let manager = LockManager::new(true, Duration::from_secs(5));
+        let mutex = Mutex::new(42);
+        let guard = mutex.lock().unwrap();
+
+        let optimized = manager.optimize(guard, "test_resource");
+        assert!(!optimized.is_released());
+
+        let mut optimized = optimized;
+        optimized.early_release();
+        assert!(optimized.is_released());
+    }
+}

crates/concurrency/src/manager.rs

@@ -0,0 +1,315 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+//! Main concurrency manager
+
+use crate::config::ConcurrencyConfig;
+use std::sync::Arc;
+
+/// Snapshot of disk permit queue usage for GetObject orchestration.
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub struct GetObjectQueueSnapshot {
+    /// Total permits configured for disk reads.
+    pub total_permits: usize,
+    /// Permits currently in use.
+    pub permits_in_use: usize,
+}
+
+impl GetObjectQueueSnapshot {
+    /// Create a queue snapshot from total and available permits.
+    pub fn from_available_permits(total_permits: usize, available_permits: usize) -> Self {
+        Self {
+            total_permits,
+            permits_in_use: total_permits.saturating_sub(available_permits),
+        }
+    }
+
+    /// Return currently available permits.
+    pub fn permits_available(&self) -> usize {
+        self.total_permits.saturating_sub(self.permits_in_use)
+    }
+
+    /// Return queue utilization percentage in the 0-100 range.
+    pub fn utilization_percent(&self) -> f64 {
+        if self.total_permits == 0 {
+            0.0
+        } else {
+            (self.permits_in_use as f64 / self.total_permits as f64) * 100.0
+        }
+    }
+
+    /// Return whether the queue is considered congested.
+    pub fn is_congested(&self, threshold_percent: f64) -> bool {
+        self.utilization_percent() > threshold_percent
+    }
+}
+
+/// Main concurrency manager that provides access to all concurrency features
+pub struct ConcurrencyManager {
+    config: ConcurrencyConfig,
+
+    #[cfg(feature = "timeout")]
+    timeout: Arc<crate::timeout::TimeoutManager>,
+
+    #[cfg(feature = "lock")]
+    lock: Arc<crate::lock::LockManager>,
+
+    #[cfg(feature = "deadlock")]
+    deadlock: Arc<crate::deadlock::DeadlockManager>,
+
+    #[cfg(feature = "backpressure")]
+    backpressure: Arc<crate::backpressure::BackpressureManager>,
+
+    #[cfg(feature = "scheduler")]
+    scheduler: Arc<crate::scheduler::SchedulerManager>,
+}
+
+impl ConcurrencyManager {
+    /// Create a new concurrency manager with the given configuration
+    pub fn new(config: ConcurrencyConfig) -> Self {
+        // Validate configuration
+        if let Err(e) = config.validate() {
+            panic!("Invalid concurrency configuration: {}", e);
+        }
+
+        Self {
+            #[cfg(feature = "timeout")]
+            timeout: Arc::new(crate::timeout::TimeoutManager::new(
+                config.default_timeout,
+                config.max_timeout,
+                config.enable_dynamic_timeout,
+            )),
+
+            #[cfg(feature = "lock")]
+            lock: Arc::new(crate::lock::LockManager::new(
+                config.enable_lock_optimization,
+                config.lock_acquire_timeout,
+            )),
+
+            #[cfg(feature = "deadlock")]
+            deadlock: Arc::new(crate::deadlock::DeadlockManager::new(
+                config.enable_deadlock_detection,
+                config.deadlock_check_interval,
+                config.hang_threshold,
+            )),
+
+            #[cfg(feature = "backpressure")]
+            backpressure: Arc::new(crate::backpressure::BackpressureManager::new(
+                config.backpressure_buffer_size,
+                config.high_watermark,
+                config.low_watermark,
+            )),
+
+            #[cfg(feature = "scheduler")]
+            scheduler: Arc::new(crate::scheduler::SchedulerManager::new(
+                config.io_buffer_size,
+                config.max_buffer_size,
+                config.high_priority_threshold,
+                config.low_priority_threshold,
+            )),
+
+            config,
+        }
+    }
+
+    /// Create with default configuration
+    pub fn with_defaults() -> Self {
+        Self::new(ConcurrencyConfig::default())
+    }
+
+    /// Create from environment variables
+    pub fn from_env() -> Self {
+        Self::new(ConcurrencyConfig::from_env())
+    }
+
+    /// Get the configuration
+    pub fn config(&self) -> &ConcurrencyConfig {
+        &self.config
+    }
+
+    // ============================================
+    // Feature enable checks
+    // ============================================
+
+    /// Check if timeout feature is enabled
+    pub fn is_timeout_enabled(&self) -> bool {
+        #[cfg(feature = "timeout")]
+        {
+            self.config.features.timeout
+        }
+        #[cfg(not(feature = "timeout"))]
+        {
+            false
+        }
+    }
+
+    /// Check if lock feature is enabled
+    pub fn is_lock_enabled(&self) -> bool {
+        #[cfg(feature = "lock")]
+        {
+            self.config.features.lock
+        }
+        #[cfg(not(feature = "lock"))]
+        {
+            false
+        }
+    }
+
+    /// Check if deadlock feature is enabled
+    pub fn is_deadlock_enabled(&self) -> bool {
+        #[cfg(feature = "deadlock")]
+        {
+            self.config.features.deadlock
+        }
+        #[cfg(not(feature = "deadlock"))]
+        {
+            false
+        }
+    }
+
+    /// Check if backpressure feature is enabled
+    pub fn is_backpressure_enabled(&self) -> bool {
+        #[cfg(feature = "backpressure")]
+        {
+            self.config.features.backpressure
+        }
+        #[cfg(not(feature = "backpressure"))]
+        {
+            false
+        }
+    }
+
+    /// Check if scheduler feature is enabled
+    pub fn is_scheduler_enabled(&self) -> bool {
+        #[cfg(feature = "scheduler")]
+        {
+            self.config.features.scheduler
+        }
+        #[cfg(not(feature = "scheduler"))]
+        {
+            false
+        }
+    }
+
+    // ============================================
+    // Feature accessors
+    // ============================================
+
+    /// Get timeout manager
+    #[cfg(feature = "timeout")]
+    pub fn timeout(&self) -> Arc<crate::timeout::TimeoutManager> {
+        self.timeout.clone()
+    }
+
+    /// Get lock manager
+    #[cfg(feature = "lock")]
+    pub fn lock(&self) -> Arc<crate::lock::LockManager> {
+        self.lock.clone()
+    }
+
+    /// Get deadlock manager
+    #[cfg(feature = "deadlock")]
+    pub fn deadlock(&self) -> Arc<crate::deadlock::DeadlockManager> {
+        self.deadlock.clone()
+    }
+
+    /// Get backpressure manager
+    #[cfg(feature = "backpressure")]
+    pub fn backpressure(&self) -> Arc<crate::backpressure::BackpressureManager> {
+        self.backpressure.clone()
+    }
+
+    /// Get scheduler manager
+    #[cfg(feature = "scheduler")]
+    pub fn scheduler(&self) -> Arc<crate::scheduler::SchedulerManager> {
+        self.scheduler.clone()
+    }
+
+    // ============================================
+    // Lifecycle management
+    // ============================================
+
+    /// Start all enabled services (e.g., deadlock detection background task)
+    pub async fn start(&self) {
+        #[cfg(feature = "deadlock")]
+        {
+            if self.config.enable_deadlock_detection {
+                self.deadlock.start().await;
+            }
+        }
+
+        tracing::info!(
+            "Concurrency manager started (timeout={}, lock={}, deadlock={}, backpressure={}, scheduler={})",
+            self.is_timeout_enabled(),
+            self.is_lock_enabled(),
+            self.is_deadlock_enabled(),
+            self.is_backpressure_enabled(),
+            self.is_scheduler_enabled()
+        );
+    }
+
+    /// Stop all services
+    pub async fn stop(&self) {
+        #[cfg(feature = "deadlock")]
+        {
+            self.deadlock.stop().await;
+        }
+
+        tracing::info!("Concurrency manager stopped");
+    }
+}
+
+impl Default for ConcurrencyManager {
+    fn default() -> Self {
+        Self::with_defaults()
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_queue_snapshot() {
+        let snapshot = GetObjectQueueSnapshot::from_available_permits(64, 16);
+        assert_eq!(snapshot.permits_in_use, 48);
+        assert_eq!(snapshot.permits_available(), 16);
+        assert!(snapshot.is_congested(70.0));
+    }
+
+    #[test]
+    fn test_manager_creation() {
+        let manager = ConcurrencyManager::with_defaults();
+        assert!(manager.config().validate().is_ok());
+    }
+
+    #[tokio::test]
+    async fn test_manager_lifecycle() {
+        let manager = ConcurrencyManager::with_defaults();
+        manager.start().await;
+        manager.stop().await;
+    }
+
+    #[test]
+    fn test_feature_checks() {
+        let manager = ConcurrencyManager::with_defaults();
+
+        // These should return the feature flag status
+        let _ = manager.is_timeout_enabled();
+        let _ = manager.is_lock_enabled();
+        let _ = manager.is_deadlock_enabled();
+        let _ = manager.is_backpressure_enabled();
+        let _ = manager.is_scheduler_enabled();
+    }
+}

crates/concurrency/src/scheduler.rs

@@ -0,0 +1,225 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+//! I/O scheduler management
+
+use rustfs_io_core::{
+    IoLoadLevel, IoPriority, IoScheduler as CoreIoScheduler, IoSchedulingContext,
+    io_profile::{AccessPattern, StorageMedia},
+};
+use rustfs_io_metrics::io_metrics;
+use std::sync::Arc;
+use std::time::Duration;
+
+/// Scheduler configuration
+#[derive(Debug, Clone)]
+pub struct SchedulerConfig {
+    /// Base buffer size
+    pub base_buffer_size: usize,
+    /// Maximum buffer size
+    pub max_buffer_size: usize,
+    /// High priority threshold
+    pub high_priority_threshold: usize,
+    /// Low priority threshold
+    pub low_priority_threshold: usize,
+}
+
+impl Default for SchedulerConfig {
+    fn default() -> Self {
+        Self {
+            base_buffer_size: 64 * 1024,              // 64KB
+            max_buffer_size: 4 * 1024 * 1024,         // 4MB
+            high_priority_threshold: 1024 * 1024,     // 1MB
+            low_priority_threshold: 10 * 1024 * 1024, // 10MB
+        }
+    }
+}
+
+/// Scheduler manager
+pub struct SchedulerManager {
+    config: SchedulerConfig,
+    scheduler: Arc<CoreIoScheduler>,
+}
+
+impl SchedulerManager {
+    /// Create a new scheduler manager
+    pub fn new(
+        base_buffer_size: usize,
+        max_buffer_size: usize,
+        high_priority_threshold: usize,
+        low_priority_threshold: usize,
+    ) -> Self {
+        let config = SchedulerConfig {
+            base_buffer_size,
+            max_buffer_size,
+            high_priority_threshold,
+            low_priority_threshold,
+        };
+
+        let core_config = rustfs_io_core::IoSchedulerConfig::default();
+
+        Self {
+            config,
+            scheduler: Arc::new(CoreIoScheduler::new(core_config)),
+        }
+    }
+
+    /// Get the configuration
+    pub fn config(&self) -> &SchedulerConfig {
+        &self.config
+    }
+
+    /// Get the scheduler
+    pub fn scheduler(&self) -> Arc<CoreIoScheduler> {
+        self.scheduler.clone()
+    }
+
+    /// Create an I/O strategy
+    pub fn create_strategy(&self) -> IoStrategy {
+        IoStrategy::new(self.config.clone(), self.scheduler.clone())
+    }
+
+    /// Calculate buffer size
+    pub fn calculate_buffer_size(
+        &self,
+        file_size: i64,
+        media: StorageMedia,
+        pattern: AccessPattern,
+        load: IoLoadLevel,
+        concurrent: usize,
+    ) -> usize {
+        let strategy = self.create_strategy();
+        strategy.calculate_buffer_size(file_size, media, pattern, load, concurrent)
+    }
+
+    /// Get I/O priority
+    pub fn get_priority(&self, size: i64) -> IoPriority {
+        IoPriority::from_size(size, self.config.high_priority_threshold, self.config.low_priority_threshold)
+    }
+}
+
+/// I/O strategy
+pub struct IoStrategy {
+    config: SchedulerConfig,
+    scheduler: Arc<CoreIoScheduler>,
+}
+
+impl IoStrategy {
+    fn new(config: SchedulerConfig, scheduler: Arc<CoreIoScheduler>) -> Self {
+        Self { config, scheduler }
+    }
+
+    /// Calculate buffer size with multi-factor strategy
+    pub fn calculate_buffer_size(
+        &self,
+        file_size: i64,
+        media: StorageMedia,
+        pattern: AccessPattern,
+        load: IoLoadLevel,
+        concurrent: usize,
+    ) -> usize {
+        // Create scheduling context
+        let _ctx = IoSchedulingContext::new(file_size, self.config.base_buffer_size)
+            .with_sequential(matches!(pattern, AccessPattern::Sequential))
+            .with_media(media);
+
+        // Get base buffer size from core scheduler
+        let permit_wait = Duration::from_millis(10); // Default wait time
+        let is_sequential = matches!(pattern, AccessPattern::Sequential);
+        let core_strategy = self.scheduler.calculate_strategy(file_size, permit_wait, is_sequential);
+        let base_size = core_strategy.buffer_size;
+
+        // Apply multi-factor adjustments
+        let adjusted_size = self.apply_adjustments(base_size, media, pattern, load, concurrent);
+
+        // Record metrics
+        io_metrics::record_io_scheduler_decision(adjusted_size, load.as_str(), pattern.as_str());
+
+        adjusted_size.min(self.config.max_buffer_size)
+    }
+
+    fn apply_adjustments(
+        &self,
+        base_size: usize,
+        media: StorageMedia,
+        pattern: AccessPattern,
+        load: IoLoadLevel,
+        concurrent: usize,
+    ) -> usize {
+        let mut size = base_size;
+
+        // Media adjustment
+        size = match media {
+            StorageMedia::Nvme => (size as f64 * 1.5) as usize,
+            StorageMedia::Ssd => (size as f64 * 1.2) as usize,
+            StorageMedia::Hdd => size,
+            _ => size,
+        };
+
+        // Pattern adjustment
+        size = match pattern {
+            AccessPattern::Sequential => (size as f64 * 1.5) as usize,
+            AccessPattern::Random => (size as f64 * 0.5) as usize,
+            _ => size,
+        };
+
+        // Load adjustment
+        size = match load {
+            IoLoadLevel::Low => (size as f64 * 1.2) as usize,
+            IoLoadLevel::Medium => size,
+            IoLoadLevel::High => (size as f64 * 0.7) as usize,
+            IoLoadLevel::Critical => (size as f64 * 0.5) as usize,
+        };
+
+        // Concurrency adjustment
+        if concurrent > 10 {
+            size = (size as f64 * 0.8) as usize;
+        }
+
+        size
+    }
+
+    /// Get the configuration
+    pub fn config(&self) -> &SchedulerConfig {
+        &self.config
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_scheduler_config() {
+        let config = SchedulerConfig::default();
+        assert!(config.base_buffer_size < config.max_buffer_size);
+    }
+
+    #[test]
+    fn test_scheduler_manager() {
+        let manager = SchedulerManager::new(1024, 4096, 512, 2048);
+        let priority = manager.get_priority(100);
+        assert!(priority.is_high());
+    }
+
+    #[test]
+    fn test_io_strategy() {
+        let manager = SchedulerManager::new(1024, 4096, 512, 2048);
+        let strategy = manager.create_strategy();
+
+        let size = strategy.calculate_buffer_size(1024 * 1024, StorageMedia::Ssd, AccessPattern::Sequential, IoLoadLevel::Low, 1);
+
+        assert!(size > 0);
+    }
+}

crates/concurrency/src/timeout.rs

@@ -0,0 +1,150 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+//! Timeout management for operations
+
+use rustfs_io_core::{TimeoutError, calculate_adaptive_timeout};
+use std::time::{Duration, Instant};
+use tokio_util::sync::CancellationToken;
+
+/// Timeout configuration
+#[derive(Debug, Clone)]
+pub struct TimeoutConfig {
+    /// Default timeout duration
+    pub default_timeout: Duration,
+    /// Maximum timeout duration
+    pub max_timeout: Duration,
+    /// Enable dynamic timeout calculation
+    pub enable_dynamic: bool,
+}
+
+impl Default for TimeoutConfig {
+    fn default() -> Self {
+        Self {
+            default_timeout: Duration::from_secs(30),
+            max_timeout: Duration::from_secs(300),
+            enable_dynamic: true,
+        }
+    }
+}
+
+/// Timeout manager
+pub struct TimeoutManager {
+    config: TimeoutConfig,
+}
+
+impl TimeoutManager {
+    /// Create a new timeout manager
+    pub fn new(default_timeout: Duration, max_timeout: Duration, enable_dynamic: bool) -> Self {
+        Self {
+            config: TimeoutConfig {
+                default_timeout,
+                max_timeout,
+                enable_dynamic,
+            },
+        }
+    }
+
+    /// Get the configuration
+    pub fn config(&self) -> &TimeoutConfig {
+        &self.config
+    }
+
+    /// Calculate timeout for a given size
+    pub fn calculate_timeout(&self, size: u64, _history: &[Duration]) -> Duration {
+        if !self.config.enable_dynamic {
+            return self.config.default_timeout;
+        }
+
+        calculate_adaptive_timeout(self.config.default_timeout, None, 0, size).min(self.config.max_timeout)
+    }
+
+    /// Wrap an operation with timeout control
+    pub async fn wrap_operation<F, T, E>(&self, operation: F, timeout: Option<Duration>) -> Result<T, TimeoutError>
+    where
+        F: std::future::Future<Output = Result<T, E>>,
+        E: Into<TimeoutError>,
+    {
+        let timeout = timeout.unwrap_or(self.config.default_timeout);
+
+        match tokio::time::timeout(timeout, operation).await {
+            Ok(Ok(result)) => Ok(result),
+            Ok(Err(e)) => Err(e.into()),
+            Err(_) => Err(TimeoutError::TimedOut(timeout)),
+        }
+    }
+
+    /// Create a timeout guard for manual timeout control
+    pub fn create_guard(&self, timeout: Option<Duration>) -> TimeoutGuard {
+        TimeoutGuard::new(timeout.unwrap_or(self.config.default_timeout))
+    }
+}
+
+/// Timeout guard for manual timeout control
+pub struct TimeoutGuard {
+    timeout: Duration,
+    start: Instant,
+    cancel_token: CancellationToken,
+}
+
+impl TimeoutGuard {
+    fn new(timeout: Duration) -> Self {
+        Self {
+            timeout,
+            start: Instant::now(),
+            cancel_token: CancellationToken::new(),
+        }
+    }
+
+    /// Check if timeout has elapsed
+    pub fn is_timed_out(&self) -> bool {
+        self.start.elapsed() > self.timeout
+    }
+
+    /// Get remaining time
+    pub fn remaining(&self) -> Duration {
+        self.timeout.saturating_sub(self.start.elapsed())
+    }
+
+    /// Get the cancellation token
+    pub fn cancel_token(&self) -> CancellationToken {
+        self.cancel_token.clone()
+    }
+
+    /// Cancel the operation
+    pub fn cancel(&self) {
+        self.cancel_token.cancel();
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_timeout_config() {
+        let config = TimeoutConfig::default();
+        assert!(config.default_timeout < config.max_timeout);
+    }
+
+    #[tokio::test]
+    async fn test_wrap_operation_success() {
+        let manager = TimeoutManager::new(Duration::from_secs(5), Duration::from_secs(10), true);
+
+        let result = manager.wrap_operation(async { Ok::<_, TimeoutError>(42) }, None).await;
+
+        assert!(result.is_ok());
+        assert_eq!(result.unwrap(), 42);
+    }
+}

crates/config/Cargo.toml

@@ -37,3 +37,6 @@ constants = ["dep:const-str"]
 notify = ["dep:const-str", "constants"]
 observability = ["constants"]
 opa = ["constants"]
+
+[lib]
+doctest = false

crates/config/README.md

@@ -53,8 +53,24 @@ Current guidance:
 
 ## Scanner environment aliases
 
+- `RUSTFS_SCANNER_SPEED` (canonical, also accepts `MINIO_SCANNER_SPEED`)
+- `RUSTFS_SCANNER_CYCLE` (canonical, also accepts `MINIO_SCANNER_CYCLE`)
 - `RUSTFS_SCANNER_START_DELAY_SECS` (canonical)
 - `RUSTFS_DATA_SCANNER_START_DELAY_SECS` (deprecated alias for compatibility)
+- `RUSTFS_SCANNER_IDLE_MODE` (canonical)
+- `RUSTFS_SCANNER_CACHE_SAVE_TIMEOUT_SECS` (canonical)
+
+## Drive timeout environment variables
+
+- `RUSTFS_DRIVE_METADATA_TIMEOUT_SECS`
+- `RUSTFS_DRIVE_DISK_INFO_TIMEOUT_SECS`
+- `RUSTFS_DRIVE_LIST_DIR_TIMEOUT_SECS`
+- `RUSTFS_DRIVE_WALKDIR_TIMEOUT_SECS`
+- `RUSTFS_DRIVE_WALKDIR_STALL_TIMEOUT_SECS`
+
+Legacy compatibility fallback:
+- `RUSTFS_DRIVE_MAX_TIMEOUT_DURATION`
+  This legacy variable is treated as a deprecated fallback for the operation-specific drive timeout variables above when a canonical variable is unset.
 
