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

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

.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

@@ -16,10 +16,8 @@ services:
 
   # --- Tracing ---
   tempo:
-    image: grafana/tempo:2.10.3
+    image: grafana/tempo:2.10.5
     container_name: tempo
-    depends_on:
-      - redpanda
     command: [ "-config.file=/etc/tempo.yaml" ]
     volumes:
       - ./tempo.yaml:/etc/tempo.yaml:ro
@@ -33,31 +31,11 @@ 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
-  redpanda:
-    image: redpandadata/redpanda:latest
-    ports:
-      - "9092:9092" # Kafka API for clients
-    command: >
-      redpanda start --overprovisioned
-      --mode=dev-container
-      --kafka-addr=PLAINTEXT://0.0.0.0:9092
-      --advertise-kafka-addr=PLAINTEXT://redpanda:9092
-
-  redpanda-console:
-    image: docker.redpanda.com/redpandadata/console:latest
-    environment:
-      - CONFIG_FILEPATH=/etc/redpanda/redpanda-console-config.yaml
-    volumes:
-      - ./redpanda-console.yaml:/etc/redpanda/redpanda-console-config.yaml
-    ports:
-      - "8080:8080"
-    depends_on:
-      - redpanda
 
   vulture:
     image: grafana/tempo-vulture:latest
@@ -106,6 +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"
@@ -168,7 +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:

.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:
@@ -33,33 +32,17 @@ distributor:
           endpoint: "0.0.0.0:4317"
         http:
           endpoint: "0.0.0.0:4318"
-  #log_received_spans:
-  #    enabled: true
-  #  log_discarded_spans:
-  #    enabled: true
 
-backend_scheduler:
-  provider:
-    compaction:
-      compaction:
-        block_retention: 1h
-
-backend_worker:
-  backend_scheduler_addr: localhost:3200
-  compaction:
-    block_retention: 1h
-  ring:
-    kvstore:
-      store: memberlist
-
-querier:
-  query_live_store: true
+ingester:
+  max_block_duration: 5m
 
 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:
@@ -85,15 +68,5 @@ overrides:
       processors: [ "span-metrics", "service-graphs", "local-blocks" ]
       generate_native_histograms: both
 
-ingest:
-  enabled: true
-  kafka:
-    address: redpanda:9092
-    topic: tempo-ingest
-
-block_builder:
-  consume_cycle_duration: 30s
-
 usage_report:
   reporting_enabled: false
-

.github/dependabot.yml

@@ -27,9 +27,8 @@ updates:
       timezone: "Asia/Shanghai"
       time: "08:00"
     assignees:
-      - "heihutu"
-    reviewers:
       - "houseme"
+    reviewers:
       - "overtrue"
       - "majinghe"
     ignore:
@@ -39,6 +38,8 @@ updates:
         versions: [ "0.23.x" ]
       - dependency-name: "ratelimit"
         versions: [ "1.x" ]
+      - dependency-name: "ratelimit"
+        versions: [ "2.x" ]
     groups:
       s3s:
         update-types:

AGENTS.md

@@ -6,24 +6,39 @@ 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
 
@@ -44,6 +59,8 @@ Reference the source files above instead.
 
 ## Verification Before PR
 
+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
@@ -69,6 +86,7 @@ Do not open a PR with code changes when the 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.

ARCHITECTURE.md

@@ -86,7 +86,7 @@ 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, mcp, s3-common, workers, zip
+  madmin, s3-common, workers, zip
 
 Depth 1:
   io-core (→ io-metrics)
@@ -195,7 +195,6 @@ Depth 8 — TOP:
 | `trusted-proxies` | 4.0K | Trusted proxy / IP forwarding |
 | `zip` | 986 | ZIP archive support for bulk downloads |
 | `workers` | 136 | Simple worker abstraction |
-| `mcp` | 2.0K | Model Context Protocol server (AI tooling) |
 
 ## Architecture Invariants
 
@@ -224,7 +223,7 @@ Depth 8 — TOP:
 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`;
-     `mcp` and `heal` use `anyhow` in library code.
+      `heal` use `anyhow` in library code.
 
 ## Known Structural Issues
 
@@ -285,9 +284,9 @@ anyhow::Result<T>             // in library code (OK in tests/CLI)
 
 ### Metrics
 
-- Prometheus-style metrics via `rustfs-metrics` crate
+- Prometheus-style metrics via `rustfs-obs` runtime and schema
 - I/O-specific counters via `rustfs-io-metrics`
-- Registration happens at crate level, collection in `metrics` crate
+- Registration happens at crate level, collection/reporting in `rustfs-obs`
 
 ### Testing
 
@@ -369,7 +368,7 @@ The binary (`main.rs`) boots in this order:
   Add handler in `admin/handlers/`, register in `admin/router.rs`
 
 - **"Where do I add a new metric?"**
-  Define in `crates/metrics/`, register collector, expose via `/minio/v2/metrics`
+  Define descriptor/collector in `crates/obs/src/metrics/`, expose via `/minio/v2/metrics`
 
 ---
 

Cargo.toml

@@ -31,8 +31,6 @@ 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
@@ -40,7 +38,7 @@ members = [
     "crates/protocols", # Protocol implementations (FTPS, SFTP, etc.)
     "crates/protos", # Protocol buffer definitions
     "crates/rio", # Rust I/O utilities and abstractions
-    "crates/concurrency", # 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
@@ -61,7 +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"]
@@ -78,64 +76,65 @@ 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-concurrency = { path = "./crates/concurrency", 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-io-metrics = { path = "crates/io-metrics", version = "0.0.5" }
-rustfs-io-core = { path = "crates/io-core", version = "0.0.5" }
-rustfs-object-capacity = { path = "crates/object-capacity", 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"
+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"
+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 = "1.52.0", features = ["fs", "rt-multi-thread", "io-uring"] }
+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"
@@ -154,27 +153,25 @@ flatbuffers = "25.12.19"
 form_urlencoded = "1.2.2"
 prost = "0.14.3"
 quick-xml = "0.39.2"
-rmcp = { version = "1.4.0" }
 rmp = { version = "0.8.15" }
 rmp-serde = { version = "1.3.1" }
 serde = { version = "1.0.228", features = ["derive"] }
 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.8" }
-blake2 = "0.11.0-rc.5"
+blake2 = "0.11.0-rc.6"
 chacha20poly1305 = { version = "0.11.0-rc.3" }
 crc-fast = "1.9.0"
 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.10"
-rsa = { version = "0.10.0-rc.17" }
-rustls = { version = "0.23.38", default-features = false, features = ["aws-lc-rs", "logging", "tls12", "prefer-post-quantum", "std"] }
-rustls-pki-types = "1.14.0"
+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"
@@ -183,18 +180,18 @@ 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 = "1.9.1"
-astral-tokio-tar = "0.6.0"
+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 = "1.129.0", 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 = "1.4.7" }
 backtrace = "0.3.76"
@@ -202,7 +199,7 @@ 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"] }
@@ -210,7 +207,7 @@ crossbeam-queue = "0.3.12"
 crossbeam-channel = "0.5.15"
 crossbeam-deque = "0.8.6"
 crossbeam-utils = "0.8.21"
-datafusion = "53.0.0"
+datafusion = "53.1.0"
 derive_builder = "0.20.2"
 enumset = "1.1.10"
 faster-hex = "0.10.0"
@@ -224,9 +221,9 @@ hex-simd = "0.8.0"
 highway = { version = "1.3.0" }
 ipnetwork = { version = "0.21.1", features = ["serde"] }
 lazy_static = "1.5.0"
-libc = "0.2.185"
+libc = "0.2.186"
 libsystemd = "0.7.2"
-local-ip-address = "0.6.11"
+local-ip-address = "0.6.12"
 memmap2 = "0.9.10"
 lz4 = "1.28.1"
 matchit = "0.9.2"
@@ -250,13 +247,13 @@ 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 = { package = "rumqttc-next", version = "0.29.0", features = ["websocket"] }
+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 = "2.1.2" }
-s3s = { git = "https://github.com/rustfs/s3s", rev = "0dd8fcaaa72eda68fafd49c38daea43bb8697558", features = ["minio"] }
+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"
@@ -270,37 +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 = "1.23.0", features = ["v4", "fast-rng", "macro-diagnostics"] }
+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 = "8.5.1"
+zip = "8.6.0"
 zstd = "0.13.3"
 
 # Observability and Metrics
 metrics = "0.24.3"
-dial9-tokio-telemetry = "0.2"
+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 = "0.31.1", 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
@@ -320,7 +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
@@ -73,7 +73,7 @@ LABEL name="RustFS" \
       license="Apache-2.0"
 
 RUN apk update && \
-    apk add --no-cache ca-certificates coreutils curl "zlib>=1.3.2-r0"
+    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

@@ -115,7 +115,7 @@ chown -R 10001:10001 data logs
 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
+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

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：

_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"

crates/audit/Cargo.toml

@@ -41,11 +41,10 @@ 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

crates/audit/src/factory.rs

@@ -14,28 +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_TLS_CA, MQTT_TLS_CLIENT_CERT, MQTT_TLS_CLIENT_KEY, MQTT_TLS_POLICY,
-    MQTT_TLS_TRUST_LEAF_AS_CA, MQTT_TOPIC, MQTT_USERNAME, MQTT_WS_PATH_ALLOWLIST, 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,
-    error::TargetError,
-    target::{
-        mqtt::{MQTTArgs, MQTTTlsConfig, validate_mqtt_broker_url},
-        webhook::WebhookArgs,
+    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::TargetType,
 };
-use std::time::Duration;
-use tracing::{debug, warn};
-use url::Url;
+use std::collections::HashSet;
 