 ## 📄 License
 

crates/config/src/audit/kafka.rs

@@ -0,0 +1,53 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+// Kafka Environment Variables
+pub const ENV_AUDIT_KAFKA_ENABLE: &str = "RUSTFS_AUDIT_KAFKA_ENABLE";
+pub const ENV_AUDIT_KAFKA_BROKERS: &str = "RUSTFS_AUDIT_KAFKA_BROKERS";
+pub const ENV_AUDIT_KAFKA_TOPIC: &str = "RUSTFS_AUDIT_KAFKA_TOPIC";
+pub const ENV_AUDIT_KAFKA_ACKS: &str = "RUSTFS_AUDIT_KAFKA_ACKS";
+pub const ENV_AUDIT_KAFKA_TLS_ENABLE: &str = "RUSTFS_AUDIT_KAFKA_TLS_ENABLE";
+pub const ENV_AUDIT_KAFKA_TLS_CA: &str = "RUSTFS_AUDIT_KAFKA_TLS_CA";
+pub const ENV_AUDIT_KAFKA_TLS_CLIENT_CERT: &str = "RUSTFS_AUDIT_KAFKA_TLS_CLIENT_CERT";
+pub const ENV_AUDIT_KAFKA_TLS_CLIENT_KEY: &str = "RUSTFS_AUDIT_KAFKA_TLS_CLIENT_KEY";
+pub const ENV_AUDIT_KAFKA_QUEUE_DIR: &str = "RUSTFS_AUDIT_KAFKA_QUEUE_DIR";
+pub const ENV_AUDIT_KAFKA_QUEUE_LIMIT: &str = "RUSTFS_AUDIT_KAFKA_QUEUE_LIMIT";
+
+pub const ENV_AUDIT_KAFKA_KEYS: &[&str; 10] = &[
+    ENV_AUDIT_KAFKA_ENABLE,
+    ENV_AUDIT_KAFKA_BROKERS,
+    ENV_AUDIT_KAFKA_TOPIC,
+    ENV_AUDIT_KAFKA_ACKS,
+    ENV_AUDIT_KAFKA_TLS_ENABLE,
+    ENV_AUDIT_KAFKA_TLS_CA,
+    ENV_AUDIT_KAFKA_TLS_CLIENT_CERT,
+    ENV_AUDIT_KAFKA_TLS_CLIENT_KEY,
+    ENV_AUDIT_KAFKA_QUEUE_DIR,
+    ENV_AUDIT_KAFKA_QUEUE_LIMIT,
+];
+
+/// A list of all valid configuration keys for a Kafka audit target.
+pub const AUDIT_KAFKA_KEYS: &[&str] = &[
+    crate::ENABLE_KEY,
+    crate::KAFKA_BROKERS,
+    crate::KAFKA_TOPIC,
+    crate::KAFKA_ACKS,
+    crate::KAFKA_TLS_ENABLE,
+    crate::KAFKA_TLS_CA,
+    crate::KAFKA_TLS_CLIENT_CERT,
+    crate::KAFKA_TLS_CLIENT_KEY,
+    crate::KAFKA_QUEUE_DIR,
+    crate::KAFKA_QUEUE_LIMIT,
+    crate::COMMENT_KEY,
+];

crates/config/src/audit/mod.rs

@@ -16,10 +16,16 @@
 //! This module defines the configuration for audit systems, including
 //! webhook and MQTT audit-related settings.
 
+mod kafka;
 mod mqtt;
+mod nats;
+mod pulsar;
 mod webhook;
 
+pub use kafka::*;
 pub use mqtt::*;
+pub use nats::*;
+pub use pulsar::*;
 pub use webhook::*;
 
 use crate::DEFAULT_DELIMITER;
@@ -29,8 +35,17 @@ pub const AUDIT_PREFIX: &str = "audit";
 pub const AUDIT_ROUTE_PREFIX: &str = const_str::concat!(AUDIT_PREFIX, DEFAULT_DELIMITER);
 
 pub const AUDIT_WEBHOOK_SUB_SYS: &str = "audit_webhook";
+pub const AUDIT_KAFKA_SUB_SYS: &str = "audit_kafka";
 pub const AUDIT_MQTT_SUB_SYS: &str = "audit_mqtt";
+pub const AUDIT_NATS_SUB_SYS: &str = "audit_nats";
+pub const AUDIT_PULSAR_SUB_SYS: &str = "audit_pulsar";
 
 pub const AUDIT_STORE_EXTENSION: &str = ".audit";
 #[allow(dead_code)]
-pub const AUDIT_SUB_SYSTEMS: &[&str] = &[AUDIT_MQTT_SUB_SYS, AUDIT_WEBHOOK_SUB_SYS];
+pub const AUDIT_SUB_SYSTEMS: &[&str] = &[
+    AUDIT_KAFKA_SUB_SYS,
+    AUDIT_MQTT_SUB_SYS,
+    AUDIT_NATS_SUB_SYS,
+    AUDIT_PULSAR_SUB_SYS,
+    AUDIT_WEBHOOK_SUB_SYS,
+];

crates/config/src/audit/mqtt.rs

@@ -23,9 +23,15 @@ pub const ENV_AUDIT_MQTT_RECONNECT_INTERVAL: &str = "RUSTFS_AUDIT_MQTT_RECONNECT
 pub const ENV_AUDIT_MQTT_KEEP_ALIVE_INTERVAL: &str = "RUSTFS_AUDIT_MQTT_KEEP_ALIVE_INTERVAL";
 pub const ENV_AUDIT_MQTT_QUEUE_DIR: &str = "RUSTFS_AUDIT_MQTT_QUEUE_DIR";
 pub const ENV_AUDIT_MQTT_QUEUE_LIMIT: &str = "RUSTFS_AUDIT_MQTT_QUEUE_LIMIT";
+pub const ENV_AUDIT_MQTT_TLS_POLICY: &str = "RUSTFS_AUDIT_MQTT_TLS_POLICY";
+pub const ENV_AUDIT_MQTT_TLS_CA: &str = "RUSTFS_AUDIT_MQTT_TLS_CA";
+pub const ENV_AUDIT_MQTT_TLS_CLIENT_CERT: &str = "RUSTFS_AUDIT_MQTT_TLS_CLIENT_CERT";
+pub const ENV_AUDIT_MQTT_TLS_CLIENT_KEY: &str = "RUSTFS_AUDIT_MQTT_TLS_CLIENT_KEY";
+pub const ENV_AUDIT_MQTT_TLS_TRUST_LEAF_AS_CA: &str = "RUSTFS_AUDIT_MQTT_TLS_TRUST_LEAF_AS_CA";
+pub const ENV_AUDIT_MQTT_WS_PATH_ALLOWLIST: &str = "RUSTFS_AUDIT_MQTT_WS_PATH_ALLOWLIST";
 
 /// A list of all valid configuration keys for an MQTT target.
-pub const ENV_AUDIT_MQTT_KEYS: &[&str; 10] = &[
+pub const ENV_AUDIT_MQTT_KEYS: &[&str; 16] = &[
     ENV_AUDIT_MQTT_ENABLE,
     ENV_AUDIT_MQTT_BROKER,
     ENV_AUDIT_MQTT_TOPIC,
@@ -36,6 +42,12 @@ pub const ENV_AUDIT_MQTT_KEYS: &[&str; 10] = &[
     ENV_AUDIT_MQTT_KEEP_ALIVE_INTERVAL,
     ENV_AUDIT_MQTT_QUEUE_DIR,
     ENV_AUDIT_MQTT_QUEUE_LIMIT,
+    ENV_AUDIT_MQTT_TLS_POLICY,
+    ENV_AUDIT_MQTT_TLS_CA,
+    ENV_AUDIT_MQTT_TLS_CLIENT_CERT,
+    ENV_AUDIT_MQTT_TLS_CLIENT_KEY,
+    ENV_AUDIT_MQTT_TLS_TRUST_LEAF_AS_CA,
+    ENV_AUDIT_MQTT_WS_PATH_ALLOWLIST,
 ];
 
 /// A list of all valid configuration keys for an MQTT target.
@@ -50,5 +62,11 @@ pub const AUDIT_MQTT_KEYS: &[&str] = &[
     crate::MQTT_KEEP_ALIVE_INTERVAL,
     crate::MQTT_QUEUE_DIR,
     crate::MQTT_QUEUE_LIMIT,
+    crate::MQTT_TLS_POLICY,
+    crate::MQTT_TLS_CA,
+    crate::MQTT_TLS_CLIENT_CERT,
+    crate::MQTT_TLS_CLIENT_KEY,
+    crate::MQTT_TLS_TRUST_LEAF_AS_CA,
+    crate::MQTT_WS_PATH_ALLOWLIST,
     crate::COMMENT_KEY,
 ];

crates/config/src/audit/nats.rs

@@ -0,0 +1,60 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+pub const ENV_AUDIT_NATS_ENABLE: &str = "RUSTFS_AUDIT_NATS_ENABLE";
+pub const ENV_AUDIT_NATS_ADDRESS: &str = "RUSTFS_AUDIT_NATS_ADDRESS";
+pub const ENV_AUDIT_NATS_SUBJECT: &str = "RUSTFS_AUDIT_NATS_SUBJECT";
+pub const ENV_AUDIT_NATS_USERNAME: &str = "RUSTFS_AUDIT_NATS_USERNAME";
+pub const ENV_AUDIT_NATS_PASSWORD: &str = "RUSTFS_AUDIT_NATS_PASSWORD";
+pub const ENV_AUDIT_NATS_TOKEN: &str = "RUSTFS_AUDIT_NATS_TOKEN";
+pub const ENV_AUDIT_NATS_CREDENTIALS_FILE: &str = "RUSTFS_AUDIT_NATS_CREDENTIALS_FILE";
+pub const ENV_AUDIT_NATS_TLS_CA: &str = "RUSTFS_AUDIT_NATS_TLS_CA";
+pub const ENV_AUDIT_NATS_TLS_CLIENT_CERT: &str = "RUSTFS_AUDIT_NATS_TLS_CLIENT_CERT";
+pub const ENV_AUDIT_NATS_TLS_CLIENT_KEY: &str = "RUSTFS_AUDIT_NATS_TLS_CLIENT_KEY";
+pub const ENV_AUDIT_NATS_TLS_REQUIRED: &str = "RUSTFS_AUDIT_NATS_TLS_REQUIRED";
+pub const ENV_AUDIT_NATS_QUEUE_DIR: &str = "RUSTFS_AUDIT_NATS_QUEUE_DIR";
+pub const ENV_AUDIT_NATS_QUEUE_LIMIT: &str = "RUSTFS_AUDIT_NATS_QUEUE_LIMIT";
+
+pub const ENV_AUDIT_NATS_KEYS: &[&str; 13] = &[
+    ENV_AUDIT_NATS_ENABLE,
+    ENV_AUDIT_NATS_ADDRESS,
+    ENV_AUDIT_NATS_SUBJECT,
+    ENV_AUDIT_NATS_USERNAME,
+    ENV_AUDIT_NATS_PASSWORD,
+    ENV_AUDIT_NATS_TOKEN,
+    ENV_AUDIT_NATS_CREDENTIALS_FILE,
+    ENV_AUDIT_NATS_TLS_CA,
+    ENV_AUDIT_NATS_TLS_CLIENT_CERT,
+    ENV_AUDIT_NATS_TLS_CLIENT_KEY,
+    ENV_AUDIT_NATS_TLS_REQUIRED,
+    ENV_AUDIT_NATS_QUEUE_DIR,
+    ENV_AUDIT_NATS_QUEUE_LIMIT,
+];
+
+pub const AUDIT_NATS_KEYS: &[&str] = &[
+    crate::ENABLE_KEY,
+    crate::NATS_ADDRESS,
+    crate::NATS_SUBJECT,
+    crate::NATS_USERNAME,
+    crate::NATS_PASSWORD,
+    crate::NATS_TOKEN,
+    crate::NATS_CREDENTIALS_FILE,
+    crate::NATS_TLS_CA,
+    crate::NATS_TLS_CLIENT_CERT,
+    crate::NATS_TLS_CLIENT_KEY,
+    crate::NATS_TLS_REQUIRED,
+    crate::NATS_QUEUE_DIR,
+    crate::NATS_QUEUE_LIMIT,
+    crate::COMMENT_KEY,
+];

crates/config/src/audit/pulsar.rs

@@ -0,0 +1,54 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+pub const ENV_AUDIT_PULSAR_ENABLE: &str = "RUSTFS_AUDIT_PULSAR_ENABLE";
+pub const ENV_AUDIT_PULSAR_BROKER: &str = "RUSTFS_AUDIT_PULSAR_BROKER";
+pub const ENV_AUDIT_PULSAR_TOPIC: &str = "RUSTFS_AUDIT_PULSAR_TOPIC";
+pub const ENV_AUDIT_PULSAR_AUTH_TOKEN: &str = "RUSTFS_AUDIT_PULSAR_AUTH_TOKEN";
+pub const ENV_AUDIT_PULSAR_USERNAME: &str = "RUSTFS_AUDIT_PULSAR_USERNAME";
+pub const ENV_AUDIT_PULSAR_PASSWORD: &str = "RUSTFS_AUDIT_PULSAR_PASSWORD";
+pub const ENV_AUDIT_PULSAR_TLS_CA: &str = "RUSTFS_AUDIT_PULSAR_TLS_CA";
+pub const ENV_AUDIT_PULSAR_TLS_ALLOW_INSECURE: &str = "RUSTFS_AUDIT_PULSAR_TLS_ALLOW_INSECURE";
+pub const ENV_AUDIT_PULSAR_TLS_HOSTNAME_VERIFICATION: &str = "RUSTFS_AUDIT_PULSAR_TLS_HOSTNAME_VERIFICATION";
+pub const ENV_AUDIT_PULSAR_QUEUE_DIR: &str = "RUSTFS_AUDIT_PULSAR_QUEUE_DIR";
+pub const ENV_AUDIT_PULSAR_QUEUE_LIMIT: &str = "RUSTFS_AUDIT_PULSAR_QUEUE_LIMIT";
+
+pub const ENV_AUDIT_PULSAR_KEYS: &[&str; 11] = &[
+    ENV_AUDIT_PULSAR_ENABLE,
+    ENV_AUDIT_PULSAR_BROKER,
+    ENV_AUDIT_PULSAR_TOPIC,
+    ENV_AUDIT_PULSAR_AUTH_TOKEN,
+    ENV_AUDIT_PULSAR_USERNAME,
+    ENV_AUDIT_PULSAR_PASSWORD,
+    ENV_AUDIT_PULSAR_TLS_CA,
+    ENV_AUDIT_PULSAR_TLS_ALLOW_INSECURE,
+    ENV_AUDIT_PULSAR_TLS_HOSTNAME_VERIFICATION,
+    ENV_AUDIT_PULSAR_QUEUE_DIR,
+    ENV_AUDIT_PULSAR_QUEUE_LIMIT,
+];
+
+pub const AUDIT_PULSAR_KEYS: &[&str] = &[
+    crate::ENABLE_KEY,
+    crate::PULSAR_BROKER,
+    crate::PULSAR_TOPIC,
+    crate::PULSAR_AUTH_TOKEN,
+    crate::PULSAR_USERNAME,
+    crate::PULSAR_PASSWORD,
+    crate::PULSAR_TLS_CA,
+    crate::PULSAR_TLS_ALLOW_INSECURE,
+    crate::PULSAR_TLS_HOSTNAME_VERIFICATION,
+    crate::PULSAR_QUEUE_DIR,
+    crate::PULSAR_QUEUE_LIMIT,
+    crate::COMMENT_KEY,
+];

crates/config/src/constants/app.rs

@@ -131,9 +131,76 @@ pub const ENV_RUSTFS_ADDRESS: &str = "RUSTFS_ADDRESS";
 /// Environment variable for server volumes.
 pub const ENV_RUSTFS_VOLUMES: &str = "RUSTFS_VOLUMES";
 
+/// Environment variable to explicitly bypass local physical disk independence checks.
+pub const ENV_UNSAFE_BYPASS_DISK_CHECK: &str = "RUSTFS_UNSAFE_BYPASS_DISK_CHECK";
+
+/// Compatibility alias used by legacy MinIO CI pipelines.
+///
+/// RustFS keeps this alias for backward compatibility only. Prefer
+/// `ENV_UNSAFE_BYPASS_DISK_CHECK` for explicit bypass control.
+pub const ENV_MINIO_CI: &str = "MINIO_CI";
+
+/// Default flag value for bypassing local physical disk independence checks.
+pub const DEFAULT_UNSAFE_BYPASS_DISK_CHECK: bool = false;
+
+/// Environment variable for server access key.
+pub const ENV_RUSTFS_ACCESS_KEY: &str = "RUSTFS_ACCESS_KEY";
+
+/// Environment variable for server access key file.
+pub const ENV_RUSTFS_ACCESS_KEY_FILE: &str = "RUSTFS_ACCESS_KEY_FILE";
+
+/// Environment variable for server root user.
+pub const ENV_RUSTFS_ROOT_USER: &str = "RUSTFS_ROOT_USER";
+
+/// Environment variable for server secret key.
+pub const ENV_RUSTFS_SECRET_KEY: &str = "RUSTFS_SECRET_KEY";
+
+/// Environment variable for server secret key file.
+pub const ENV_RUSTFS_SECRET_KEY_FILE: &str = "RUSTFS_SECRET_KEY_FILE";
+
+/// Environment variable for server root password.
+pub const ENV_RUSTFS_ROOT_PASSWORD: &str = "RUSTFS_ROOT_PASSWORD";
+
+/// Environment variable for server OBS endpoint.
+pub const ENV_RUSTFS_OBS_ENDPOINT: &str = "RUSTFS_OBS_ENDPOINT";
+
+/// Environment variable for console server enable.
+pub const ENV_RUSTFS_CONSOLE_ENABLE: &str = "RUSTFS_CONSOLE_ENABLE";
+
+/// Environment variable for console server address.
+pub const ENV_RUSTFS_CONSOLE_ADDRESS: &str = "RUSTFS_CONSOLE_ADDRESS";
+
 /// Environment variable for server tls path.
 pub const ENV_RUSTFS_TLS_PATH: &str = "RUSTFS_TLS_PATH";
 