 /// Trait for creating targets from configuration
 #[async_trait]
@@ -49,10 +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
@@ -61,71 +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
@@ -134,112 +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)),
-            tls: MQTTTlsConfig::from_values(
-                config.lookup(MQTT_TLS_POLICY).as_deref(),
-                config.lookup(MQTT_TLS_CA).as_deref(),
-                config.lookup(MQTT_TLS_CLIENT_CERT).as_deref(),
-                config.lookup(MQTT_TLS_CLIENT_KEY).as_deref(),
-                config.lookup(MQTT_TLS_TRUST_LEAF_AS_CA).as_deref(),
-                config.lookup(MQTT_WS_PATH_ALLOWLIST).as_deref(),
-            )?,
-            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}')")))?;
-
-        let tls = MQTTTlsConfig::from_values(
-            config.lookup(MQTT_TLS_POLICY).as_deref(),
-            config.lookup(MQTT_TLS_CA).as_deref(),
-            config.lookup(MQTT_TLS_CLIENT_CERT).as_deref(),
-            config.lookup(MQTT_TLS_CLIENT_KEY).as_deref(),
-            config.lookup(MQTT_TLS_TRUST_LEAF_AS_CA).as_deref(),
-            config.lookup(MQTT_WS_PATH_ALLOWLIST).as_deref(),
-        )?;
-        validate_mqtt_broker_url(&url, &tls)?;
-
-        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
     }
@@ -105,143 +109,22 @@ 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();
-        // 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");
-
-            // *** 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 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)
-                    });
-                } else {
-                    info!(instance_id = %id, "Skip disabled target");
-                }
+            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();
         while let Some((id, result)) = tasks.next().await {
             match result {
@@ -317,8 +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

@@ -24,6 +24,14 @@ use std::sync::Arc;
 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 {
@@ -537,8 +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;
                                     }
@@ -552,6 +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();
                     }
                 }
@@ -664,6 +679,25 @@ impl AuditSystem {
         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

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 {

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

@@ -51,7 +51,7 @@ impl InternodeMetrics {
             return;
         }
         self.sent_bytes_total.fetch_add(bytes, Ordering::Relaxed);
-        counter!("rustfs.internode.sent.bytes.total").increment(bytes);
+        counter!("rustfs_system_network_internode_sent_bytes_total").increment(bytes);
     }
 
     pub fn record_recv_bytes(&self, bytes: usize) {
@@ -60,22 +60,22 @@ impl InternodeMetrics {
             return;
         }
         self.recv_bytes_total.fetch_add(bytes, Ordering::Relaxed);
-        counter!("rustfs.internode.recv.bytes.total").increment(bytes);
+        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.internode.requests.outgoing.total").increment(1);
+        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.internode.requests.incoming.total").increment(1);
+        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.internode.errors.total").increment(1);
+        counter!("rustfs_system_network_internode_errors_total").increment(1);
     }
 
     pub fn record_dial_result(&self, duration: Duration, success: bool) {
@@ -83,11 +83,11 @@ impl InternodeMetrics {
         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.internode.dial.avg_time.nanos").set(total as f64 / samples as f64);
+        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.internode.dial.errors.total").increment(1);
+            counter!("rustfs_system_network_internode_dial_errors_total").increment(1);
         }
 
         let now_ms = SystemTime::now()
@@ -101,11 +101,7 @@ impl InternodeMetrics {
     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 = if dial_samples_total == 0 {
-            0
-        } else {
-            dial_total_time_nanos / dial_samples_total
-        };
+        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),

crates/concurrency/Cargo.toml

@@ -26,7 +26,7 @@ thiserror = { workspace = true }
 tracing = { workspace = true }
 
 [dev-dependencies]
-tokio = { workspace = true, features = ["full"] }
+tokio = { workspace = true, features = ["test-util","macros","rt-multi-thread"] }
 
 [features]
 default = ["timeout", "lock", "deadlock", "backpressure", "scheduler"]

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/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

@@ -131,6 +131,18 @@ 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";
 
@@ -284,9 +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

@@ -39,6 +39,9 @@ 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";
 
@@ -89,6 +92,10 @@ pub const DEFAULT_STAT_TIMEOUT_SECS: u64 = 3;
 /// 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;
@@ -130,6 +137,7 @@ mod tests {
         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");
@@ -147,6 +155,7 @@ mod tests {
         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);

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

@@ -17,8 +17,11 @@ 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;

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";
@@ -34,6 +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";
@@ -46,12 +48,13 @@ pub const ENV_IDENTITY_OPENID_EMAIL_CLAIM: &str = "RUSTFS_IDENTITY_OPENID_EMAIL_
 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; 15] = &[
+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,
@@ -71,6 +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,

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

@@ -59,9 +59,28 @@ 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

@@ -40,6 +40,39 @@ 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.

crates/config/src/lib.rs

@@ -25,10 +25,16 @@ 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::*;

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/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");

crates/credentials/src/credentials.rs

@@ -20,7 +20,7 @@ use std::collections::HashMap;
 use std::env;
 use std::fmt;
 use std::io::Error;
-use std::sync::OnceLock;
+use std::sync::{LazyLock, OnceLock};
 use time::OffsetDateTime;
 