+/// Environment variable for server KMS enable.
+pub const ENV_RUSTFS_KMS_ENABLE: &str = "RUSTFS_KMS_ENABLE";
+
+/// Default KMS enable for server-side encryption
+/// This is the default value for enabling KMS encryption for server-side encryption.
+/// Default value: false
+pub const DEFAULT_KMS_ENABLE: bool = false;
+
+/// Environment variable for server KMS backend.
+pub const ENV_RUSTFS_KMS_BACKEND: &str = "RUSTFS_KMS_BACKEND";
+
+/// Environment variable for Vault Transit mount path.
+pub const ENV_RUSTFS_KMS_VAULT_MOUNT_PATH: &str = "RUSTFS_KMS_VAULT_MOUNT_PATH";
+
+/// Default KMS backend for server-side encryption
+/// This is the default KMS backend for server-side encryption.
+/// Default value: local
+pub const DEFAULT_KMS_BACKEND: &str = "local";
+
+/// Environment variable for selecting the buffer profile used for adaptive buffer sizing.
+pub const ENV_RUSTFS_BUFFER_PROFILE: &str = "RUSTFS_BUFFER_PROFILE";
+
+/// Default buffer profile for adaptive buffer sizing
+/// This is the default buffer profile for adaptive buffer sizing.
+/// It is used to identify the workload profile for adaptive buffer sizing.
+/// Default value: GeneralPurpose
+pub const DEFAULT_BUFFER_PROFILE: &str = "GeneralPurpose";
+
 /// Default value for the server TLS path if `ENV_RUSTFS_TLS_PATH` is not set.
 pub const DEFAULT_RUSTFS_TLS_PATH: &str = "";
 
@@ -229,9 +296,9 @@ pub const DEFAULT_OBS_LOGS_EXPORT_ENABLED: bool = true;
 
 /// Default profiling export enabled
 /// It is used to enable or disable exporting profiles
-/// Default value: true
+/// Default value: false
 /// Environment variable: RUSTFS_OBS_PROFILING_EXPORT_ENABLED
-pub const DEFAULT_OBS_PROFILING_EXPORT_ENABLED: bool = true;
+pub const DEFAULT_OBS_PROFILING_EXPORT_ENABLED: bool = false;
 
 /// Default log local logging enabled for rustfs
 /// This is the default log local logging enabled for rustfs.

crates/config/src/constants/capacity.rs

@@ -0,0 +1,164 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+//! Capacity calculation configuration constants
+
+// ============================================================================
+// Environment Variable Names
+// ============================================================================
+
+/// Environment variable for scheduled update interval
+pub const ENV_CAPACITY_SCHEDULED_INTERVAL: &str = "RUSTFS_CAPACITY_SCHEDULED_INTERVAL";
+
+/// Environment variable for write trigger delay
+pub const ENV_CAPACITY_WRITE_TRIGGER_DELAY: &str = "RUSTFS_CAPACITY_WRITE_TRIGGER_DELAY";
+
+/// Environment variable for write frequency threshold
+pub const ENV_CAPACITY_WRITE_FREQUENCY_THRESHOLD: &str = "RUSTFS_CAPACITY_WRITE_FREQUENCY_THRESHOLD";
+
+/// Environment variable for fast update threshold
+pub const ENV_CAPACITY_FAST_UPDATE_THRESHOLD: &str = "RUSTFS_CAPACITY_FAST_UPDATE_THRESHOLD";
+
+/// Environment variable for max files threshold
+pub const ENV_CAPACITY_MAX_FILES_THRESHOLD: &str = "RUSTFS_CAPACITY_MAX_FILES_THRESHOLD";
+
+/// Environment variable for statistics timeout
+pub const ENV_CAPACITY_STAT_TIMEOUT: &str = "RUSTFS_CAPACITY_STAT_TIMEOUT";
+
+/// Environment variable for sample rate
+pub const ENV_CAPACITY_SAMPLE_RATE: &str = "RUSTFS_CAPACITY_SAMPLE_RATE";
+
+/// Environment variable for metrics logging interval
+pub const ENV_CAPACITY_METRICS_INTERVAL: &str = "RUSTFS_CAPACITY_METRICS_INTERVAL";
+
+/// Environment variable for following symbolic links during capacity calculation
+pub const ENV_CAPACITY_FOLLOW_SYMLINKS: &str = "RUSTFS_CAPACITY_FOLLOW_SYMLINKS";
+
+/// Environment variable for maximum symlink follow depth
+pub const ENV_CAPACITY_MAX_SYMLINK_DEPTH: &str = "RUSTFS_CAPACITY_MAX_SYMLINK_DEPTH";
+
+/// Environment variable for enabling dynamic timeout calculation
+pub const ENV_CAPACITY_ENABLE_DYNAMIC_TIMEOUT: &str = "RUSTFS_CAPACITY_ENABLE_DYNAMIC_TIMEOUT";
+
+/// Environment variable for minimum capacity calculation timeout
+pub const ENV_CAPACITY_MIN_TIMEOUT: &str = "RUSTFS_CAPACITY_MIN_TIMEOUT";
+
+/// Environment variable for maximum capacity calculation timeout
+pub const ENV_CAPACITY_MAX_TIMEOUT: &str = "RUSTFS_CAPACITY_MAX_TIMEOUT";
+
+/// Environment variable for progress stall detection timeout
+pub const ENV_CAPACITY_STALL_TIMEOUT: &str = "RUSTFS_CAPACITY_STALL_TIMEOUT";
+
+// ============================================================================
+// Default Values
+// ============================================================================
+
+/// Scheduled update interval in seconds
+/// Default: 120 seconds (2 minutes)
+pub const DEFAULT_SCHEDULED_UPDATE_INTERVAL_SECS: u64 = 120;
+
+/// Write trigger delay in seconds
+/// Default: 5 seconds
+pub const DEFAULT_WRITE_TRIGGER_DELAY_SECS: u64 = 5;
+
+/// Write frequency threshold (writes per minute)
+/// Default: 5 writes/minute
+pub const DEFAULT_WRITE_FREQUENCY_THRESHOLD: usize = 5;
+
+/// Fast update threshold in seconds
+/// Default: 30 seconds
+pub const DEFAULT_FAST_UPDATE_THRESHOLD_SECS: u64 = 30;
+
+/// Maximum files threshold for sampling
+/// Default: 200,000 files
+pub const DEFAULT_MAX_FILES_THRESHOLD: usize = 200_000;
+
+/// Statistics timeout in seconds
+/// Default: 3 seconds
+pub const DEFAULT_STAT_TIMEOUT_SECS: u64 = 3;
+
+/// Sampling rate (1 in every N files)
+/// Default: 200
+pub const DEFAULT_SAMPLE_RATE: usize = 200;
+
+/// Capacity metrics logging interval in seconds
+/// Default: 600 seconds (10 minutes)
+pub const DEFAULT_CAPACITY_METRICS_INTERVAL_SECS: u64 = 600;
+
+/// Follow symbolic links during capacity calculation
+/// Default: false (disabled for safety)
+pub const DEFAULT_CAPACITY_FOLLOW_SYMLINKS: bool = false;
+
+/// Maximum symlink follow depth
+/// Default: 3 levels
+pub const DEFAULT_CAPACITY_MAX_SYMLINK_DEPTH: u8 = 3;
+
+/// Enable dynamic timeout calculation based on directory characteristics
+/// Default: true (enabled)
+pub const DEFAULT_CAPACITY_ENABLE_DYNAMIC_TIMEOUT: bool = true;
+
+/// Minimum capacity calculation timeout in seconds
+/// Default: 2 seconds
+pub const DEFAULT_CAPACITY_MIN_TIMEOUT_SECS: u64 = 2;
+
+/// Maximum capacity calculation timeout in seconds
+/// Default: 15 seconds
+pub const DEFAULT_CAPACITY_MAX_TIMEOUT_SECS: u64 = 15;
+
+/// Progress stall detection timeout in seconds
+/// Default: 20 seconds
+pub const DEFAULT_CAPACITY_STALL_TIMEOUT_SECS: u64 = 20;
+
+// ============================================================================
+// Tests
+// ============================================================================
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_env_var_names() {
+        assert_eq!(ENV_CAPACITY_SCHEDULED_INTERVAL, "RUSTFS_CAPACITY_SCHEDULED_INTERVAL");
+        assert_eq!(ENV_CAPACITY_WRITE_TRIGGER_DELAY, "RUSTFS_CAPACITY_WRITE_TRIGGER_DELAY");
+        assert_eq!(ENV_CAPACITY_WRITE_FREQUENCY_THRESHOLD, "RUSTFS_CAPACITY_WRITE_FREQUENCY_THRESHOLD");
+        assert_eq!(ENV_CAPACITY_FAST_UPDATE_THRESHOLD, "RUSTFS_CAPACITY_FAST_UPDATE_THRESHOLD");
+        assert_eq!(ENV_CAPACITY_MAX_FILES_THRESHOLD, "RUSTFS_CAPACITY_MAX_FILES_THRESHOLD");
+        assert_eq!(ENV_CAPACITY_STAT_TIMEOUT, "RUSTFS_CAPACITY_STAT_TIMEOUT");
+        assert_eq!(ENV_CAPACITY_SAMPLE_RATE, "RUSTFS_CAPACITY_SAMPLE_RATE");
+        assert_eq!(ENV_CAPACITY_METRICS_INTERVAL, "RUSTFS_CAPACITY_METRICS_INTERVAL");
+        assert_eq!(ENV_CAPACITY_FOLLOW_SYMLINKS, "RUSTFS_CAPACITY_FOLLOW_SYMLINKS");
+        assert_eq!(ENV_CAPACITY_MAX_SYMLINK_DEPTH, "RUSTFS_CAPACITY_MAX_SYMLINK_DEPTH");
+        assert_eq!(ENV_CAPACITY_ENABLE_DYNAMIC_TIMEOUT, "RUSTFS_CAPACITY_ENABLE_DYNAMIC_TIMEOUT");
+        assert_eq!(ENV_CAPACITY_MIN_TIMEOUT, "RUSTFS_CAPACITY_MIN_TIMEOUT");
+        assert_eq!(ENV_CAPACITY_MAX_TIMEOUT, "RUSTFS_CAPACITY_MAX_TIMEOUT");
+        assert_eq!(ENV_CAPACITY_STALL_TIMEOUT, "RUSTFS_CAPACITY_STALL_TIMEOUT");
+    }
+
+    #[test]
+    fn test_default_values() {
+        assert_eq!(DEFAULT_SCHEDULED_UPDATE_INTERVAL_SECS, 120);
+        assert_eq!(DEFAULT_WRITE_TRIGGER_DELAY_SECS, 5);
+        assert_eq!(DEFAULT_WRITE_FREQUENCY_THRESHOLD, 5);
+        assert_eq!(DEFAULT_FAST_UPDATE_THRESHOLD_SECS, 30);
+        assert_eq!(DEFAULT_MAX_FILES_THRESHOLD, 200_000);
+        assert_eq!(DEFAULT_STAT_TIMEOUT_SECS, 3);
+        assert_eq!(DEFAULT_SAMPLE_RATE, 200);
+        assert_eq!(DEFAULT_CAPACITY_METRICS_INTERVAL_SECS, 600);
+        assert_eq!(DEFAULT_CAPACITY_MAX_SYMLINK_DEPTH, 3);
+        assert_eq!(DEFAULT_CAPACITY_MIN_TIMEOUT_SECS, 2);
+        assert_eq!(DEFAULT_CAPACITY_MAX_TIMEOUT_SECS, 15);
+        assert_eq!(DEFAULT_CAPACITY_STALL_TIMEOUT_SECS, 20);
+    }
+}

crates/config/src/constants/drive.rs

@@ -0,0 +1,68 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+/// Legacy global drive timeout fallback.
+/// Deprecated in favor of per-operation drive timeout knobs.
+pub const ENV_DRIVE_MAX_TIMEOUT_DURATION: &str = "RUSTFS_DRIVE_MAX_TIMEOUT_DURATION";
+
+/// Default timeout in seconds for the legacy global drive timeout fallback.
+pub const DEFAULT_DRIVE_MAX_TIMEOUT_DURATION_SECS: u64 = 30;
+
+/// Timeout for metadata-oriented drive operations such as `read_metadata`.
+pub const ENV_DRIVE_METADATA_TIMEOUT_SECS: &str = "RUSTFS_DRIVE_METADATA_TIMEOUT_SECS";
+pub const DEFAULT_DRIVE_METADATA_TIMEOUT_SECS: u64 = 5;
+
+/// Timeout for `disk_info()` calls on local and remote drives.
+pub const ENV_DRIVE_DISK_INFO_TIMEOUT_SECS: &str = "RUSTFS_DRIVE_DISK_INFO_TIMEOUT_SECS";
+pub const DEFAULT_DRIVE_DISK_INFO_TIMEOUT_SECS: u64 = 5;
+
+/// Timeout for `list_dir()` style metadata listing operations.
+pub const ENV_DRIVE_LIST_DIR_TIMEOUT_SECS: &str = "RUSTFS_DRIVE_LIST_DIR_TIMEOUT_SECS";
+pub const DEFAULT_DRIVE_LIST_DIR_TIMEOUT_SECS: u64 = 5;
+
+/// Total timeout for `walk_dir()` operations.
+pub const ENV_DRIVE_WALKDIR_TIMEOUT_SECS: &str = "RUSTFS_DRIVE_WALKDIR_TIMEOUT_SECS";
+pub const DEFAULT_DRIVE_WALKDIR_TIMEOUT_SECS: u64 = 5;
+
+/// Maximum time without forward progress while consuming a `walk_dir()` stream.
+pub const ENV_DRIVE_WALKDIR_STALL_TIMEOUT_SECS: &str = "RUSTFS_DRIVE_WALKDIR_STALL_TIMEOUT_SECS";
+pub const DEFAULT_DRIVE_WALKDIR_STALL_TIMEOUT_SECS: u64 = 5;
+
+/// Interval in seconds between active health probes for local and remote drives.
+pub const ENV_DRIVE_ACTIVE_CHECK_INTERVAL_SECS: &str = "RUSTFS_DRIVE_ACTIVE_CHECK_INTERVAL_SECS";
+pub const DEFAULT_DRIVE_ACTIVE_CHECK_INTERVAL_SECS: u64 = 15;
+
+/// Timeout in seconds for a single active health probe.
+pub const ENV_DRIVE_ACTIVE_CHECK_TIMEOUT_SECS: &str = "RUSTFS_DRIVE_ACTIVE_CHECK_TIMEOUT_SECS";
+pub const DEFAULT_DRIVE_ACTIVE_CHECK_TIMEOUT_SECS: u64 = 5;
+
+/// Number of consecutive failures before a suspect drive is classified as offline.
+pub const ENV_DRIVE_SUSPECT_FAILURE_THRESHOLD: &str = "RUSTFS_DRIVE_SUSPECT_FAILURE_THRESHOLD";
+pub const DEFAULT_DRIVE_SUSPECT_FAILURE_THRESHOLD: u64 = 2;
+
+/// Number of consecutive successful recovery probes before a returning drive is considered online again.
+pub const ENV_DRIVE_RETURNING_SUCCESS_THRESHOLD: &str = "RUSTFS_DRIVE_RETURNING_SUCCESS_THRESHOLD";
+pub const DEFAULT_DRIVE_RETURNING_SUCCESS_THRESHOLD: u64 = 3;
+
+/// Probe interval in seconds while a drive is in the recovery path.
+pub const ENV_DRIVE_RETURNING_PROBE_INTERVAL_SECS: &str = "RUSTFS_DRIVE_RETURNING_PROBE_INTERVAL_SECS";
+pub const DEFAULT_DRIVE_RETURNING_PROBE_INTERVAL_SECS: u64 = 2;
+
+/// Duration in seconds for classifying a recovered drive as a short offline event.
+pub const ENV_DRIVE_OFFLINE_GRACE_PERIOD_SECS: &str = "RUSTFS_DRIVE_OFFLINE_GRACE_PERIOD_SECS";
+pub const DEFAULT_DRIVE_OFFLINE_GRACE_PERIOD_SECS: u64 = 30;
+
+/// Duration in seconds after which a recovered drive is classified as long offline.
+pub const ENV_DRIVE_LONG_OFFLINE_THRESHOLD_SECS: &str = "RUSTFS_DRIVE_LONG_OFFLINE_THRESHOLD_SECS";
+pub const DEFAULT_DRIVE_LONG_OFFLINE_THRESHOLD_SECS: u64 = 172_800;

crates/config/src/constants/env.rs

@@ -26,6 +26,15 @@ pub const RUSTFS_WEBHOOK_SKIP_TLS_VERIFY_DEFAULT: bool = false;
 pub const ENABLE_KEY: &str = "enable";
 pub const COMMENT_KEY: &str = "comment";
 
+/// Global switch for enabling the audit module.
+pub const ENV_AUDIT_ENABLE: &str = "RUSTFS_AUDIT_ENABLE";
+/// Global switch for enabling the notify module.
+pub const ENV_NOTIFY_ENABLE: &str = "RUSTFS_NOTIFY_ENABLE";
+/// Default global audit switch (disabled by default).
+pub const DEFAULT_AUDIT_ENABLE: bool = false;
+/// Default global notify switch (disabled by default).
+pub const DEFAULT_NOTIFY_ENABLE: bool = false;
+
 /// Medium-drawn lines separator
 /// This is used to separate words in environment variable names.
 pub const ENV_WORD_DELIMITER_DASH: &str = "-";
@@ -290,4 +299,10 @@ mod tests {
             assert!(state.is_disabled());
         }
     }
+
+    #[test]
+    fn test_global_audit_notify_switch_constants() {
+        assert_eq!(ENV_AUDIT_ENABLE, "RUSTFS_AUDIT_ENABLE");
+        assert_eq!(ENV_NOTIFY_ENABLE, "RUSTFS_NOTIFY_ENABLE");
+    }
 }

crates/config/src/constants/heal.rs

@@ -56,6 +56,15 @@ pub const ENV_HEAL_TASK_TIMEOUT_SECS: &str = "RUSTFS_HEAL_TASK_TIMEOUT_SECS";
 /// - Note: A higher concurrency limit can speed up healing but may lead to resource contention.
 pub const ENV_HEAL_MAX_CONCURRENT_HEALS: &str = "RUSTFS_HEAL_MAX_CONCURRENT_HEALS";
 
+/// Environment variable name that specifies the maximum number of concurrent heal operations
+/// allowed for a single erasure set.
+///
+/// - Purpose: Prevent one degraded set from consuming all global heal slots.
+/// - Unit: number of operations (usize).
+/// - Valid values: any positive integer.
+/// - Example: `export RUSTFS_HEAL_MAX_CONCURRENT_PER_SET=1`
+pub const ENV_HEAL_MAX_CONCURRENT_PER_SET: &str = "RUSTFS_HEAL_MAX_CONCURRENT_PER_SET";
+
 /// Default value for enabling authentication for heal operations if not specified in the environment variable.
 /// - Value: true (authentication enabled).
 /// - Rationale: Enabling authentication by default enhances security for heal operations.
@@ -86,3 +95,46 @@ pub const DEFAULT_HEAL_TASK_TIMEOUT_SECS: u64 = 300; // 5 minutes
 /// - Rationale: This default concurrency limit helps balance healing speed with resource usage, preventing system overload.
 /// - Adjustments: Users may modify this value via the `RUSTFS_HEAL_MAX_CONCURRENT_HEALS` environment variable based on their system capacity and expected heal workload.
 pub const DEFAULT_HEAL_MAX_CONCURRENT_HEALS: usize = 4;
+
+/// Default maximum number of concurrent heal operations per erasure set.
+///
+/// - Value: 1 concurrent heal operation per set.
+/// - Rationale: Keeps a degraded set from monopolizing the global heal scheduler.
+pub const DEFAULT_HEAL_MAX_CONCURRENT_PER_SET: usize = 1;
+
+/// Environment variable that controls whether low-priority heal requests should merge into
+/// an existing queued request with the same deduplication key.
+pub const ENV_HEAL_LOW_PRIORITY_MERGE_ENABLE: &str = "RUSTFS_HEAL_LOW_PRIORITY_MERGE_ENABLE";
+
+/// Environment variable that allows low-priority heal requests to be dropped when the queue is full.
+pub const ENV_HEAL_LOW_PRIORITY_DROP_WHEN_FULL: &str = "RUSTFS_HEAL_LOW_PRIORITY_DROP_WHEN_FULL";
+
+/// Environment variable that controls concurrent object heals within a single erasure-set page.
+pub const ENV_HEAL_PAGE_OBJECT_CONCURRENCY: &str = "RUSTFS_HEAL_PAGE_OBJECT_CONCURRENCY";
+
+/// Environment variable that toggles notify-driven scheduler wakeups.
+pub const ENV_HEAL_EVENT_DRIVEN_SCHEDULER_ENABLE: &str = "RUSTFS_HEAL_EVENT_DRIVEN_SCHEDULER_ENABLE";
+
+/// Environment variable that toggles per-set bulkhead scheduling.
+pub const ENV_HEAL_SET_BULKHEAD_ENABLE: &str = "RUSTFS_HEAL_SET_BULKHEAD_ENABLE";
+
+/// Environment variable that toggles page-level parallel object healing for erasure-set repair.
+pub const ENV_HEAL_PAGE_PARALLEL_ENABLE: &str = "RUSTFS_HEAL_PAGE_PARALLEL_ENABLE";
+
+/// Default behavior is to merge duplicate low-priority requests.
+pub const DEFAULT_HEAL_LOW_PRIORITY_MERGE_ENABLE: bool = true;
+
+/// Default behavior is to drop low-priority requests instead of blocking when the queue is full.
+pub const DEFAULT_HEAL_LOW_PRIORITY_DROP_WHEN_FULL: bool = true;
+
+/// Default per-page object heal concurrency for erasure-set healing.
+pub const DEFAULT_HEAL_PAGE_OBJECT_CONCURRENCY: usize = 8;
+
+/// Default behavior is to keep notify-driven scheduler wakeups enabled.
+pub const DEFAULT_HEAL_EVENT_DRIVEN_SCHEDULER_ENABLE: bool = true;
+
+/// Default behavior is to keep per-set bulkhead scheduling enabled.
+pub const DEFAULT_HEAL_SET_BULKHEAD_ENABLE: bool = true;
+
+/// Default behavior is to keep erasure-set page parallelism enabled.
+pub const DEFAULT_HEAL_PAGE_PARALLEL_ENABLE: bool = true;

crates/config/src/constants/health.rs

@@ -0,0 +1,28 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+/// Enable or disable public `/health` and `/health/ready` endpoints.
+/// When disabled, the routes are not registered and return 404.
+pub const ENV_HEALTH_ENDPOINT_ENABLE: &str = "RUSTFS_HEALTH_ENDPOINT_ENABLE";
+pub const DEFAULT_HEALTH_ENDPOINT_ENABLE: bool = true;
+
+/// Cache TTL for storage readiness runtime-state evaluation (milliseconds).
+/// This reduces storage-layer pressure when probes are called at high frequency.
+pub const ENV_HEALTH_READINESS_CACHE_TTL_MS: &str = "RUSTFS_HEALTH_READINESS_CACHE_TTL_MS";
+pub const DEFAULT_HEALTH_READINESS_CACHE_TTL_MS: u64 = 1000;
+
+/// Enable minimal health payload mode for GET `/health*` responses.
+/// When enabled, only `status` and `ready` fields are returned.
+pub const ENV_HEALTH_MINIMAL_RESPONSE_ENABLE: &str = "RUSTFS_HEALTH_MINIMAL_RESPONSE_ENABLE";
+pub const DEFAULT_HEALTH_MINIMAL_RESPONSE_ENABLE: bool = false;

crates/config/src/constants/internode.rs

@@ -0,0 +1,62 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+/// Timeout for establishing a new internode gRPC connection.
+pub const ENV_INTERNODE_CONNECT_TIMEOUT_SECS: &str = "RUSTFS_INTERNODE_CONNECT_TIMEOUT_SECS";
+pub const DEFAULT_INTERNODE_CONNECT_TIMEOUT_SECS: u64 = 3;
+
+/// TCP keepalive interval for internode gRPC channels.
+pub const ENV_INTERNODE_TCP_KEEPALIVE_SECS: &str = "RUSTFS_INTERNODE_TCP_KEEPALIVE_SECS";
+pub const DEFAULT_INTERNODE_TCP_KEEPALIVE_SECS: u64 = 10;
+
+/// HTTP/2 keepalive interval for internode gRPC channels.
+pub const ENV_INTERNODE_HTTP2_KEEPALIVE_INTERVAL_SECS: &str = "RUSTFS_INTERNODE_HTTP2_KEEPALIVE_INTERVAL_SECS";
+pub const DEFAULT_INTERNODE_HTTP2_KEEPALIVE_INTERVAL_SECS: u64 = 5;
+
+/// HTTP/2 keepalive timeout for internode gRPC channels.
+pub const ENV_INTERNODE_HTTP2_KEEPALIVE_TIMEOUT_SECS: &str = "RUSTFS_INTERNODE_HTTP2_KEEPALIVE_TIMEOUT_SECS";
+pub const DEFAULT_INTERNODE_HTTP2_KEEPALIVE_TIMEOUT_SECS: u64 = 3;
+
+/// Overall timeout for a single internode gRPC request.
+pub const ENV_INTERNODE_RPC_TIMEOUT_SECS: &str = "RUSTFS_INTERNODE_RPC_TIMEOUT_SECS";
+pub const DEFAULT_INTERNODE_RPC_TIMEOUT_SECS: u64 = 10;
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn internode_timeout_defaults_stay_in_expected_bounds() {
+        assert_eq!(DEFAULT_INTERNODE_CONNECT_TIMEOUT_SECS, 3);
+        assert_eq!(DEFAULT_INTERNODE_TCP_KEEPALIVE_SECS, 10);
+        assert_eq!(DEFAULT_INTERNODE_HTTP2_KEEPALIVE_INTERVAL_SECS, 5);
+        assert_eq!(DEFAULT_INTERNODE_HTTP2_KEEPALIVE_TIMEOUT_SECS, 3);
+        assert_eq!(DEFAULT_INTERNODE_RPC_TIMEOUT_SECS, 10);
+    }
+
+    #[test]
+    fn internode_timeout_env_names_are_stable() {
+        assert_eq!(ENV_INTERNODE_CONNECT_TIMEOUT_SECS, "RUSTFS_INTERNODE_CONNECT_TIMEOUT_SECS");
+        assert_eq!(ENV_INTERNODE_TCP_KEEPALIVE_SECS, "RUSTFS_INTERNODE_TCP_KEEPALIVE_SECS");
+        assert_eq!(
+            ENV_INTERNODE_HTTP2_KEEPALIVE_INTERVAL_SECS,
+            "RUSTFS_INTERNODE_HTTP2_KEEPALIVE_INTERVAL_SECS"
+        );
+        assert_eq!(
+            ENV_INTERNODE_HTTP2_KEEPALIVE_TIMEOUT_SECS,
+            "RUSTFS_INTERNODE_HTTP2_KEEPALIVE_TIMEOUT_SECS"
+        );
+        assert_eq!(ENV_INTERNODE_RPC_TIMEOUT_SECS, "RUSTFS_INTERNODE_RPC_TIMEOUT_SECS");
+    }
+}

crates/config/src/constants/mod.rs

@@ -14,10 +14,14 @@
 
 pub(crate) mod app;
 pub(crate) mod body_limits;
+pub(crate) mod capacity;
 pub(crate) mod compress;
 pub(crate) mod console;
+pub(crate) mod drive;
 pub(crate) mod env;
 pub(crate) mod heal;
+pub(crate) mod health;
+pub(crate) mod internode;
 pub(crate) mod object;
 pub(crate) mod oidc;
 pub(crate) mod profiler;
@@ -28,3 +32,5 @@ pub(crate) mod runtime;
 pub(crate) mod scanner;
 pub(crate) mod targets;
 pub(crate) mod tls;
+pub(crate) mod workload;
+pub(crate) mod zero_copy;

crates/config/src/constants/object.rs

@@ -12,62 +12,6 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-/// Environment variable name to toggle object-level in-memory caching.
-///
-/// - Purpose: Enable or disable the object-level in-memory cache (moka).
-/// - Acceptable values: `"true"` / `"false"` (case-insensitive) or a boolean typed config.
-/// - Semantics: When enabled, the system keeps fully-read objects in memory to reduce backend requests; when disabled, reads bypass the object cache.
-/// - Example: `export RUSTFS_OBJECT_CACHE_ENABLE=true`
-/// - Note: Evaluate together with `RUSTFS_OBJECT_CACHE_CAPACITY_MB`, TTL/TTI and concurrency thresholds to balance memory usage and throughput.
-pub const ENV_OBJECT_CACHE_ENABLE: &str = "RUSTFS_OBJECT_CACHE_ENABLE";
-
-/// Environment variable name that specifies the object cache capacity in megabytes.
-///
-/// - Purpose: Set the maximum total capacity of the object cache (in MB).
-/// - Unit: MB (1 MB = 1_048_576 bytes).
-/// - Valid values: any positive integer (0 may indicate disabled or alternative handling).
-/// - Semantics: When the moka cache reaches this capacity, eviction policies will remove entries; tune according to available memory and object size distribution.
-/// - Example: `export RUSTFS_OBJECT_CACHE_CAPACITY_MB=512`
-/// - Note: Actual memory usage will be slightly higher due to object headers and indexing overhead.
-pub const ENV_OBJECT_CACHE_CAPACITY_MB: &str = "RUSTFS_OBJECT_CACHE_CAPACITY_MB";
-
-/// Environment variable name for maximum object size eligible for caching in megabytes.
-///
-/// - Purpose: Define the upper size limit for individual objects to be considered for caching.
-/// - Unit: MB (1 MB = 1_048_576 bytes).
-/// - Valid values: any positive integer; objects larger than this size will not be cached.
-/// - Semantics: Prevents caching of excessively large objects that could monopolize cache capacity; tune based on typical object size distribution.
-/// - Example: `export RUSTFS_OBJECT_CACHE_MAX_OBJECT_SIZE_MB=50`
-/// - Note: Setting this too low may reduce cache effectiveness; setting it too high may lead to inefficient memory usage.
-pub const ENV_OBJECT_CACHE_MAX_OBJECT_SIZE_MB: &str = "RUSTFS_OBJECT_CACHE_MAX_OBJECT_SIZE_MB";
-
-/// Environment variable name for object cache TTL (time-to-live) in seconds.
-///
-/// - Purpose: Specify the maximum lifetime of a cached entry from the moment it is written.
-/// - Unit: seconds (u64).
-/// - Semantics: TTL acts as a hard upper bound; entries older than TTL are considered expired and removed by periodic cleanup.
-/// - Example: `export RUSTFS_OBJECT_CACHE_TTL_SECS=300`
-/// - Note: TTL and TTI both apply; either policy can cause eviction.
-pub const ENV_OBJECT_CACHE_TTL_SECS: &str = "RUSTFS_OBJECT_CACHE_TTL_SECS";
-
-/// Environment variable name for object cache TTI (time-to-idle) in seconds.
-///
-/// - Purpose: Specify how long an entry may remain in cache without being accessed before it is evicted.
-/// - Unit: seconds (u64).
-/// - Semantics: TTI helps remove one-time or infrequently used entries; frequent accesses reset idle timers but do not extend beyond TTL unless additional logic exists.
-/// - Example: `export RUSTFS_OBJECT_CACHE_TTI_SECS=120`
-/// - Note: Works together with TTL to keep the cache populated with actively used objects.
-pub const ENV_OBJECT_CACHE_TTI_SECS: &str = "RUSTFS_OBJECT_CACHE_TTI_SECS";
-
-/// Environment variable name for threshold of "hot" object hit count used to extend life.
-///
-/// - Purpose: Define a hit-count threshold to mark objects as "hot" so they may be treated preferentially near expiration.
-/// - Valid values: positive integer (usize).
-/// - Semantics: Objects reaching this hit count can be considered for relaxed eviction to avoid thrashing hot items.
-/// - Example: `export RUSTFS_OBJECT_HOT_MIN_HITS_TO_EXTEND=5`
-/// - Note: This is an optional enhancement and requires cache-layer statistics and extension logic to take effect.
-pub const ENV_OBJECT_HOT_MIN_HITS_TO_EXTEND: &str = "RUSTFS_OBJECT_HOT_MIN_HITS_TO_EXTEND";
-
 /// Environment variable name for high concurrency threshold used in adaptive buffering.
 ///
 /// - Purpose: When concurrent request count exceeds this threshold, the system enters a "high concurrency" optimization mode to reduce per-request buffer sizes.
@@ -94,24 +38,6 @@ pub const ENV_OBJECT_MEDIUM_CONCURRENCY_THRESHOLD: &str = "RUSTFS_OBJECT_MEDIUM_
 /// - Note: This setting may interact with OS-level I/O scheduling and should be tuned based on hardware capabilities.
 pub const ENV_OBJECT_MAX_CONCURRENT_DISK_READS: &str = "RUSTFS_OBJECT_MAX_CONCURRENT_DISK_READS";
 
-/// Default: object caching is disabled.
-///
-/// - Semantics: Safe default to avoid unexpected memory usage or cache consistency concerns when not explicitly enabled.
-/// - Default is set to false (disabled).
-pub const DEFAULT_OBJECT_CACHE_ENABLE: bool = false;
-
-/// Default object cache capacity in MB.
-///
-/// - Default: 100 MB (can be overridden by `RUSTFS_OBJECT_CACHE_CAPACITY_MB`).
-/// - Note: Choose a conservative default to reduce memory pressure in development/testing.
-pub const DEFAULT_OBJECT_CACHE_CAPACITY_MB: u64 = 100;
-
-/// Default maximum object size eligible for caching in MB.
-///
-/// - Default: 10 MB (can be overridden by `RUSTFS_OBJECT_CACHE_MAX_OBJECT_SIZE_MB`).
-/// - Note: Balances caching effectiveness with memory usage.
-pub const DEFAULT_OBJECT_CACHE_MAX_OBJECT_SIZE_MB: usize = 10;
-
 /// Maximum concurrent requests before applying aggressive optimization.
 ///
 /// When concurrent requests exceed this threshold (>8), the system switches to
@@ -141,33 +67,6 @@ pub const DEFAULT_OBJECT_MEDIUM_CONCURRENCY_THRESHOLD: usize = 4;
 /// Default is set to 64 concurrent reads.
 pub const DEFAULT_OBJECT_MAX_CONCURRENT_DISK_READS: usize = 64;
 
-/// Time-to-live for cached objects (5 minutes = 300 seconds).
-///
-/// After this duration, cached objects are automatically expired by Moka's
-/// background cleanup process, even if they haven't been accessed. This prevents
-/// stale data from consuming cache capacity indefinitely.
-///
-/// Default is set to 300 seconds.
-pub const DEFAULT_OBJECT_CACHE_TTL_SECS: u64 = 300;
-
-/// Time-to-idle for cached objects (2 minutes = 120 seconds).
-///
-/// Objects that haven't been accessed for this duration are automatically evicted,
-/// even if their TTL hasn't expired. This ensures cache is populated with actively
-/// used objects and clears out one-time reads efficiently.
-///
-/// Default is set to 120 seconds.
-pub const DEFAULT_OBJECT_CACHE_TTI_SECS: u64 = 120;
-
-/// Minimum hit count to extend object lifetime beyond TTL.
-///
-/// "Hot" objects that have been accessed at least this many times are treated
-/// specially - they can survive longer in cache even as they approach TTL expiration.
-/// This prevents frequently accessed objects from being evicted prematurely.
-///
-/// Default is set to 5 hits.
-pub const DEFAULT_OBJECT_HOT_MIN_HITS_TO_EXTEND: usize = 5;
-
 /// Skip bitrot hash verification on GetObject reads.
 ///
 /// When enabled, GetObject reads skip the per-shard hash
@@ -180,3 +79,425 @@ pub const ENV_OBJECT_GET_SKIP_BITROT_VERIFY: &str = "RUSTFS_OBJECT_GET_SKIP_BITR
 