 /// Global active credentials
@@ -314,6 +314,17 @@ impl fmt::Debug for Credentials {
 }
 
 impl Credentials {
+    /// Returns a reference to this credential's claims, or a shared empty map
+    /// when the credential has no claims attached. Avoids per-call allocation
+    /// at call sites that need an `&HashMap<String, Value>`.
+    pub fn claims_or_empty(&self) -> &HashMap<String, Value> {
+        static EMPTY: LazyLock<HashMap<String, Value>> = LazyLock::new(HashMap::new);
+        match &self.claims {
+            Some(c) => c,
+            None => &EMPTY,
+        }
+    }
+
     pub fn is_expired(&self) -> bool {
         if self.expiration.is_none() {
             return false;

crates/e2e_test/Cargo.toml

@@ -41,7 +41,6 @@ serde_json.workspace = true
 tonic = { workspace = true }
 tokio = { workspace = true }
 tokio-stream = { workspace = true }
-clap = { workspace = true }
 rustfs-madmin.workspace = true
 rustfs-filemeta.workspace = true
 bytes.workspace = true
@@ -53,8 +52,6 @@ async-compression = { workspace = true, features = ["tokio", "bzip2", "xz"] }
 async-trait = { workspace = true }
 flate2.workspace = true
 http.workspace = true
-http-body.workspace = true
-http-body-util.workspace = true
 reqwest = { workspace = true }
 rustfs-signer.workspace = true
 tracing = { workspace = true }
@@ -76,3 +73,4 @@ rcgen.workspace = true
 anyhow.workspace = true
 rustls.workspace = true
 zip.workspace = true
+clap.workspace = true

crates/e2e_test/src/archive_download_integrity_test.rs

@@ -218,6 +218,34 @@ mod tests {
         Ok(builder.send().await?)
     }
 
+    async fn signed_get_request_with_headers(
+        url: &str,
+        access_key: &str,
+        secret_key: &str,
+        extra_headers: &[(&str, &str)],
+    ) -> Result<reqwest::Response, Box<dyn Error + Send + Sync>> {
+        let uri = url.parse::<http::Uri>()?;
+        let authority = uri.authority().ok_or("request URL missing authority")?.to_string();
+        let mut request = http::Request::builder()
+            .method(http::Method::GET)
+            .uri(uri)
+            .header(HOST, authority)
+            .header("x-amz-content-sha256", UNSIGNED_PAYLOAD);
+        for (name, value) in extra_headers {
+            request = request.header(*name, *value);
+        }
+
+        let signed = sign_v4(request.body(Body::empty())?, 0, access_key, secret_key, "", "us-east-1");
+
+        let client = local_http_client();
+        let mut builder = client.get(url);
+        for (name, value) in signed.headers() {
+            builder = builder.header(name, value);
+        }
+
+        Ok(builder.send().await?)
+    }
+
     async fn assert_archive_object_content_encoding(
         client: &S3Client,
         bucket: &str,
@@ -655,6 +683,42 @@ mod tests {
         Ok(())
     }
 
+    #[tokio::test]
+    #[serial]
+    async fn test_multipart_get_ignores_empty_conditional_etag_headers() -> Result<(), Box<dyn Error + Send + Sync>> {
+        init_logging();
+        let mut env = RustFSTestEnvironment::new().await?;
+        env.start_rustfs_server(vec![]).await?;
+        env.create_test_bucket(MULTIPART_ARCHIVE_TEST_BUCKET).await?;
+
+        let client = env.create_s3_client();
+        let key = "multipart-empty-conditional-headers.zip";
+        let zip_bytes =
+            complete_archive_multipart_upload_with_content_encoding(&client, MULTIPART_ARCHIVE_TEST_BUCKET, key, None).await?;
+        let object_url = format!("{}/{}/{}", env.url, MULTIPART_ARCHIVE_TEST_BUCKET, key);
+
+        let response = signed_get_request_with_headers(
+            &object_url,
+            &env.access_key,
+            &env.secret_key,
+            &[("if-match", ""), ("if-none-match", "")],
+        )
+        .await?;
+        let status = response.status();
+        let body = response.bytes().await?;
+
+        assert_eq!(
+            status,
+            StatusCode::OK,
+            "unexpected multipart GET status {status}, body: {}",
+            String::from_utf8_lossy(body.as_ref())
+        );
+        assert_eq!(body.as_ref(), zip_bytes.as_slice());
+
+        env.stop_server();
+        Ok(())
+    }
+
     #[tokio::test]
     #[serial]
     async fn test_archive_multipart_with_aws_chunked_and_effective_encoding_roundtrips_by_default()

crates/e2e_test/src/bin/tls_gen.rs

@@ -12,12 +12,11 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-pub mod collectors;
-pub mod constants;
-pub mod format;
-mod global;
-mod metrics_type;
+use clap::Parser;
+use e2e_test::tls_gen::{Args, run};
 
-pub use format::report_metrics;
-pub use global::init_metrics_system;
-pub use metrics_type::*;
+fn main() -> anyhow::Result<()> {
+    let out_dir = run(Args::parse())?;
+    println!("Generated RustFS TLS bundle in {}", out_dir.display());
+    Ok(())
+}

crates/e2e_test/src/kms/test_runner.rs

@@ -450,11 +450,7 @@ impl KMSTestSuite {
         if failed > 0 {
             warn!("❌ Failing tests:");
             for result in results.iter().filter(|r| !r.success) {
-                warn!(
-                    "  - {}: {}",
-                    result.test_name,
-                    result.error_message.as_ref().unwrap_or(&"Unknown error".to_string())
-                );
+                warn!("  - {}: {}", result.test_name, result.error_message.as_deref().unwrap_or("Unknown error"));
             }
         }
     }

crates/e2e_test/src/lib.rs

@@ -125,3 +125,5 @@ mod replication_extension_test;
 