 /// Default: bitrot verification is enabled on GetObject reads (do not skip).
 pub const DEFAULT_OBJECT_GET_SKIP_BITROT_VERIFY: bool = false;
+
+// =============================================================================
+// Concurrent Request Fix - Timeout and Backpressure Configuration
+// =============================================================================
+
+/// Environment variable for GetObject request timeout in seconds.
+///
+/// When a GetObject request exceeds this duration, it will be cancelled
+/// and return a 504 Gateway Timeout error. This prevents requests from
+/// hanging indefinitely due to deadlocks or resource exhaustion.
+///
+/// Default: 30 seconds (can be overridden by `RUSTFS_OBJECT_GET_TIMEOUT`).
+/// Set to 0 to disable timeout (not recommended for production).
+pub const ENV_OBJECT_GET_TIMEOUT: &str = "RUSTFS_OBJECT_GET_TIMEOUT";
+
+/// Default GetObject request timeout in seconds.
+///
+/// This value balances between allowing large object transfers to complete
+/// and preventing indefinite hangs. For 20-26MB objects with concurrent
+/// range reads, 30 seconds should be sufficient under normal conditions.
+pub const DEFAULT_OBJECT_GET_TIMEOUT: u64 = 30;
+
+/// Environment variable for disk read operation timeout in seconds.
+///
+/// Individual disk read operations that exceed this duration will be
+/// cancelled and treated as failures. This helps detect slow or hung
+/// disks without waiting indefinitely.
+///
+/// Default: 10 seconds (can be overridden by `RUSTFS_OBJECT_DISK_READ_TIMEOUT`).
+pub const ENV_OBJECT_DISK_READ_TIMEOUT: &str = "RUSTFS_OBJECT_DISK_READ_TIMEOUT";
+
+/// Default disk read timeout in seconds.
+pub const DEFAULT_OBJECT_DISK_READ_TIMEOUT: u64 = 10;
+
+/// Environment variable for minimum GetObject timeout in seconds.
+///
+/// When dynamic timeout calculation is enabled, this is the minimum timeout
+/// that will be used regardless of object size. This prevents excessively
+/// short timeouts for very small objects.
+///
+/// Default: 5 seconds (can be overridden by `RUSTFS_OBJECT_MIN_TIMEOUT`).
+pub const ENV_OBJECT_MIN_TIMEOUT: &str = "RUSTFS_OBJECT_MIN_TIMEOUT";
+
+/// Default minimum GetObject timeout: 5 seconds.
+pub const DEFAULT_OBJECT_MIN_TIMEOUT: u64 = 5;
+
+/// Environment variable for maximum GetObject timeout in seconds.
+///
+/// When dynamic timeout calculation is enabled, this is the maximum timeout
+/// that will be used regardless of object size. This prevents excessively
+/// long timeouts for very large objects.
+///
+/// Default: 300 seconds (5 minutes, can be overridden by `RUSTFS_OBJECT_MAX_TIMEOUT`).
+pub const ENV_OBJECT_MAX_TIMEOUT: &str = "RUSTFS_OBJECT_MAX_TIMEOUT";
+
+/// Default maximum GetObject timeout: 300 seconds (5 minutes).
+pub const DEFAULT_OBJECT_MAX_TIMEOUT: u64 = 300;
+
+/// Environment variable for default bytes per second for timeout estimation.
+///
+/// This value is used to estimate timeout duration based on object size when
+/// dynamic timeout calculation is enabled. The timeout is calculated as:
+/// (object_size / bytes_per_second) * buffer_factor
+///
+/// Default: 1048576 (1 MB/s, can be overridden by `RUSTFS_OBJECT_BYTES_PER_SECOND`).
+pub const ENV_OBJECT_BYTES_PER_SECOND: &str = "RUSTFS_OBJECT_BYTES_PER_SECOND";
+
+/// Default bytes per second for timeout estimation: 1 MB/s.
+pub const DEFAULT_OBJECT_BYTES_PER_SECOND: u64 = 1024 * 1024;
+
+/// Environment variable to enable dynamic timeout calculation.
+///
+/// When enabled, timeout is calculated based on object size and transfer speed
+/// rather than using a fixed timeout value. This provides better timeout
+/// handling for objects of varying sizes.
+///
+/// Default: true (enabled, can be overridden by `RUSTFS_OBJECT_DYNAMIC_TIMEOUT_ENABLE`).
+pub const ENV_OBJECT_DYNAMIC_TIMEOUT_ENABLE: &str = "RUSTFS_OBJECT_DYNAMIC_TIMEOUT_ENABLE";
+
+/// Default: dynamic timeout calculation is enabled.
+pub const DEFAULT_OBJECT_DYNAMIC_TIMEOUT_ENABLE: bool = true;
+
+/// Environment variable for duplex pipe buffer size in bytes.
+///
+/// The duplex pipe connects the disk read task to the HTTP response stream.
+/// A larger buffer reduces backpressure but increases memory usage.
+/// For large objects (20-26MB), a 4MB buffer provides good throughput.
+///
+/// Default: 4194304 (4 MB, can be overridden by `RUSTFS_OBJECT_DUPLEX_BUFFER_SIZE`).
+/// Minimum recommended: 1048576 (1 MB).
+pub const ENV_OBJECT_DUPLEX_BUFFER_SIZE: &str = "RUSTFS_OBJECT_DUPLEX_BUFFER_SIZE";
+
+/// Default duplex buffer size: 4 MB.
+///
+/// This is 4x larger than the original 1 MB buffer, providing better
+/// handling of large objects and reducing backpressure-related hangs.
+pub const DEFAULT_OBJECT_DUPLEX_BUFFER_SIZE: usize = 4 * 1024 * 1024;
+
+/// Environment variable for I/O buffer size in bytes.
+///
+/// This controls the buffer size used for individual I/O operations.
+/// A larger buffer improves throughput for sequential reads but may
+/// increase latency for small random reads.
+///
+/// Default: 131072 (128 KB, can be overridden by `RUSTFS_OBJECT_IO_BUFFER_SIZE`).
+pub const ENV_OBJECT_IO_BUFFER_SIZE: &str = "RUSTFS_OBJECT_IO_BUFFER_SIZE";
+
+/// Default I/O buffer size: 128 KB.
+pub const DEFAULT_OBJECT_IO_BUFFER_SIZE: usize = 128 * 1024;
+
+/// Environment variable to enable/disable lock optimization.
+///
+/// When enabled, read locks are released immediately after metadata
+/// is read, rather than being held for the entire data transfer.
+/// This significantly reduces lock contention under high concurrency.
+///
+/// Default: true (enabled, can be overridden by `RUSTFS_OBJECT_LOCK_OPTIMIZATION_ENABLE`).
+pub const ENV_OBJECT_LOCK_OPTIMIZATION_ENABLE: &str = "RUSTFS_OBJECT_LOCK_OPTIMIZATION_ENABLE";
+
+/// Default: lock optimization is enabled.
+pub const DEFAULT_OBJECT_LOCK_OPTIMIZATION_ENABLE: bool = true;
+
+/// Environment variable to enable/disable priority-based I/O scheduling.
+///
+/// When enabled, smaller requests (< 1MB) are given higher priority
+/// than larger requests (> 10MB), preventing "starvation" of small
+/// requests by large ones.
+///
+/// Default: true (enabled, can be overridden by `RUSTFS_OBJECT_PRIORITY_SCHEDULING_ENABLE`).
+pub const ENV_OBJECT_PRIORITY_SCHEDULING_ENABLE: &str = "RUSTFS_OBJECT_PRIORITY_SCHEDULING_ENABLE";
+
+/// Default: priority scheduling is enabled.
+pub const DEFAULT_OBJECT_PRIORITY_SCHEDULING_ENABLE: bool = true;
+
+/// Environment variable to enable/disable deadlock detection.
+///
+/// When enabled, the system monitors active requests and detects
+/// potential deadlock situations (circular lock wait chains).
+/// This has some performance overhead and is intended for debugging.
+///
+/// Default: false (disabled, can be overridden by `RUSTFS_OBJECT_DEADLOCK_DETECTION_ENABLE`).
+pub const ENV_OBJECT_DEADLOCK_DETECTION_ENABLE: &str = "RUSTFS_OBJECT_DEADLOCK_DETECTION_ENABLE";
+
+/// Default: deadlock detection is disabled for performance.
+pub const DEFAULT_OBJECT_DEADLOCK_DETECTION_ENABLE: bool = false;
+
+/// Environment variable for deadlock detection check interval in seconds.
+///
+/// How often the deadlock detector analyzes the lock wait graph.
+/// More frequent checks detect deadlocks faster but use more CPU.
+///
+/// Default: 5 seconds (can be overridden by `RUSTFS_OBJECT_DEADLOCK_CHECK_INTERVAL`).
+pub const ENV_OBJECT_DEADLOCK_CHECK_INTERVAL: &str = "RUSTFS_OBJECT_DEADLOCK_CHECK_INTERVAL";
+
+/// Default deadlock check interval: 5 seconds.
+pub const DEFAULT_OBJECT_DEADLOCK_CHECK_INTERVAL: u64 = 5;
+
+/// Environment variable for deadlock detection hang threshold in seconds.
+///
+/// Requests that have been running longer than this threshold are
+/// considered "potentially hung" and included in deadlock analysis.
+///
+/// Default: 10 seconds (can be overridden by `RUSTFS_OBJECT_DEADLOCK_HANG_THRESHOLD`).
+pub const ENV_OBJECT_DEADLOCK_HANG_THRESHOLD: &str = "RUSTFS_OBJECT_DEADLOCK_HANG_THRESHOLD";
+
+/// Default hang threshold: 10 seconds.
+pub const DEFAULT_OBJECT_DEADLOCK_HANG_THRESHOLD: u64 = 10;
+
+/// Environment variable for backpressure high watermark percentage.
+///
+/// When buffer usage exceeds this percentage, the system enters
+/// "high watermark" state and may apply backpressure to producers.
+///
+/// Default: 80 (80%, can be overridden by `RUSTFS_OBJECT_BACKPRESSURE_HIGH_WATERMARK`).
+pub const ENV_OBJECT_BACKPRESSURE_HIGH_WATERMARK: &str = "RUSTFS_OBJECT_BACKPRESSURE_HIGH_WATERMARK";
+
+/// Default high watermark: 80%.
+pub const DEFAULT_OBJECT_BACKPRESSURE_HIGH_WATERMARK: u32 = 80;
+
+/// Environment variable for backpressure low watermark percentage.
+///
+/// When buffer usage drops below this percentage after being in
+/// high watermark state, backpressure is released.
+///
+/// Default: 50 (50%, can be overridden by `RUSTFS_OBJECT_BACKPRESSURE_LOW_WATERMARK`).
+pub const ENV_OBJECT_BACKPRESSURE_LOW_WATERMARK: &str = "RUSTFS_OBJECT_BACKPRESSURE_LOW_WATERMARK";
+
+/// Default low watermark: 50%.
+pub const DEFAULT_OBJECT_BACKPRESSURE_LOW_WATERMARK: u32 = 50;
+
+/// Environment variable for lock acquisition timeout in seconds.
+///
+/// When a lock cannot be acquired within this duration, the operation
+/// will fail with a timeout error. This prevents indefinite waiting
+/// for locks that may never be released due to deadlocks.
+///
+/// Default: 5 seconds (can be overridden by `RUSTFS_OBJECT_LOCK_ACQUIRE_TIMEOUT`).
+pub const ENV_OBJECT_LOCK_ACQUIRE_TIMEOUT: &str = "RUSTFS_OBJECT_LOCK_ACQUIRE_TIMEOUT";
+
+/// Default lock acquisition timeout: 5 seconds.
+pub const DEFAULT_OBJECT_LOCK_ACQUIRE_TIMEOUT: u64 = 5;
+
+// ============================================================================
+// I/O priority scheduling configuration
+// ============================================================================
+
+/// Environment variable for I/O high priority size threshold in bytes.
+///
+/// Requests smaller than this threshold are classified as high priority.
+/// High priority requests are processed first to prevent starvation of small requests.
+///
+/// Default: 1048576 (1 MB, can be overridden by `RUSTFS_OBJECT_IO_HIGH_PRIORITY_SIZE_THRESHOLD`).
+pub const ENV_OBJECT_IO_HIGH_PRIORITY_SIZE_THRESHOLD: &str = "RUSTFS_OBJECT_IO_HIGH_PRIORITY_SIZE_THRESHOLD";
+
+/// Default high priority size threshold: 1 MB.
+pub const DEFAULT_OBJECT_IO_HIGH_PRIORITY_SIZE_THRESHOLD: usize = 1024 * 1024;
+
+/// Environment variable for I/O low priority size threshold in bytes.
+///
+/// Requests larger than this threshold are classified as low priority.
+/// Low priority requests are processed last to avoid blocking small requests.
+///
+/// Default: 10485760 (10 MB, can be overridden by `RUSTFS_OBJECT_IO_LOW_PRIORITY_SIZE_THRESHOLD`).
+pub const ENV_OBJECT_IO_LOW_PRIORITY_SIZE_THRESHOLD: &str = "RUSTFS_OBJECT_IO_LOW_PRIORITY_SIZE_THRESHOLD";
+
+/// Default low priority size threshold: 10 MB.
+pub const DEFAULT_OBJECT_IO_LOW_PRIORITY_SIZE_THRESHOLD: usize = 10 * 1024 * 1024;
+
+/// Environment variable for high priority queue capacity.
+///
+/// Maximum number of requests that can be queued in the high priority queue.
+///
+/// Default: 32 (can be overridden by `RUSTFS_OBJECT_IO_QUEUE_HIGH_CAPACITY`).
+pub const ENV_OBJECT_IO_QUEUE_HIGH_CAPACITY: &str = "RUSTFS_OBJECT_IO_QUEUE_HIGH_CAPACITY";
+
+/// Default high priority queue capacity: 32.
+pub const DEFAULT_OBJECT_IO_QUEUE_HIGH_CAPACITY: usize = 32;
+
+/// Environment variable for normal priority queue capacity.
+///
+/// Maximum number of requests that can be queued in the normal priority queue.
+///
+/// Default: 64 (can be overridden by `RUSTFS_OBJECT_IO_QUEUE_NORMAL_CAPACITY`).
+pub const ENV_OBJECT_IO_QUEUE_NORMAL_CAPACITY: &str = "RUSTFS_OBJECT_IO_QUEUE_NORMAL_CAPACITY";
+
+/// Default normal priority queue capacity: 64.
+pub const DEFAULT_OBJECT_IO_QUEUE_NORMAL_CAPACITY: usize = 64;
+
+/// Environment variable for low priority queue capacity.
+///
+/// Maximum number of requests that can be queued in the low priority queue.
+///
+/// Default: 16 (can be overridden by `RUSTFS_OBJECT_IO_QUEUE_LOW_CAPACITY`).
+pub const ENV_OBJECT_IO_QUEUE_LOW_CAPACITY: &str = "RUSTFS_OBJECT_IO_QUEUE_LOW_CAPACITY";
+
+/// Default low priority queue capacity: 16.
+pub const DEFAULT_OBJECT_IO_QUEUE_LOW_CAPACITY: usize = 16;
+
+/// Environment variable for starvation prevention check interval in milliseconds.
+///
+/// How often the system checks for starving low-priority requests.
+/// When a low-priority request has been waiting longer than the starvation threshold,
+/// it is promoted to normal priority.
+///
+/// Default: 100 ms (can be overridden by `RUSTFS_OBJECT_IO_STARVATION_PREVENTION_INTERVAL`).
+pub const ENV_OBJECT_IO_STARVATION_PREVENTION_INTERVAL: &str = "RUSTFS_OBJECT_IO_STARVATION_PREVENTION_INTERVAL";
+
+/// Default starvation prevention interval: 100 ms.
+pub const DEFAULT_OBJECT_IO_STARVATION_PREVENTION_INTERVAL: u64 = 100;
+
+/// Environment variable for starvation threshold in seconds.
+///
+/// Maximum time a low-priority request can wait before being promoted to normal priority.
+/// This prevents indefinite starvation of low-priority requests.
+///
+/// Default: 5 seconds (can be overridden by `RUSTFS_OBJECT_IO_STARVATION_THRESHOLD_SECS`).
+pub const ENV_OBJECT_IO_STARVATION_THRESHOLD_SECS: &str = "RUSTFS_OBJECT_IO_STARVATION_THRESHOLD_SECS";
+
+/// Default starvation threshold: 5 seconds.
+pub const DEFAULT_OBJECT_IO_STARVATION_THRESHOLD_SECS: u64 = 5;
+
+/// Environment variable for load sampling window size.
+///
+/// Number of recent samples used to calculate I/O load metrics.
+///
+/// Default: 100 samples (can be overridden by `RUSTFS_OBJECT_IO_LOAD_SAMPLE_WINDOW`).
+pub const ENV_OBJECT_IO_LOAD_SAMPLE_WINDOW: &str = "RUSTFS_OBJECT_IO_LOAD_SAMPLE_WINDOW";
+
+/// Default load sampling window: 100 samples.
+pub const DEFAULT_OBJECT_IO_LOAD_SAMPLE_WINDOW: usize = 100;
+
+/// Environment variable for high load wait time threshold in milliseconds.
+///
+/// When average wait time exceeds this threshold, the system is considered to be under high load.
+///
+/// Default: 50 ms (can be overridden by `RUSTFS_OBJECT_IO_LOAD_HIGH_THRESHOLD_MS`).
+pub const ENV_OBJECT_IO_LOAD_HIGH_THRESHOLD_MS: &str = "RUSTFS_OBJECT_IO_LOAD_HIGH_THRESHOLD_MS";
+
+/// Default high load threshold: 50 ms.
+pub const DEFAULT_OBJECT_IO_LOAD_HIGH_THRESHOLD_MS: u64 = 50;
+
+/// Environment variable for low load wait time threshold in milliseconds.
+///
+/// When average wait time is below this threshold, the system is considered to be under low load.
+///
+/// Default: 10 ms (can be overridden by `RUSTFS_OBJECT_IO_LOAD_LOW_THRESHOLD_MS`).
+pub const ENV_OBJECT_IO_LOAD_LOW_THRESHOLD_MS: &str = "RUSTFS_OBJECT_IO_LOAD_LOW_THRESHOLD_MS";
+
+/// Default low load threshold: 10 ms.
+pub const DEFAULT_OBJECT_IO_LOAD_LOW_THRESHOLD_MS: u64 = 10;
+
+/// Environment variable for enabling storage media detection for adaptive I/O scheduling.
+///
+/// When disabled, the scheduler falls back to `Unknown` storage media unless an explicit
+/// override is provided.
+///
+/// Default: true (can be overridden by `RUSTFS_OBJECT_IO_STORAGE_DETECTION_ENABLE`).
+pub const ENV_OBJECT_IO_STORAGE_DETECTION_ENABLE: &str = "RUSTFS_OBJECT_IO_STORAGE_DETECTION_ENABLE";
+
+/// Default storage media detection setting: enabled.
+pub const DEFAULT_OBJECT_IO_STORAGE_DETECTION_ENABLE: bool = true;
+
+/// Environment variable for overriding detected storage media.
+///
+/// Supported values: `nvme`, `ssd`, `hdd`, `unknown`.
+/// Empty value means auto-detect or fallback to `Unknown`.
+///
+/// Default: empty string (can be overridden by `RUSTFS_OBJECT_IO_STORAGE_MEDIA_OVERRIDE`).
+pub const ENV_OBJECT_IO_STORAGE_MEDIA_OVERRIDE: &str = "RUSTFS_OBJECT_IO_STORAGE_MEDIA_OVERRIDE";
+
+/// Default storage media override: no override.
+pub const DEFAULT_OBJECT_IO_STORAGE_MEDIA_OVERRIDE: &str = "";
+
+/// Environment variable for access-pattern history size.
+///
+/// Controls how many recent offset/length observations are used to classify
+/// sequential, random, or mixed reads.
+///
+/// Default: 8 (can be overridden by `RUSTFS_OBJECT_IO_PATTERN_HISTORY_SIZE`).
+pub const ENV_OBJECT_IO_PATTERN_HISTORY_SIZE: &str = "RUSTFS_OBJECT_IO_PATTERN_HISTORY_SIZE";
+
+/// Default access-pattern history size: 8 samples.
+pub const DEFAULT_OBJECT_IO_PATTERN_HISTORY_SIZE: usize = 8;
+
+/// Environment variable for sequential access step tolerance in bytes.
+///
+/// Small gaps between adjacent reads within this tolerance are still treated as sequential.
+///
+/// Default: 131072 bytes (128 KiB, can be overridden by `RUSTFS_OBJECT_IO_SEQUENTIAL_STEP_TOLERANCE_BYTES`).
+pub const ENV_OBJECT_IO_SEQUENTIAL_STEP_TOLERANCE_BYTES: &str = "RUSTFS_OBJECT_IO_SEQUENTIAL_STEP_TOLERANCE_BYTES";
+
+/// Default sequential step tolerance: 128 KiB.
+pub const DEFAULT_OBJECT_IO_SEQUENTIAL_STEP_TOLERANCE_BYTES: u64 = 128 * 1024;
+
+/// Environment variable for bandwidth EMA beta.
+///
+/// Lower values react faster to recent throughput changes while higher values smooth
+/// short-term fluctuations more aggressively.
+///
+/// Default: 0.1 (can be overridden by `RUSTFS_OBJECT_IO_BANDWIDTH_EMA_BETA`).
+pub const ENV_OBJECT_IO_BANDWIDTH_EMA_BETA: &str = "RUSTFS_OBJECT_IO_BANDWIDTH_EMA_BETA";
+
+/// Default bandwidth EMA beta: 0.1.
+pub const DEFAULT_OBJECT_IO_BANDWIDTH_EMA_BETA: f64 = 0.1;
+
+/// Environment variable for the low bandwidth threshold in bytes per second.
+///
+/// Observed throughput below this value causes the scheduler to be more conservative
+/// with buffer growth and read-ahead.
+///
+/// Default: 67108864 bytes/sec (64 MiB/s, can be overridden by `RUSTFS_OBJECT_IO_BANDWIDTH_LOW_THRESHOLD_BPS`).
+pub const ENV_OBJECT_IO_BANDWIDTH_LOW_THRESHOLD_BPS: &str = "RUSTFS_OBJECT_IO_BANDWIDTH_LOW_THRESHOLD_BPS";
+
+/// Default low bandwidth threshold: 64 MiB/s.
+pub const DEFAULT_OBJECT_IO_BANDWIDTH_LOW_THRESHOLD_BPS: u64 = 64 * 1024 * 1024;
+
+/// Environment variable for the high bandwidth threshold in bytes per second.
+///
+/// Observed throughput above this value allows the scheduler to be more aggressive
+/// for sequential workloads.
+///
+/// Default: 536870912 bytes/sec (512 MiB/s, can be overridden by `RUSTFS_OBJECT_IO_BANDWIDTH_HIGH_THRESHOLD_BPS`).
+pub const ENV_OBJECT_IO_BANDWIDTH_HIGH_THRESHOLD_BPS: &str = "RUSTFS_OBJECT_IO_BANDWIDTH_HIGH_THRESHOLD_BPS";
+
+/// Default high bandwidth threshold: 512 MiB/s.
+pub const DEFAULT_OBJECT_IO_BANDWIDTH_HIGH_THRESHOLD_BPS: u64 = 512 * 1024 * 1024;
+
+/// Environment variable for NVMe buffer cap in bytes.
+///
+/// Sequential reads on NVMe can scale up to this buffer cap.
+///
+/// Default: 2097152 bytes (2 MiB, can be overridden by `RUSTFS_OBJECT_IO_NVME_BUFFER_CAP`).
+pub const ENV_OBJECT_IO_NVME_BUFFER_CAP: &str = "RUSTFS_OBJECT_IO_NVME_BUFFER_CAP";
+
+/// Default NVMe buffer cap: 2 MiB.
+pub const DEFAULT_OBJECT_IO_NVME_BUFFER_CAP: usize = 2 * 1024 * 1024;
+
+/// Environment variable for SSD buffer cap in bytes.
+///
+/// Default: 1048576 bytes (1 MiB, can be overridden by `RUSTFS_OBJECT_IO_SSD_BUFFER_CAP`).
+pub const ENV_OBJECT_IO_SSD_BUFFER_CAP: &str = "RUSTFS_OBJECT_IO_SSD_BUFFER_CAP";
+
+/// Default SSD buffer cap: 1 MiB.
+pub const DEFAULT_OBJECT_IO_SSD_BUFFER_CAP: usize = 1024 * 1024;
+
+/// Environment variable for HDD buffer cap in bytes.
+///
+/// Default: 524288 bytes (512 KiB, can be overridden by `RUSTFS_OBJECT_IO_HDD_BUFFER_CAP`).
+pub const ENV_OBJECT_IO_HDD_BUFFER_CAP: &str = "RUSTFS_OBJECT_IO_HDD_BUFFER_CAP";
+
+/// Default HDD buffer cap: 512 KiB.
+pub const DEFAULT_OBJECT_IO_HDD_BUFFER_CAP: usize = 512 * 1024;
+
+/// Environment variable for disabling read-ahead under random or mixed access with concurrency.
+///
+/// When concurrent requests reach this threshold, random-heavy workloads stop using read-ahead.
+///
+/// Default: 4 (can be overridden by `RUSTFS_OBJECT_IO_RANDOM_READAHEAD_DISABLE_CONCURRENCY`).
+pub const ENV_OBJECT_IO_RANDOM_READAHEAD_DISABLE_CONCURRENCY: &str = "RUSTFS_OBJECT_IO_RANDOM_READAHEAD_DISABLE_CONCURRENCY";
+
+/// Default read-ahead disable concurrency threshold: 4.
+pub const DEFAULT_OBJECT_IO_RANDOM_READAHEAD_DISABLE_CONCURRENCY: usize = 4;

crates/config/src/constants/oidc.rs

@@ -17,6 +17,7 @@ pub const OIDC_CONFIG_URL: &str = "config_url";
 pub const OIDC_CLIENT_ID: &str = "client_id";
 pub const OIDC_CLIENT_SECRET: &str = "client_secret";
 pub const OIDC_SCOPES: &str = "scopes";
+pub const OIDC_OTHER_AUDIENCES: &str = "other_audiences";
 pub const OIDC_REDIRECT_URI: &str = "redirect_uri";
 pub const OIDC_REDIRECT_URI_DYNAMIC: &str = "redirect_uri_dynamic";
 pub const OIDC_CLAIM_NAME: &str = "claim_name";
@@ -24,6 +25,7 @@ pub const OIDC_CLAIM_PREFIX: &str = "claim_prefix";
 pub const OIDC_ROLE_POLICY: &str = "role_policy";
 pub const OIDC_DISPLAY_NAME: &str = "display_name";
 pub const OIDC_GROUPS_CLAIM: &str = "groups_claim";
+pub const OIDC_ROLES_CLAIM: &str = "roles_claim";
 pub const OIDC_EMAIL_CLAIM: &str = "email_claim";
 pub const OIDC_USERNAME_CLAIM: &str = "username_claim";
 
@@ -33,6 +35,7 @@ pub const ENV_IDENTITY_OPENID_CONFIG_URL: &str = "RUSTFS_IDENTITY_OPENID_CONFIG_
 pub const ENV_IDENTITY_OPENID_CLIENT_ID: &str = "RUSTFS_IDENTITY_OPENID_CLIENT_ID";
 pub const ENV_IDENTITY_OPENID_CLIENT_SECRET: &str = "RUSTFS_IDENTITY_OPENID_CLIENT_SECRET";
 pub const ENV_IDENTITY_OPENID_SCOPES: &str = "RUSTFS_IDENTITY_OPENID_SCOPES";
+pub const ENV_IDENTITY_OPENID_OTHER_AUDIENCES: &str = "RUSTFS_IDENTITY_OPENID_OTHER_AUDIENCES";
 pub const ENV_IDENTITY_OPENID_REDIRECT_URI: &str = "RUSTFS_IDENTITY_OPENID_REDIRECT_URI";
 pub const ENV_IDENTITY_OPENID_REDIRECT_URI_DYNAMIC: &str = "RUSTFS_IDENTITY_OPENID_REDIRECT_URI_DYNAMIC";
 pub const ENV_IDENTITY_OPENID_CLAIM_NAME: &str = "RUSTFS_IDENTITY_OPENID_CLAIM_NAME";
@@ -40,16 +43,18 @@ pub const ENV_IDENTITY_OPENID_CLAIM_PREFIX: &str = "RUSTFS_IDENTITY_OPENID_CLAIM
 pub const ENV_IDENTITY_OPENID_ROLE_POLICY: &str = "RUSTFS_IDENTITY_OPENID_ROLE_POLICY";
 pub const ENV_IDENTITY_OPENID_DISPLAY_NAME: &str = "RUSTFS_IDENTITY_OPENID_DISPLAY_NAME";
 pub const ENV_IDENTITY_OPENID_GROUPS_CLAIM: &str = "RUSTFS_IDENTITY_OPENID_GROUPS_CLAIM";
+pub const ENV_IDENTITY_OPENID_ROLES_CLAIM: &str = "RUSTFS_IDENTITY_OPENID_ROLES_CLAIM";
 pub const ENV_IDENTITY_OPENID_EMAIL_CLAIM: &str = "RUSTFS_IDENTITY_OPENID_EMAIL_CLAIM";
 pub const ENV_IDENTITY_OPENID_USERNAME_CLAIM: &str = "RUSTFS_IDENTITY_OPENID_USERNAME_CLAIM";
 
 /// List of all environment variable keys for an OIDC provider.
-pub const ENV_IDENTITY_OPENID_KEYS: &[&str; 14] = &[
+pub const ENV_IDENTITY_OPENID_KEYS: &[&str; 16] = &[
     ENV_IDENTITY_OPENID_ENABLE,
     ENV_IDENTITY_OPENID_CONFIG_URL,
     ENV_IDENTITY_OPENID_CLIENT_ID,
     ENV_IDENTITY_OPENID_CLIENT_SECRET,
     ENV_IDENTITY_OPENID_SCOPES,
+    ENV_IDENTITY_OPENID_OTHER_AUDIENCES,
     ENV_IDENTITY_OPENID_REDIRECT_URI,
     ENV_IDENTITY_OPENID_REDIRECT_URI_DYNAMIC,
     ENV_IDENTITY_OPENID_CLAIM_NAME,
@@ -57,6 +62,7 @@ pub const ENV_IDENTITY_OPENID_KEYS: &[&str; 14] = &[
     ENV_IDENTITY_OPENID_ROLE_POLICY,
     ENV_IDENTITY_OPENID_DISPLAY_NAME,
     ENV_IDENTITY_OPENID_GROUPS_CLAIM,
+    ENV_IDENTITY_OPENID_ROLES_CLAIM,
     ENV_IDENTITY_OPENID_EMAIL_CLAIM,
     ENV_IDENTITY_OPENID_USERNAME_CLAIM,
 ];
@@ -68,6 +74,7 @@ pub const IDENTITY_OPENID_KEYS: &[&str] = &[
     OIDC_CLIENT_ID,
     OIDC_CLIENT_SECRET,
     OIDC_SCOPES,
+    OIDC_OTHER_AUDIENCES,
     OIDC_REDIRECT_URI,
     OIDC_REDIRECT_URI_DYNAMIC,
     OIDC_CLAIM_NAME,
@@ -75,6 +82,7 @@ pub const IDENTITY_OPENID_KEYS: &[&str] = &[
     OIDC_ROLE_POLICY,
     OIDC_DISPLAY_NAME,
     OIDC_GROUPS_CLAIM,
+    OIDC_ROLES_CLAIM,
     OIDC_EMAIL_CLAIM,
     OIDC_USERNAME_CLAIM,
     crate::COMMENT_KEY,
@@ -84,6 +92,8 @@ pub const IDENTITY_OPENID_KEYS: &[&str] = &[
 pub const OIDC_DEFAULT_SCOPES: &str = "openid,profile,email";
 pub const OIDC_DEFAULT_CLAIM_NAME: &str = "groups";
 pub const OIDC_DEFAULT_GROUPS_CLAIM: &str = "groups";
+/// Empty means do not merge a secondary claim into groups (legacy behavior). Set to e.g. `roles` to opt in.
+pub const OIDC_DEFAULT_ROLES_CLAIM: &str = "";
 pub const OIDC_DEFAULT_EMAIL_CLAIM: &str = "email";
 pub const OIDC_DEFAULT_USERNAME_CLAIM: &str = "preferred_username";
 

crates/config/src/constants/proxy.rs

@@ -20,6 +20,11 @@ pub const ENV_TRUSTED_PROXY_ENABLED: &str = "RUSTFS_TRUSTED_PROXY_ENABLED";
 /// Trusted proxy middleware is enabled by default.
 pub const DEFAULT_TRUSTED_PROXY_ENABLED: bool = true;
 
+/// Environment variable to select the trusted proxy implementation.
+pub const ENV_TRUSTED_PROXY_IMPLEMENTATION: &str = "RUSTFS_TRUSTED_PROXY_IMPLEMENTATION";
+/// The simplified implementation is used by default.
+pub const DEFAULT_TRUSTED_PROXY_IMPLEMENTATION: &str = "simple";
+
 /// Environment variable for the proxy validation mode.
 pub const ENV_TRUSTED_PROXY_VALIDATION_MODE: &str = "RUSTFS_TRUSTED_PROXY_VALIDATION_MODE";
 /// Default validation mode is "hop_by_hop".

crates/config/src/constants/runtime.rs

@@ -27,6 +27,16 @@ pub const ENV_RNG_SEED: &str = "RUSTFS_RUNTIME_RNG_SEED";
 /// Event polling interval
 pub const ENV_EVENT_INTERVAL: &str = "RUSTFS_RUNTIME_EVENT_INTERVAL";
 
+// Dial9 Tokio Telemetry Configuration
+pub const ENV_RUNTIME_DIAL9_ENABLED: &str = "RUSTFS_RUNTIME_DIAL9_ENABLED";
+pub const ENV_RUNTIME_DIAL9_OUTPUT_DIR: &str = "RUSTFS_RUNTIME_DIAL9_OUTPUT_DIR";
+pub const ENV_RUNTIME_DIAL9_FILE_PREFIX: &str = "RUSTFS_RUNTIME_DIAL9_FILE_PREFIX";
+pub const ENV_RUNTIME_DIAL9_MAX_FILE_SIZE: &str = "RUSTFS_RUNTIME_DIAL9_MAX_FILE_SIZE";
+pub const ENV_RUNTIME_DIAL9_ROTATION_COUNT: &str = "RUSTFS_RUNTIME_DIAL9_ROTATION_COUNT";
+pub const ENV_RUNTIME_DIAL9_S3_BUCKET: &str = "RUSTFS_RUNTIME_DIAL9_S3_BUCKET";
+pub const ENV_RUNTIME_DIAL9_S3_PREFIX: &str = "RUSTFS_RUNTIME_DIAL9_S3_PREFIX";
+pub const ENV_RUNTIME_DIAL9_SAMPLING_RATE: &str = "RUSTFS_RUNTIME_DIAL9_SAMPLING_RATE";
+
 // Default values for Tokio runtime
 pub const DEFAULT_WORKER_THREADS: usize = 16;
 pub const DEFAULT_MAX_BLOCKING_THREADS: usize = 1024;
@@ -40,9 +50,37 @@ pub const DEFAULT_MAX_IO_EVENTS_PER_TICK: usize = 1024;
 pub const DEFAULT_EVENT_INTERVAL: u32 = 61;
 pub const DEFAULT_RNG_SEED: Option<u64> = None; // None means random
 
+// Dial9 Tokio Telemetry Default values
+pub const DEFAULT_RUNTIME_DIAL9_ENABLED: bool = false; // Disabled by default
+pub const DEFAULT_RUNTIME_DIAL9_OUTPUT_DIR: &str = "/var/log/rustfs/telemetry";
+pub const DEFAULT_RUNTIME_DIAL9_FILE_PREFIX: &str = "rustfs-tokio";
+pub const DEFAULT_RUNTIME_DIAL9_MAX_FILE_SIZE: u64 = 100 * 1024 * 1024; // 100MB
+pub const DEFAULT_RUNTIME_DIAL9_ROTATION_COUNT: usize = 10;
+pub const DEFAULT_RUNTIME_DIAL9_SAMPLING_RATE: f64 = 1.0; // 100% sampling
+// Note: S3 bucket/prefix have no default; absence means upload is disabled (modeled as Option<String>)
+
+// Allocator reclaim configuration
+pub const ENV_ALLOCATOR_RECLAIM_ENABLED: &str = "RUSTFS_ALLOCATOR_RECLAIM_ENABLED";
+pub const ENV_ALLOCATOR_RECLAIM_INTERVAL_SECS: &str = "RUSTFS_ALLOCATOR_RECLAIM_INTERVAL_SECS";
+pub const ENV_ALLOCATOR_RECLAIM_FORCE: &str = "RUSTFS_ALLOCATOR_RECLAIM_FORCE";
+pub const ENV_ALLOCATOR_RECLAIM_IDLE_INTERVALS: &str = "RUSTFS_ALLOCATOR_RECLAIM_IDLE_INTERVALS";
+pub const DEFAULT_ALLOCATOR_RECLAIM_ENABLED: bool = false;
+pub const DEFAULT_ALLOCATOR_RECLAIM_INTERVAL_SECS: u64 = 30;
+pub const DEFAULT_ALLOCATOR_RECLAIM_FORCE: bool = true;
+pub const DEFAULT_ALLOCATOR_RECLAIM_IDLE_INTERVALS: u64 = 3;
+
+// File page-cache reclaim configuration
+pub const ENV_OBJECT_FILE_CACHE_RECLAIM_WRITE_ENABLE: &str = "RUSTFS_OBJECT_FILE_CACHE_RECLAIM_WRITE_ENABLE";
+pub const ENV_OBJECT_FILE_CACHE_RECLAIM_READ_ENABLE: &str = "RUSTFS_OBJECT_FILE_CACHE_RECLAIM_READ_ENABLE";
+pub const ENV_OBJECT_FILE_CACHE_RECLAIM_THRESHOLD: &str = "RUSTFS_OBJECT_FILE_CACHE_RECLAIM_THRESHOLD";
+pub const DEFAULT_OBJECT_FILE_CACHE_RECLAIM_WRITE_ENABLE: bool = false;
+pub const DEFAULT_OBJECT_FILE_CACHE_RECLAIM_READ_ENABLE: bool = false;
+pub const DEFAULT_OBJECT_FILE_CACHE_RECLAIM_THRESHOLD: usize = 4 * 1024 * 1024;
+
 /// Threshold for small object seek support in megabytes.
 ///
 /// When an object is smaller than this size, rustfs will provide seek support.
 ///
 /// Default is set to 10MB.
+pub const ENV_OBJECT_SEEK_SUPPORT_THRESHOLD: &str = "RUSTFS_OBJECT_SEEK_SUPPORT_THRESHOLD";
 pub const DEFAULT_OBJECT_SEEK_SUPPORT_THRESHOLD: usize = 10 * 1024 * 1024;

crates/config/src/constants/scanner.rs

@@ -25,6 +25,12 @@ pub const ENV_SCANNER_START_DELAY_SECS: &str = "RUSTFS_SCANNER_START_DELAY_SECS"
 #[deprecated(note = "Use RUSTFS_SCANNER_START_DELAY_SECS instead")]
 pub const ENV_DATA_SCANNER_START_DELAY_SECS: &str = "RUSTFS_DATA_SCANNER_START_DELAY_SECS";
 
+/// Environment variable that specifies the scanner cycle interval in seconds.
+/// If set, this overrides the cycle interval derived from `RUSTFS_SCANNER_SPEED`.
+/// - Unit: seconds (u64).
+/// - Example: `export RUSTFS_SCANNER_CYCLE=3600` (1 hour)
+pub const ENV_SCANNER_CYCLE: &str = "RUSTFS_SCANNER_CYCLE";
+
 /// Environment variable that selects the scanner speed preset.
 /// Valid values: `fastest`, `fast`, `default`, `slow`, `slowest`.
 /// Controls the sleep factor, maximum sleep duration, and cycle interval.
@@ -39,9 +45,24 @@ pub const DEFAULT_SCANNER_SPEED: &str = "default";
 /// - Example: `export RUSTFS_SCANNER_IDLE_MODE=false`
 pub const ENV_SCANNER_IDLE_MODE: &str = "RUSTFS_SCANNER_IDLE_MODE";
 
+/// Environment variable that controls scanner cache save timeout in seconds.
+/// The scanner enforces a minimum value of `1`.
+/// - Unit: seconds (u64).
+/// - Example: `export RUSTFS_SCANNER_CACHE_SAVE_TIMEOUT_SECS=30`
+pub const ENV_SCANNER_CACHE_SAVE_TIMEOUT_SECS: &str = "RUSTFS_SCANNER_CACHE_SAVE_TIMEOUT_SECS";
+
 /// Default scanner idle mode.
 pub const DEFAULT_SCANNER_IDLE_MODE: bool = true;
 