 #[cfg(test)]
 mod snowball_auto_extract_test;
+
+pub mod tls_gen;

crates/e2e_test/src/list_object_versions_regression_test.rs

@@ -19,6 +19,7 @@
 mod tests {
     use crate::common::{RustFSTestEnvironment, init_logging};
     use aws_sdk_s3::Client;
+    use aws_sdk_s3::primitives::ByteStream;
     use aws_sdk_s3::types::{BucketVersioningStatus, VersioningConfiguration};
     use serial_test::serial;
     use tracing::info;
@@ -179,4 +180,84 @@ mod tests {
             "Delete marker should no longer be latest after the second put"
         );
     }
+
+    #[tokio::test]
+    #[serial]
+    async fn test_list_object_versions_prefix_with_marker_object_returns_children() {
+        init_logging();
+        info!("🧪 TEST: ListObjectVersions returns prefix children when a marker object also exists");
+
+        let mut env = RustFSTestEnvironment::new().await.expect("Failed to create test environment");
+        env.start_rustfs_server(vec![]).await.expect("Failed to start RustFS");
+
+        let client = create_s3_client(&env);
+        let bucket = "test-list-versions-prefix-marker";
+        let marker_key = "data01";
+        let child_keys = [
+            "data01/meta/dump-2026-04-08-053205.json.gz",
+            "data01/meta/dump-2026-04-08-063209.json.gz",
+        ];
+
+        client
+            .create_bucket()
+            .bucket(bucket)
+            .send()
+            .await
+            .expect("Failed to create bucket");
+
+        client
+            .put_bucket_versioning()
+            .bucket(bucket)
+            .versioning_configuration(
+                VersioningConfiguration::builder()
+                    .status(BucketVersioningStatus::Suspended)
+                    .build(),
+            )
+            .send()
+            .await
+            .expect("Failed to suspend versioning");
+
+        client
+            .put_object()
+            .bucket(bucket)
+            .key(marker_key)
+            .body(ByteStream::from_static(b""))
+            .send()
+            .await
+            .expect("Failed to put marker object");
+
+        for key in child_keys {
+            client
+                .put_object()
+                .bucket(bucket)
+                .key(key)
+                .body(ByteStream::from_static(b"payload"))
+                .send()
+                .await
+                .expect("Failed to put child object");
+        }
+
+        let listing = client
+            .list_object_versions()
+            .bucket(bucket)
+            .prefix("data01/")
+            .send()
+            .await
+            .expect("Failed to list object versions by prefix");
+
+        let version_keys: Vec<_> = listing.versions().iter().filter_map(|version| version.key()).collect();
+
+        assert_eq!(
+            version_keys.len(),
+            child_keys.len(),
+            "ListObjectVersions with a trailing slash prefix should include child objects even when the marker object exists"
+        );
+
+        for key in child_keys {
+            assert!(
+                version_keys.contains(&key),
+                "ListObjectVersions(prefix=data01/) should include child object {key}"
+            );
+        }
+    }
 }

crates/e2e_test/src/reliant/conditional_writes.rs

@@ -259,7 +259,7 @@ async fn test_conditional_multi_part_upload() -> Result<(), Box<dyn std::error::
 
     let upload_id = initiate_response
         .upload_id()
-        .ok_or(std::io::Error::other("No upload ID returned"))?;
+        .ok_or_else(|| std::io::Error::other("No upload ID returned"))?;
 
     // Upload parts
     for part_number in 1..=num_parts {
@@ -277,7 +277,7 @@ async fn test_conditional_multi_part_upload() -> Result<(), Box<dyn std::error::
 
         let part_etag = upload_part_response
             .e_tag()
-            .ok_or(std::io::Error::other("Do not have etag"))?
+            .ok_or_else(|| std::io::Error::other("Do not have etag"))?
             .to_string();
 
         let completed_part = CompletedPart::builder().part_number(part_number).e_tag(part_etag).build();

crates/e2e_test/src/reliant/grpc_lock_client.rs

@@ -120,11 +120,6 @@ impl LockClient for GrpcLockClient {
             .map_err(|e| LockError::internal(e.to_string()))?
             .into_inner();
 
-        // Check for explicit error first
-        if let Some(error_info) = resp.error_info {
-            return Err(LockError::internal(error_info));
-        }
-
         // Check if the lock acquisition was successful
         if resp.success {
             Ok(LockResponse::success(
@@ -134,7 +129,8 @@ impl LockClient for GrpcLockClient {
         } else {
             // Lock acquisition failed
             Ok(LockResponse::failure(
-                "Lock acquisition failed on remote server".to_string(),
+                resp.error_info
+                    .unwrap_or_else(|| "Lock acquisition failed on remote server".to_string()),
                 std::time::Duration::ZERO,
             ))
         }

crates/e2e_test/src/reliant/grpc_lock_server.rs

@@ -50,6 +50,18 @@ fn lock_result_from_error(error: impl Into<String>) -> GenerallyLockResult {
     }
 }
 