+/// Compatibility flag kept for Patch 3 rollback windows.
+///
+/// Inline scanner heal execution has been removed in favor of heal-candidate enqueue.
+/// When this flag is enabled, RustFS logs a warning and continues to use enqueue-based heal.
+pub const ENV_SCANNER_INLINE_HEAL_ENABLE: &str = "RUSTFS_SCANNER_INLINE_HEAL_ENABLE";
+
+/// Default inline scanner heal compatibility mode.
+pub const DEFAULT_SCANNER_INLINE_HEAL_ENABLE: bool = false;
+
 /// Scanner speed preset controlling throttling behavior.
 ///
 /// Each preset defines three parameters:
@@ -56,6 +77,8 @@ pub const DEFAULT_SCANNER_IDLE_MODE: bool = true;
 /// | `default` | 2x     | 1 second  | 1 minute       |
 /// | `slow`    | 10x    | 15 seconds| 1 minute       |
 /// | `slowest` | 100x   | 15 seconds| 30 minutes     |
+///
+/// The cycle interval can be overridden by `RUSTFS_SCANNER_CYCLE`.
 #[derive(Debug, Clone, Copy, PartialEq, Eq, Default)]
 pub enum ScannerSpeed {
     Fastest,

crates/config/src/constants/targets.rs

@@ -34,3 +34,49 @@ pub const MQTT_RECONNECT_INTERVAL: &str = "reconnect_interval";
 pub const MQTT_KEEP_ALIVE_INTERVAL: &str = "keep_alive_interval";
 pub const MQTT_QUEUE_DIR: &str = "queue_dir";
 pub const MQTT_QUEUE_LIMIT: &str = "queue_limit";
+pub const MQTT_TLS_POLICY: &str = "tls_policy";
+pub const MQTT_TLS_CA: &str = "tls_ca";
+pub const MQTT_TLS_CLIENT_CERT: &str = "tls_client_cert";
+pub const MQTT_TLS_CLIENT_KEY: &str = "tls_client_key";
+pub const MQTT_TLS_TRUST_LEAF_AS_CA: &str = "tls_trust_leaf_as_ca";
+pub const MQTT_WS_PATH_ALLOWLIST: &str = "ws_path_allowlist";
+pub const KAFKA_BROKERS: &str = "brokers";
+pub const KAFKA_TOPIC: &str = "topic";
+pub const KAFKA_ACKS: &str = "acks";
+pub const KAFKA_QUEUE_DIR: &str = "queue_dir";
+pub const KAFKA_QUEUE_LIMIT: &str = "queue_limit";
+pub const KAFKA_TLS_ENABLE: &str = "tls_enable";
+pub const KAFKA_TLS_CA: &str = "tls_ca";
+pub const KAFKA_TLS_CLIENT_CERT: &str = "tls_client_cert";
+pub const KAFKA_TLS_CLIENT_KEY: &str = "tls_client_key";
+
+pub const NATS_ADDRESS: &str = "address";
+pub const NATS_SUBJECT: &str = "subject";
+pub const NATS_USERNAME: &str = "username";
+pub const NATS_PASSWORD: &str = "password";
+pub const NATS_TOKEN: &str = "token";
+pub const NATS_CREDENTIALS_FILE: &str = "credentials_file";
+pub const NATS_TLS_CA: &str = "tls_ca";
+pub const NATS_TLS_CLIENT_CERT: &str = "tls_client_cert";
+pub const NATS_TLS_CLIENT_KEY: &str = "tls_client_key";
+pub const NATS_TLS_REQUIRED: &str = "tls_required";
+pub const NATS_QUEUE_DIR: &str = "queue_dir";
+pub const NATS_QUEUE_LIMIT: &str = "queue_limit";
+
+pub const PULSAR_BROKER: &str = "broker";
+pub const PULSAR_TOPIC: &str = "topic";
+pub const PULSAR_AUTH_TOKEN: &str = "auth_token";
+pub const PULSAR_USERNAME: &str = "username";
+pub const PULSAR_PASSWORD: &str = "password";
+pub const PULSAR_TLS_CA: &str = "tls_ca";
+pub const PULSAR_TLS_ALLOW_INSECURE: &str = "tls_allow_insecure";
+pub const PULSAR_TLS_HOSTNAME_VERIFICATION: &str = "tls_hostname_verification";
+pub const PULSAR_QUEUE_DIR: &str = "queue_dir";
+pub const PULSAR_QUEUE_LIMIT: &str = "queue_limit";
+
+/// Environment variable controlling whether target queue files are Snappy-compressed.
+/// Applies to both notify and audit target queue stores.
+pub const ENV_TARGET_STORE_COMPRESS: &str = "RUSTFS_TARGET_STORE_COMPRESS";
+
+/// Queue-store compression is enabled by default to reduce disk footprint.
+pub const DEFAULT_TARGET_STORE_COMPRESS: bool = true;

crates/config/src/constants/tls.rs

@@ -84,3 +84,69 @@ pub const ENV_SERVER_MTLS_ENABLE: &str = "RUSTFS_SERVER_MTLS_ENABLE";
 /// By default, RustFS server mTLS is disabled.
 /// To change this behavior, set the environment variable RUSTFS_SERVER_MTLS_ENABLE=1
 pub const DEFAULT_SERVER_MTLS_ENABLE: bool = false;
+
+// ── HTTP Transport Tuning Parameters ──
+
+/// Environment variable for HTTP/2 initial stream window size (bytes)
+/// Default: 4194304 (4 MB)
+pub const ENV_H2_INITIAL_STREAM_WINDOW_SIZE: &str = "RUSTFS_H2_INITIAL_STREAM_WINDOW_SIZE";
+pub const DEFAULT_H2_INITIAL_STREAM_WINDOW_SIZE: u32 = 4 * 1024 * 1024; // 4 MB
+
+/// Environment variable for HTTP/2 initial connection window size (bytes)
+/// Default: 8388608 (8 MB)
+pub const ENV_H2_INITIAL_CONN_WINDOW_SIZE: &str = "RUSTFS_H2_INITIAL_CONN_WINDOW_SIZE";
+pub const DEFAULT_H2_INITIAL_CONN_WINDOW_SIZE: u32 = 8 * 1024 * 1024; // 8 MB
+
+/// Environment variable for HTTP/2 max frame size (bytes)
+/// Range: 16384 (16 KB) to 16777216 (16 MB) per RFC 7540
+/// Default: 524288 (512 KB)
+pub const ENV_H2_MAX_FRAME_SIZE: &str = "RUSTFS_H2_MAX_FRAME_SIZE";
+pub const DEFAULT_H2_MAX_FRAME_SIZE: u32 = 512 * 1024; // 512 KB
+
+/// Environment variable for HTTP/2 max header list size (bytes)
+/// Default: 65536 (64 KB)
+pub const ENV_H2_MAX_HEADER_LIST_SIZE: &str = "RUSTFS_H2_MAX_HEADER_LIST_SIZE";
+pub const DEFAULT_H2_MAX_HEADER_LIST_SIZE: u32 = 64 * 1024; // 64 KB
+
+/// Environment variable for HTTP/2 max concurrent streams
+/// Default: 2048
+pub const ENV_H2_MAX_CONCURRENT_STREAMS: &str = "RUSTFS_H2_MAX_CONCURRENT_STREAMS";
+pub const DEFAULT_H2_MAX_CONCURRENT_STREAMS: u32 = 2048;
+
+/// Environment variable for HTTP/2 keep-alive interval (seconds)
+/// Default: 20
+pub const ENV_H2_KEEP_ALIVE_INTERVAL: &str = "RUSTFS_H2_KEEP_ALIVE_INTERVAL";
+pub const DEFAULT_H2_KEEP_ALIVE_INTERVAL: u64 = 20;
+
+/// Environment variable for HTTP/2 keep-alive timeout (seconds)
+/// Default: 10
+pub const ENV_H2_KEEP_ALIVE_TIMEOUT: &str = "RUSTFS_H2_KEEP_ALIVE_TIMEOUT";
+pub const DEFAULT_H2_KEEP_ALIVE_TIMEOUT: u64 = 10;
+
+/// Environment variable for HTTP/1.1 header read timeout (seconds)
+/// Default: 5
+pub const ENV_HTTP1_HEADER_READ_TIMEOUT: &str = "RUSTFS_HTTP1_HEADER_READ_TIMEOUT";
+pub const DEFAULT_HTTP1_HEADER_READ_TIMEOUT: u64 = 5;
+
+/// Environment variable for HTTP/1.1 max buffer size (bytes)
+/// Default: 65536 (64 KB)
+pub const ENV_HTTP1_MAX_BUF_SIZE: &str = "RUSTFS_HTTP1_MAX_BUF_SIZE";
+pub const DEFAULT_HTTP1_MAX_BUF_SIZE: usize = 64 * 1024; // 64 KB
+
+// ── TLS Hot Reload Parameters ──
+
+/// Environment variable to enable TLS certificate hot reload
+/// Default: false
+/// To enable, set the environment variable RUSTFS_TLS_RELOAD_ENABLE=1
+pub const ENV_TLS_RELOAD_ENABLE: &str = "RUSTFS_TLS_RELOAD_ENABLE";
+
+/// Default value for TLS certificate hot reload
+/// By default, RustFS does not reload TLS certificates automatically.
+pub const DEFAULT_TLS_RELOAD_ENABLE: bool = false;
+
+/// Environment variable for TLS certificate reload interval (seconds)
+/// Default: 30 seconds. Minimum: 5 seconds.
+pub const ENV_TLS_RELOAD_INTERVAL: &str = "RUSTFS_TLS_RELOAD_INTERVAL";
+
+/// Default interval for TLS certificate reload check
+pub const DEFAULT_TLS_RELOAD_INTERVAL: u64 = 30;

crates/config/src/constants/workload.rs

@@ -0,0 +1,70 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+//! Workload profile buffer configuration constants.
+//!
+//! This module defines environment variable keys and default values for
+//! custom buffer profile configuration.
+
+use crate::{KI_B, MI_B};
+
+/// Environment variable for minimum buffer size
+/// Default: 64KB (65536 bytes)
+pub const ENV_RUSTFS_BUFFER_MIN_SIZE: &str = "RUSTFS_BUFFER_MIN_SIZE";
+
+/// Environment variable for maximum buffer size
+/// Default: 1MB (1048576 bytes)
+pub const ENV_RUSTFS_BUFFER_MAX_SIZE: &str = "RUSTFS_BUFFER_MAX_SIZE";
+
+/// Environment variable for default buffer size (used when file size is unknown)
+/// Default: 256KB (262144 bytes)
+pub const ENV_RUSTFS_BUFFER_DEFAULT_SIZE: &str = "RUSTFS_BUFFER_DEFAULT_SIZE";
+
+/// Default minimum buffer size: 64KB
+pub const DEFAULT_BUFFER_MIN_SIZE: usize = 64 * KI_B;
+
+/// Default maximum buffer size: 1MB
+pub const DEFAULT_BUFFER_MAX_SIZE: usize = MI_B;
+
+/// Default buffer size for unknown file size: 256KB
+pub const DEFAULT_BUFFER_UNKNOWN_SIZE: usize = 256 * KI_B;
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::MI_B;
+
+    #[test]
+    fn test_default_values() {
+        assert_eq!(DEFAULT_BUFFER_MIN_SIZE, 65536); // 64KB
+        assert_eq!(DEFAULT_BUFFER_MAX_SIZE, 1048576); // 1MB
+        assert_eq!(DEFAULT_BUFFER_UNKNOWN_SIZE, 262144); // 256KB
+    }
+
+    #[test]
+    fn test_constants() {
+        assert_eq!(KI_B, 1024);
+        assert_eq!(MI_B, 1024 * 1024);
+        assert_eq!(64 * KI_B, DEFAULT_BUFFER_MIN_SIZE);
+        assert_eq!(MI_B, DEFAULT_BUFFER_MAX_SIZE);
+        assert_eq!(256 * KI_B, DEFAULT_BUFFER_UNKNOWN_SIZE);
+    }
+
+    #[test]
+    fn test_env_var_names() {
+        assert_eq!(ENV_RUSTFS_BUFFER_MIN_SIZE, "RUSTFS_BUFFER_MIN_SIZE");
+        assert_eq!(ENV_RUSTFS_BUFFER_MAX_SIZE, "RUSTFS_BUFFER_MAX_SIZE");
+        assert_eq!(ENV_RUSTFS_BUFFER_DEFAULT_SIZE, "RUSTFS_BUFFER_DEFAULT_SIZE");
+    }
+}

crates/config/src/constants/zero_copy.rs

@@ -0,0 +1,105 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+//! Zero-copy I/O configuration constants.
+//!
+//! This module defines environment variables and default values for zero-copy
+//! read operations, which use memory mapping (mmap) to avoid data copying.
+
+// =============================================================================
+// Zero-Copy Configuration
+// =============================================================================
+
+/// Environment variable for zero-copy read enable.
+///
+/// When enabled, uses mmap (Unix) or optimized reads for zero-copy data access.
+/// This reduces memory copies from 3-4 to 1, lowering CPU usage by 20-30%
+/// and improving P95 latency by 15-25%.
+///
+/// - Purpose: Enable or disable zero-copy read operations
+/// - Acceptable values: `"true"` / `"false"` (case-insensitive) or a boolean typed config
+/// - Semantics: When enabled, uses mmap on Unix systems for memory-mapped file reads;
+///   falls back to regular I/O on non-Unix platforms or when mmap fails
+/// - Example: `export RUSTFS_OBJECT_ZERO_COPY_ENABLE=true`
+/// - Note: Zero-copy is safe for all workloads and provides significant performance
+///   benefits with minimal risk. Disable only if mmap-related issues are encountered.
+pub const ENV_OBJECT_ZERO_COPY_ENABLE: &str = "RUSTFS_OBJECT_ZERO_COPY_ENABLE";
+
+/// Default: zero-copy reads are enabled.
+///
+/// Zero-copy uses memory mapping (mmap) on Unix systems to avoid data copying
+/// between kernel and user space. This provides:
+/// - Reduced memory copies: from 3-4 copies to 1 copy
+/// - Lower CPU usage: 20-30% reduction expected
+/// - Improved latency P95: 15-25% reduction expected
+/// - Increased throughput: 10-20% improvement expected
+///
+/// On non-Unix platforms or when mmap fails, the system automatically falls back
+/// to regular I/O without errors.
+pub const DEFAULT_OBJECT_ZERO_COPY_ENABLE: bool = true;
+
+// =============================================================================
+// Direct I/O Configuration
+// =============================================================================
+
+/// Environment variable for Direct I/O enable (Linux only).
+///
+/// When enabled, uses O_DIRECT flag to bypass OS page cache for large files.
+/// This is only beneficial for specific workloads (databases, large sequential reads).
+///
+/// - Purpose: Enable or disable Direct I/O for large file operations
+/// - Acceptable values: `"true"` / `"false"` (case-insensitive) or a boolean typed config
+/// - Semantics: When enabled, files larger than the threshold will use O_DIRECT flag;
+///   this bypasses the OS page cache and transfers data directly between disk and application
+/// - Example: `export RUSTFS_OBJECT_DIRECT_IO_ENABLE=true`
+/// - Note: Direct I/O is disabled by default because it's only beneficial for specific
+///   use cases. For most workloads, the OS page cache provides better performance.
+pub const ENV_OBJECT_DIRECT_IO_ENABLE: &str = "RUSTFS_OBJECT_DIRECT_IO_ENABLE";
+
+/// Default: Direct I/O is disabled.
+///
+/// Direct I/O is disabled by default because it's only beneficial for specific use cases:
+/// - Large file transfers (>128MB)
+/// - Databases with their own cache
+/// - Applications requiring predictable I/O latency
+///
+/// For most workloads, the OS page cache provides better performance through:
+/// - Read-ahead caching
+/// - Write buffering
+/// - Multi-use caching (same data cached for multiple operations)
+pub const DEFAULT_OBJECT_DIRECT_IO_ENABLE: bool = false;
+
+/// Environment variable for Direct I/O minimum file size threshold.
+///
+/// Files smaller than this size will use regular I/O even if Direct I/O is enabled.
+/// This avoids the overhead of Direct I/O for small files where the OS page cache
+/// is more effective.
+///
+/// - Purpose: Set the minimum file size for Direct I/O operations
+/// - Unit: Bytes
+/// - Valid values: any positive integer (default: 134,217,728 bytes = 128 MB)
+/// - Semantics: Only files larger than this threshold will use Direct I/O when enabled;
+///   smaller files use regular buffered I/O
+/// - Example: `export RUSTFS_OBJECT_DIRECT_IO_THRESHOLD=268435456`
+/// - Note: The default threshold of 128MB balances the overhead of Direct I/O setup
+///   against the benefits of bypassing the page cache for large files.
+pub const ENV_OBJECT_DIRECT_IO_THRESHOLD: &str = "RUSTFS_OBJECT_DIRECT_IO_THRESHOLD";
+
+/// Default Direct I/O threshold: 128 MB.
+///
+/// Only files larger than 128MB will use Direct I/O when enabled.
+/// Smaller files benefit from OS page cache.
+///
+/// Formula: 128 * 1024 * 1024 = 134,217,728 bytes
+pub const DEFAULT_OBJECT_DIRECT_IO_THRESHOLD: usize = 128 * 1024 * 1024;

crates/config/src/lib.rs

@@ -19,14 +19,22 @@ pub use constants::app::*;
 #[cfg(feature = "constants")]
 pub use constants::body_limits::*;
 #[cfg(feature = "constants")]
+pub use constants::capacity::*;
+#[cfg(feature = "constants")]
 pub use constants::compress::*;
 #[cfg(feature = "constants")]
 pub use constants::console::*;
 #[cfg(feature = "constants")]
+pub use constants::drive::*;
+#[cfg(feature = "constants")]
 pub use constants::env::*;
 #[cfg(feature = "constants")]
 pub use constants::heal::*;
 #[cfg(feature = "constants")]
+pub use constants::health::*;
+#[cfg(feature = "constants")]
+pub use constants::internode::*;
+#[cfg(feature = "constants")]
 pub use constants::object::*;
 #[cfg(feature = "constants")]
 pub use constants::profiler::*;
@@ -45,6 +53,10 @@ pub use constants::targets::*;
 #[cfg(feature = "constants")]
 pub use constants::tls::*;
 #[cfg(feature = "constants")]
+pub use constants::workload::*;
+#[cfg(feature = "constants")]
+pub use constants::zero_copy::*;
+#[cfg(feature = "constants")]
 pub mod oidc {
     pub use super::constants::oidc::*;
 }

crates/config/src/notify/kafka.rs

@@ -0,0 +1,53 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+/// A list of all valid configuration keys for a Kafka target.
+pub const NOTIFY_KAFKA_KEYS: &[&str] = &[
+    crate::ENABLE_KEY,
+    crate::KAFKA_BROKERS,
+    crate::KAFKA_TOPIC,
+    crate::KAFKA_ACKS,
+    crate::KAFKA_TLS_ENABLE,
+    crate::KAFKA_TLS_CA,
+    crate::KAFKA_TLS_CLIENT_CERT,
+    crate::KAFKA_TLS_CLIENT_KEY,
+    crate::KAFKA_QUEUE_DIR,
+    crate::KAFKA_QUEUE_LIMIT,
+    crate::COMMENT_KEY,
+];
+
+// Kafka Environment Variables
+pub const ENV_NOTIFY_KAFKA_ENABLE: &str = "RUSTFS_NOTIFY_KAFKA_ENABLE";
+pub const ENV_NOTIFY_KAFKA_BROKERS: &str = "RUSTFS_NOTIFY_KAFKA_BROKERS";
+pub const ENV_NOTIFY_KAFKA_TOPIC: &str = "RUSTFS_NOTIFY_KAFKA_TOPIC";
+pub const ENV_NOTIFY_KAFKA_ACKS: &str = "RUSTFS_NOTIFY_KAFKA_ACKS";
+pub const ENV_NOTIFY_KAFKA_TLS_ENABLE: &str = "RUSTFS_NOTIFY_KAFKA_TLS_ENABLE";
+pub const ENV_NOTIFY_KAFKA_TLS_CA: &str = "RUSTFS_NOTIFY_KAFKA_TLS_CA";
+pub const ENV_NOTIFY_KAFKA_TLS_CLIENT_CERT: &str = "RUSTFS_NOTIFY_KAFKA_TLS_CLIENT_CERT";
+pub const ENV_NOTIFY_KAFKA_TLS_CLIENT_KEY: &str = "RUSTFS_NOTIFY_KAFKA_TLS_CLIENT_KEY";
+pub const ENV_NOTIFY_KAFKA_QUEUE_DIR: &str = "RUSTFS_NOTIFY_KAFKA_QUEUE_DIR";
+pub const ENV_NOTIFY_KAFKA_QUEUE_LIMIT: &str = "RUSTFS_NOTIFY_KAFKA_QUEUE_LIMIT";
+
+pub const ENV_NOTIFY_KAFKA_KEYS: &[&str; 10] = &[
+    ENV_NOTIFY_KAFKA_ENABLE,
+    ENV_NOTIFY_KAFKA_BROKERS,
+    ENV_NOTIFY_KAFKA_TOPIC,
+    ENV_NOTIFY_KAFKA_ACKS,
+    ENV_NOTIFY_KAFKA_TLS_ENABLE,
+    ENV_NOTIFY_KAFKA_TLS_CA,
+    ENV_NOTIFY_KAFKA_TLS_CLIENT_CERT,
+    ENV_NOTIFY_KAFKA_TLS_CLIENT_KEY,
+    ENV_NOTIFY_KAFKA_QUEUE_DIR,
+    ENV_NOTIFY_KAFKA_QUEUE_LIMIT,
+];

crates/config/src/notify/mod.rs

@@ -13,12 +13,18 @@
 // limitations under the License.
 