+fn lock_result_from_release(lock_id: &rustfs_lock::LockId, success: bool) -> GenerallyLockResult {
+    if success {
+        GenerallyLockResult {
+            success: true,
+            error_info: None,
+            lock_info: None,
+        }
+    } else {
+        lock_result_from_error(format!("lock not found for release: {lock_id}"))
+    }
+}
+
 /// Minimal NodeService implementation that only supports Lock RPCs
 /// Used for testing distributed lock scenarios with real gRPC
 #[derive(Debug)]
@@ -102,7 +114,7 @@ impl NodeService for MinimalLockNodeService {
                 let lock_info_json = result.lock_info.as_ref().and_then(|info| serde_json::to_string(info).ok());
                 Ok(Response::new(GenerallyLockResponse {
                     success: result.success,
-                    error_info: None,
+                    error_info: result.error,
                     lock_info: lock_info_json,
                 }))
             }
@@ -131,11 +143,14 @@ impl NodeService for MinimalLockNodeService {
         };
 
         match self.lock_client.release(&args.lock_id).await {
-            Ok(success) => Ok(Response::new(GenerallyLockResponse {
-                success,
-                error_info: None,
-                lock_info: None,
-            })),
+            Ok(success) => {
+                let result = lock_result_from_release(&args.lock_id, success);
+                Ok(Response::new(GenerallyLockResponse {
+                    success: result.success,
+                    error_info: result.error_info,
+                    lock_info: None,
+                }))
+            }
             Err(err) => Ok(Response::new(GenerallyLockResponse {
                 success: false,
                 error_info: Some(format!(
@@ -161,11 +176,14 @@ impl NodeService for MinimalLockNodeService {
         };
 
         match self.lock_client.force_release(&args.lock_id).await {
-            Ok(success) => Ok(Response::new(GenerallyLockResponse {
-                success,
-                error_info: None,
-                lock_info: None,
-            })),
+            Ok(success) => {
+                let result = lock_result_from_release(&args.lock_id, success);
+                Ok(Response::new(GenerallyLockResponse {
+                    success: result.success,
+                    error_info: result.error_info,
+                    lock_info: None,
+                }))
+            }
             Err(err) => Ok(Response::new(GenerallyLockResponse {
                 success: false,
                 error_info: Some(format!(
@@ -271,10 +289,9 @@ impl NodeService for MinimalLockNodeService {
                 Ok(batch_results) => {
                     for (result_idx, success) in batch_results.into_iter().enumerate() {
                         if let Some(request_idx) = valid_indices.get(result_idx) {
-                            results[*request_idx] = GenerallyLockResult {
-                                success,
-                                error_info: None,
-                                lock_info: None,
+                            results[*request_idx] = match lock_ids.get(result_idx) {
+                                Some(lock_id) => lock_result_from_release(lock_id, success),
+                                None => lock_result_from_error(format!("unlock response index out of range: {result_idx}")),
                             };
                         }
                     }

crates/e2e_test/src/reliant/head_tls_bodyless_test.rs

@@ -0,0 +1,201 @@
+// 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.
+
+//! Regression test for TLS/HTTP2 `HEAD` responses on missing objects.
+//!
+//! Before the fix, RustFS returned `404` for a missing object but still wrote
+//! the XML error payload on a `HEAD` request. Under HTTP/2 this emitted DATA
+//! frames after the response headers, which clients surfaced as a protocol
+//! error. This test keeps the request at the raw HTTPS layer so it can validate
+//! the final wire-facing behavior rather than SDK-level error mapping.
+
+#![cfg(test)]
+
+use crate::common::{RustFSTestEnvironment, init_logging, rustfs_binary_path};
+use http::Version;
+use http::header::HOST;
+use rcgen::generate_simple_self_signed;
+use reqwest::{Certificate, Client, Response, StatusCode};
+use rustfs_signer::constants::UNSIGNED_PAYLOAD;
+use rustfs_signer::sign_v4;
+use s3s::Body;
+use serial_test::serial;
+use std::error::Error;
+use std::path::Path;
+use std::process::Command;
+use tokio::fs;
+use tokio::time::{Duration, sleep};
+use tracing::info;
+
+const ACCESS_KEY: &str = "rustfsadmin";
+const SECRET_KEY: &str = "rustfsadmin";
+const BUCKET: &str = "test-head-tls-bodyless-bucket";
+
+async fn generate_tls_bundle(tls_dir: &Path) -> Result<Vec<u8>, Box<dyn Error + Send + Sync>> {
+    fs::create_dir_all(tls_dir).await?;
+    let cert = generate_simple_self_signed(vec!["localhost".to_string(), "127.0.0.1".to_string()])?;
+    let cert_pem = cert.cert.pem();
+    let key_pem = cert.signing_key.serialize_pem();
+
+    fs::write(tls_dir.join("rustfs_cert.pem"), cert_pem.as_bytes()).await?;
+    fs::write(tls_dir.join("rustfs_key.pem"), key_pem.as_bytes()).await?;
+
+    Ok(cert_pem.into_bytes())
+}
+
+fn local_https_h2_client(ca_pem: &[u8]) -> Result<Client, Box<dyn Error + Send + Sync>> {
+    let _ca_cert = Certificate::from_pem(ca_pem)?;
+    Ok(Client::builder()
+        .no_proxy()
+        .no_gzip()
+        .no_brotli()
+        .no_zstd()
+        .no_deflate()
+        .danger_accept_invalid_certs(true)
+        .build()?)
+}
+
+async fn signed_empty_request(
+    client: &Client,
+    method: http::Method,
+    url: &str,
+) -> Result<Response, Box<dyn Error + Send + Sync>> {
+    let uri = url.parse::<http::Uri>()?;
+    let authority = uri.authority().ok_or("request URL missing authority")?.to_string();
+    let request = http::Request::builder()
+        .method(method.as_str())
+        .uri(uri)
+        .header(HOST, authority)
+        .header("x-amz-content-sha256", UNSIGNED_PAYLOAD)
+        .body(Body::empty())?;
+
+    let signed = sign_v4(request, 0, ACCESS_KEY, SECRET_KEY, "", "us-east-1");
+
+    let reqwest_method = reqwest::Method::from_bytes(method.as_str().as_bytes())?;
+    let mut builder = client.request(reqwest_method, url);
+    for (name, value) in signed.headers() {
+        builder = builder.header(name, value);
+    }
+
+    Ok(builder.send().await?)
+}
+
+async fn ensure_bucket_exists(client: &Client, endpoint: &str) -> Result<(), Box<dyn Error + Send + Sync>> {
+    let bucket_url = format!("{endpoint}/{BUCKET}/");
+    let response = signed_empty_request(client, http::Method::HEAD, &bucket_url).await?;
+
+    if response.status() == StatusCode::OK {
+        return Ok(());
+    }
+
+    let response = signed_empty_request(client, http::Method::PUT, &bucket_url).await?;
+    match response.status() {
+        StatusCode::OK => Ok(()),
+        StatusCode::CONFLICT => Ok(()),
+        status => Err(format!("unexpected bucket setup status: {status}").into()),
+    }
+}
+
+async fn wait_for_tls_server_ready(client: &Client, endpoint: &str) -> Result<(), Box<dyn Error + Send + Sync>> {
+    let ready_url = format!("{endpoint}/");
+    for _attempt in 0..60 {
+        match signed_empty_request(client, http::Method::GET, &ready_url).await {
+            Ok(response) if response.status().is_success() => return Ok(()),
+            Ok(_) | Err(_) => sleep(Duration::from_millis(500)).await,
+        }
+    }
+
+    Err("RustFS TLS server failed to become ready within 30 seconds".into())
+}
+
+async fn start_tls_rustfs_server(env: &mut RustFSTestEnvironment, tls_dir: &Path) -> Result<(), Box<dyn Error + Send + Sync>> {
+    let binary_path = rustfs_binary_path();
+    let mut command = Command::new(&binary_path);
+    command
+        .env("RUST_LOG", "rustfs=info,rustfs_notify=debug")
+        .env("RUSTFS_TLS_PATH", tls_dir)
+        .current_dir(&env.temp_dir);
+
+    for key in [
+        "RUSTFS_ADDRESS",
+        "RUSTFS_VOLUMES",
+        "RUSTFS_ACCESS_KEY",
+        "RUSTFS_SECRET_KEY",
+        "RUSTFS_TLS_PATH",
+        "RUSTFS_OBS_LOG_DIRECTORY",
+    ] {
+        command.env_remove(key);
+    }
+
+    let process = command
+        .env("RUSTFS_TLS_PATH", tls_dir)
+        .env("RUSTFS_CONSOLE_ENABLE", "false")
+        .args([
+            "--address",
+            &env.address,
+            "--access-key",
+            &env.access_key,
+            "--secret-key",
+            &env.secret_key,
+            &env.temp_dir,
+        ])
+        .spawn()?;
+
+    env.process = Some(process);
+    Ok(())
+}
+
+#[tokio::test]
+#[serial]
+async fn test_head_missing_object_over_tls_http2_is_bodyless() -> Result<(), Box<dyn Error + Send + Sync>> {
+    init_logging();
+
+    let mut env = RustFSTestEnvironment::new().await?;
+    let tls_dir = std::path::PathBuf::from(&env.temp_dir).join("tls");
+    let ca_pem = generate_tls_bundle(&tls_dir).await?;
+    start_tls_rustfs_server(&mut env, &tls_dir).await?;
+
+    let endpoint = format!("https://{}", env.address);
+    let client = local_https_h2_client(&ca_pem)?;
+    wait_for_tls_server_ready(&client, &endpoint).await?;
+    ensure_bucket_exists(&client, &endpoint).await?;
+
+    let missing_key = "head-does-not-exist.txt";
+    let object_url = format!("{endpoint}/{BUCKET}/{missing_key}");
+
+    let get_response = signed_empty_request(&client, http::Method::GET, &object_url).await?;
+    assert_eq!(get_response.status(), StatusCode::NOT_FOUND);
+    let get_version = get_response.version();
+    let get_body = get_response.bytes().await?;
+    let get_body_text = String::from_utf8_lossy(&get_body);
+    assert!(
+        get_body_text.contains("<Code>NoSuchKey</Code>") || get_body_text.contains("<Code>NoSuchObject</Code>"),
+        "GET missing-object error body should expose NoSuchKey/NoSuchObject, got: {}",
+        get_body_text
+    );
+    info!("GET missing object over TLS used {:?} and returned {} bytes", get_version, get_body.len());
+
+    let head_response = signed_empty_request(&client, http::Method::HEAD, &object_url).await?;
+    assert_eq!(head_response.status(), StatusCode::NOT_FOUND);
+    assert_eq!(head_response.version(), Version::HTTP_2, "HEAD regression test must exercise HTTP/2");
+    let head_body = head_response.bytes().await?;
+    assert!(
+        head_body.is_empty(),
+        "HEAD missing-object response must not send body bytes over TLS/HTTP2, got {} bytes: {:?}",
+        head_body.len(),
+        head_body
+    );
+
+    Ok(())
+}