 mod arn;
+mod kafka;
 mod mqtt;
+mod nats;
+mod pulsar;
 mod store;
 mod webhook;
 
 pub use arn::*;
+pub use kafka::*;
 pub use mqtt::*;
+pub use nats::*;
+pub use pulsar::*;
 pub use store::*;
 pub use webhook::*;
 
@@ -64,9 +70,14 @@ pub const ENV_NOTIFY_SEND_CONCURRENCY: &str = "RUSTFS_NOTIFY_SEND_CONCURRENCY";
 pub const DEFAULT_NOTIFY_SEND_CONCURRENCY: usize = 64;
 
 #[allow(dead_code)]
-pub const NOTIFY_SUB_SYSTEMS: &[&str] = &[NOTIFY_MQTT_SUB_SYS, NOTIFY_WEBHOOK_SUB_SYS];
+pub const NOTIFY_SUB_SYSTEMS: &[&str] = &[
+    NOTIFY_KAFKA_SUB_SYS,
+    NOTIFY_MQTT_SUB_SYS,
+    NOTIFY_NATS_SUB_SYS,
+    NOTIFY_PULSAR_SUB_SYS,
+    NOTIFY_WEBHOOK_SUB_SYS,
+];
 
-#[allow(dead_code)]
 pub const NOTIFY_KAFKA_SUB_SYS: &str = "notify_kafka";
 pub const NOTIFY_MQTT_SUB_SYS: &str = "notify_mqtt";
 #[allow(dead_code)]
@@ -83,4 +94,5 @@ pub const NOTIFY_AMQP_SUB_SYS: &str = "notify_amqp";
 pub const NOTIFY_POSTGRES_SUB_SYS: &str = "notify_postgres";
 #[allow(dead_code)]
 pub const NOTIFY_REDIS_SUB_SYS: &str = "notify_redis";
+pub const NOTIFY_PULSAR_SUB_SYS: &str = "notify_pulsar";
 pub const NOTIFY_WEBHOOK_SUB_SYS: &str = "notify_webhook";

crates/config/src/notify/mqtt.rs

@@ -24,6 +24,12 @@ pub const NOTIFY_MQTT_KEYS: &[&str] = &[
     crate::MQTT_KEEP_ALIVE_INTERVAL,
     crate::MQTT_QUEUE_DIR,
     crate::MQTT_QUEUE_LIMIT,
+    crate::MQTT_TLS_POLICY,
+    crate::MQTT_TLS_CA,
+    crate::MQTT_TLS_CLIENT_CERT,
+    crate::MQTT_TLS_CLIENT_KEY,
+    crate::MQTT_TLS_TRUST_LEAF_AS_CA,
+    crate::MQTT_WS_PATH_ALLOWLIST,
     crate::COMMENT_KEY,
 ];
 
@@ -38,8 +44,14 @@ pub const ENV_NOTIFY_MQTT_RECONNECT_INTERVAL: &str = "RUSTFS_NOTIFY_MQTT_RECONNE
 pub const ENV_NOTIFY_MQTT_KEEP_ALIVE_INTERVAL: &str = "RUSTFS_NOTIFY_MQTT_KEEP_ALIVE_INTERVAL";
 pub const ENV_NOTIFY_MQTT_QUEUE_DIR: &str = "RUSTFS_NOTIFY_MQTT_QUEUE_DIR";
 pub const ENV_NOTIFY_MQTT_QUEUE_LIMIT: &str = "RUSTFS_NOTIFY_MQTT_QUEUE_LIMIT";
+pub const ENV_NOTIFY_MQTT_TLS_POLICY: &str = "RUSTFS_NOTIFY_MQTT_TLS_POLICY";
+pub const ENV_NOTIFY_MQTT_TLS_CA: &str = "RUSTFS_NOTIFY_MQTT_TLS_CA";
+pub const ENV_NOTIFY_MQTT_TLS_CLIENT_CERT: &str = "RUSTFS_NOTIFY_MQTT_TLS_CLIENT_CERT";
+pub const ENV_NOTIFY_MQTT_TLS_CLIENT_KEY: &str = "RUSTFS_NOTIFY_MQTT_TLS_CLIENT_KEY";
+pub const ENV_NOTIFY_MQTT_TLS_TRUST_LEAF_AS_CA: &str = "RUSTFS_NOTIFY_MQTT_TLS_TRUST_LEAF_AS_CA";
+pub const ENV_NOTIFY_MQTT_WS_PATH_ALLOWLIST: &str = "RUSTFS_NOTIFY_MQTT_WS_PATH_ALLOWLIST";
 
-pub const ENV_NOTIFY_MQTT_KEYS: &[&str; 10] = &[
+pub const ENV_NOTIFY_MQTT_KEYS: &[&str; 16] = &[
     ENV_NOTIFY_MQTT_ENABLE,
     ENV_NOTIFY_MQTT_BROKER,
     ENV_NOTIFY_MQTT_TOPIC,
@@ -50,4 +62,10 @@ pub const ENV_NOTIFY_MQTT_KEYS: &[&str; 10] = &[
     ENV_NOTIFY_MQTT_KEEP_ALIVE_INTERVAL,
     ENV_NOTIFY_MQTT_QUEUE_DIR,
     ENV_NOTIFY_MQTT_QUEUE_LIMIT,
+    ENV_NOTIFY_MQTT_TLS_POLICY,
+    ENV_NOTIFY_MQTT_TLS_CA,
+    ENV_NOTIFY_MQTT_TLS_CLIENT_CERT,
+    ENV_NOTIFY_MQTT_TLS_CLIENT_KEY,
+    ENV_NOTIFY_MQTT_TLS_TRUST_LEAF_AS_CA,
+    ENV_NOTIFY_MQTT_WS_PATH_ALLOWLIST,
 ];

crates/config/src/notify/nats.rs

@@ -0,0 +1,60 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+pub const NOTIFY_NATS_KEYS: &[&str] = &[
+    crate::ENABLE_KEY,
+    crate::NATS_ADDRESS,
+    crate::NATS_SUBJECT,
+    crate::NATS_USERNAME,
+    crate::NATS_PASSWORD,
+    crate::NATS_TOKEN,
+    crate::NATS_CREDENTIALS_FILE,
+    crate::NATS_TLS_CA,
+    crate::NATS_TLS_CLIENT_CERT,
+    crate::NATS_TLS_CLIENT_KEY,
+    crate::NATS_TLS_REQUIRED,
+    crate::NATS_QUEUE_DIR,
+    crate::NATS_QUEUE_LIMIT,
+    crate::COMMENT_KEY,
+];
+
+pub const ENV_NOTIFY_NATS_ENABLE: &str = "RUSTFS_NOTIFY_NATS_ENABLE";
+pub const ENV_NOTIFY_NATS_ADDRESS: &str = "RUSTFS_NOTIFY_NATS_ADDRESS";
+pub const ENV_NOTIFY_NATS_SUBJECT: &str = "RUSTFS_NOTIFY_NATS_SUBJECT";
+pub const ENV_NOTIFY_NATS_USERNAME: &str = "RUSTFS_NOTIFY_NATS_USERNAME";
+pub const ENV_NOTIFY_NATS_PASSWORD: &str = "RUSTFS_NOTIFY_NATS_PASSWORD";
+pub const ENV_NOTIFY_NATS_TOKEN: &str = "RUSTFS_NOTIFY_NATS_TOKEN";
+pub const ENV_NOTIFY_NATS_CREDENTIALS_FILE: &str = "RUSTFS_NOTIFY_NATS_CREDENTIALS_FILE";
+pub const ENV_NOTIFY_NATS_TLS_CA: &str = "RUSTFS_NOTIFY_NATS_TLS_CA";
+pub const ENV_NOTIFY_NATS_TLS_CLIENT_CERT: &str = "RUSTFS_NOTIFY_NATS_TLS_CLIENT_CERT";
+pub const ENV_NOTIFY_NATS_TLS_CLIENT_KEY: &str = "RUSTFS_NOTIFY_NATS_TLS_CLIENT_KEY";
+pub const ENV_NOTIFY_NATS_TLS_REQUIRED: &str = "RUSTFS_NOTIFY_NATS_TLS_REQUIRED";
+pub const ENV_NOTIFY_NATS_QUEUE_DIR: &str = "RUSTFS_NOTIFY_NATS_QUEUE_DIR";
+pub const ENV_NOTIFY_NATS_QUEUE_LIMIT: &str = "RUSTFS_NOTIFY_NATS_QUEUE_LIMIT";
+
+pub const ENV_NOTIFY_NATS_KEYS: &[&str; 13] = &[
+    ENV_NOTIFY_NATS_ENABLE,
+    ENV_NOTIFY_NATS_ADDRESS,
+    ENV_NOTIFY_NATS_SUBJECT,
+    ENV_NOTIFY_NATS_USERNAME,
+    ENV_NOTIFY_NATS_PASSWORD,
+    ENV_NOTIFY_NATS_TOKEN,
+    ENV_NOTIFY_NATS_CREDENTIALS_FILE,
+    ENV_NOTIFY_NATS_TLS_CA,
+    ENV_NOTIFY_NATS_TLS_CLIENT_CERT,
+    ENV_NOTIFY_NATS_TLS_CLIENT_KEY,
+    ENV_NOTIFY_NATS_TLS_REQUIRED,
+    ENV_NOTIFY_NATS_QUEUE_DIR,
+    ENV_NOTIFY_NATS_QUEUE_LIMIT,
+];

crates/config/src/notify/pulsar.rs

@@ -0,0 +1,54 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+pub const NOTIFY_PULSAR_KEYS: &[&str] = &[
+    crate::ENABLE_KEY,
+    crate::PULSAR_BROKER,
+    crate::PULSAR_TOPIC,
+    crate::PULSAR_AUTH_TOKEN,
+    crate::PULSAR_USERNAME,
+    crate::PULSAR_PASSWORD,
+    crate::PULSAR_TLS_CA,
+    crate::PULSAR_TLS_ALLOW_INSECURE,
+    crate::PULSAR_TLS_HOSTNAME_VERIFICATION,
+    crate::PULSAR_QUEUE_DIR,
+    crate::PULSAR_QUEUE_LIMIT,
+    crate::COMMENT_KEY,
+];
+
+pub const ENV_NOTIFY_PULSAR_ENABLE: &str = "RUSTFS_NOTIFY_PULSAR_ENABLE";
+pub const ENV_NOTIFY_PULSAR_BROKER: &str = "RUSTFS_NOTIFY_PULSAR_BROKER";
+pub const ENV_NOTIFY_PULSAR_TOPIC: &str = "RUSTFS_NOTIFY_PULSAR_TOPIC";
+pub const ENV_NOTIFY_PULSAR_AUTH_TOKEN: &str = "RUSTFS_NOTIFY_PULSAR_AUTH_TOKEN";
+pub const ENV_NOTIFY_PULSAR_USERNAME: &str = "RUSTFS_NOTIFY_PULSAR_USERNAME";
+pub const ENV_NOTIFY_PULSAR_PASSWORD: &str = "RUSTFS_NOTIFY_PULSAR_PASSWORD";
+pub const ENV_NOTIFY_PULSAR_TLS_CA: &str = "RUSTFS_NOTIFY_PULSAR_TLS_CA";
+pub const ENV_NOTIFY_PULSAR_TLS_ALLOW_INSECURE: &str = "RUSTFS_NOTIFY_PULSAR_TLS_ALLOW_INSECURE";
+pub const ENV_NOTIFY_PULSAR_TLS_HOSTNAME_VERIFICATION: &str = "RUSTFS_NOTIFY_PULSAR_TLS_HOSTNAME_VERIFICATION";
+pub const ENV_NOTIFY_PULSAR_QUEUE_DIR: &str = "RUSTFS_NOTIFY_PULSAR_QUEUE_DIR";
+pub const ENV_NOTIFY_PULSAR_QUEUE_LIMIT: &str = "RUSTFS_NOTIFY_PULSAR_QUEUE_LIMIT";
+
+pub const ENV_NOTIFY_PULSAR_KEYS: &[&str; 11] = &[
+    ENV_NOTIFY_PULSAR_ENABLE,
+    ENV_NOTIFY_PULSAR_BROKER,
+    ENV_NOTIFY_PULSAR_TOPIC,
+    ENV_NOTIFY_PULSAR_AUTH_TOKEN,
+    ENV_NOTIFY_PULSAR_USERNAME,
+    ENV_NOTIFY_PULSAR_PASSWORD,
+    ENV_NOTIFY_PULSAR_TLS_CA,
+    ENV_NOTIFY_PULSAR_TLS_ALLOW_INSECURE,
+    ENV_NOTIFY_PULSAR_TLS_HOSTNAME_VERIFICATION,
+    ENV_NOTIFY_PULSAR_QUEUE_DIR,
+    ENV_NOTIFY_PULSAR_QUEUE_LIMIT,
+];

crates/config/src/observability/mod.rs

@@ -22,6 +22,14 @@ pub const ENV_OBS_ENDPOINT: &str = "RUSTFS_OBS_ENDPOINT";
 pub const ENV_OBS_TRACE_ENDPOINT: &str = "RUSTFS_OBS_TRACE_ENDPOINT";
 pub const ENV_OBS_METRIC_ENDPOINT: &str = "RUSTFS_OBS_METRIC_ENDPOINT";
 pub const ENV_OBS_LOG_ENDPOINT: &str = "RUSTFS_OBS_LOG_ENDPOINT";
+pub const ENV_OBS_ENDPOINT_HEADERS: &str = "RUSTFS_OBS_ENDPOINT_HEADERS";
+pub const ENV_OBS_ENDPOINT_TRACES_HEADERS: &str = "RUSTFS_OBS_ENDPOINT_TRACES_HEADERS";
+pub const ENV_OBS_ENDPOINT_METRICS_HEADERS: &str = "RUSTFS_OBS_ENDPOINT_METRICS_HEADERS";
+pub const ENV_OBS_ENDPOINT_LOGS_HEADERS: &str = "RUSTFS_OBS_ENDPOINT_LOGS_HEADERS";
+pub const ENV_OBS_ENDPOINT_TIMEOUT_MILLIS: &str = "RUSTFS_OBS_ENDPOINT_TIMEOUT_MILLIS";
+pub const ENV_OBS_ENDPOINT_TRACES_TIMEOUT_MILLIS: &str = "RUSTFS_OBS_ENDPOINT_TRACES_TIMEOUT_MILLIS";
+pub const ENV_OBS_ENDPOINT_METRICS_TIMEOUT_MILLIS: &str = "RUSTFS_OBS_ENDPOINT_METRICS_TIMEOUT_MILLIS";
+pub const ENV_OBS_ENDPOINT_LOGS_TIMEOUT_MILLIS: &str = "RUSTFS_OBS_ENDPOINT_LOGS_TIMEOUT_MILLIS";
 pub const ENV_OBS_PROFILING_ENDPOINT: &str = "RUSTFS_OBS_PROFILING_ENDPOINT";
 pub const ENV_OBS_USE_STDOUT: &str = "RUSTFS_OBS_USE_STDOUT";
 pub const ENV_OBS_SAMPLE_RATIO: &str = "RUSTFS_OBS_SAMPLE_RATIO";
@@ -108,6 +116,14 @@ mod tests {
         assert_eq!(ENV_OBS_TRACE_ENDPOINT, "RUSTFS_OBS_TRACE_ENDPOINT");
         assert_eq!(ENV_OBS_METRIC_ENDPOINT, "RUSTFS_OBS_METRIC_ENDPOINT");
         assert_eq!(ENV_OBS_LOG_ENDPOINT, "RUSTFS_OBS_LOG_ENDPOINT");
+        assert_eq!(ENV_OBS_ENDPOINT_HEADERS, "RUSTFS_OBS_ENDPOINT_HEADERS");
+        assert_eq!(ENV_OBS_ENDPOINT_TRACES_HEADERS, "RUSTFS_OBS_ENDPOINT_TRACES_HEADERS");
+        assert_eq!(ENV_OBS_ENDPOINT_METRICS_HEADERS, "RUSTFS_OBS_ENDPOINT_METRICS_HEADERS");
+        assert_eq!(ENV_OBS_ENDPOINT_LOGS_HEADERS, "RUSTFS_OBS_ENDPOINT_LOGS_HEADERS");
+        assert_eq!(ENV_OBS_ENDPOINT_TIMEOUT_MILLIS, "RUSTFS_OBS_ENDPOINT_TIMEOUT_MILLIS");
+        assert_eq!(ENV_OBS_ENDPOINT_TRACES_TIMEOUT_MILLIS, "RUSTFS_OBS_ENDPOINT_TRACES_TIMEOUT_MILLIS");
+        assert_eq!(ENV_OBS_ENDPOINT_METRICS_TIMEOUT_MILLIS, "RUSTFS_OBS_ENDPOINT_METRICS_TIMEOUT_MILLIS");
+        assert_eq!(ENV_OBS_ENDPOINT_LOGS_TIMEOUT_MILLIS, "RUSTFS_OBS_ENDPOINT_LOGS_TIMEOUT_MILLIS");
         assert_eq!(ENV_OBS_PROFILING_ENDPOINT, "RUSTFS_OBS_PROFILING_ENDPOINT");
         assert_eq!(ENV_OBS_USE_STDOUT, "RUSTFS_OBS_USE_STDOUT");
         assert_eq!(ENV_OBS_SAMPLE_RATIO, "RUSTFS_OBS_SAMPLE_RATIO");