crates/e2e_test/src/reliant/lifecycle.rs

@@ -160,7 +160,7 @@ async fn test_bucket_lifecycle_configuration() -> Result<(), Box<dyn std::error:
 #[tokio::test(flavor = "multi_thread", worker_threads = 2)]
 #[serial]
 #[ignore = "requires running RustFS server at localhost:9000"]
-async fn test_bucket_lifecycle_rejects_zero_days() -> Result<(), Box<dyn std::error::Error>> {
+async fn test_bucket_lifecycle_accepts_zero_days() -> Result<(), Box<dyn std::error::Error>> {
     use aws_sdk_s3::types::{BucketLifecycleConfiguration, LifecycleExpiration, LifecycleRule, LifecycleRuleFilter};
 
     let client = create_aws_s3_client().await?;
@@ -176,19 +176,12 @@ async fn test_bucket_lifecycle_rejects_zero_days() -> Result<(), Box<dyn std::er
         .build()?;
     let lifecycle = BucketLifecycleConfiguration::builder().rules(rule).build()?;
 
-    let err = client
+    client
         .put_bucket_lifecycle_configuration()
         .bucket(BUCKET)
         .lifecycle_configuration(lifecycle)
         .send()
-        .await
-        .expect_err("zero-day lifecycle expiration should be rejected");
-
-    let err_msg = format!("{err:?}");
-    assert!(
-        err_msg.contains("InvalidArgument") && err_msg.contains("greater than 0"),
-        "unexpected error: {err_msg}"
-    );
+        .await?;
 
     Ok(())
 }

crates/e2e_test/src/reliant/lock.rs

@@ -275,6 +275,41 @@ async fn test_grpc_lock_client_batch_acquire_and_release() {
     handle.abort();
 }
 
+#[tokio::test]
+async fn test_grpc_lock_client_uses_request_lock_id_and_reports_missing_unlock() {
+    let manager = Arc::new(GlobalLockManager::new());
+    let local_client: Arc<dyn rustfs_lock::LockClient> = Arc::new(LocalClient::with_manager(manager));
+
+    let (addr, handle) = spawn_lock_server(local_client).await.expect("Failed to spawn server");
+    tokio::time::sleep(Duration::from_millis(100)).await;
+
+    let grpc_client = GrpcLockClient::new(addr);
+    let request = LockRequest::new(test_resource(), LockType::Exclusive, "owner-a").with_acquire_timeout(Duration::from_secs(2));
+
+    let response = grpc_client.acquire_lock(&request).await.expect("gRPC acquire should succeed");
+    let lock_info = response.lock_info.expect("gRPC acquire should include lock info");
+    assert_eq!(lock_info.id, request.lock_id);
+
+    assert!(
+        grpc_client
+            .release(&request.lock_id)
+            .await
+            .expect("gRPC release should succeed"),
+        "release should find the request lock id"
+    );
+
+    let missing_release = grpc_client
+        .release(&request.lock_id)
+        .await
+        .expect_err("second release should report missing lock");
+    assert!(
+        missing_release.to_string().contains("lock not found for release"),
+        "missing release should preserve server error, got: {missing_release}"
+    );
+
+    handle.abort();
+}
+
 #[tokio::test]
 async fn test_distributed_lock_4_nodes_grpc_read_write_quorum_split_with_two_failed_nodes() {
     let manager1 = Arc::new(GlobalLockManager::new());

crates/e2e_test/src/reliant/mod.rs

@@ -17,6 +17,7 @@ mod get_deleted_object_test;
 mod grpc_lock_client;
 mod grpc_lock_server;
 mod head_deleted_object_versioning_test;
+mod head_tls_bodyless_test;
 mod lifecycle;
 mod lock;
 mod node_interact_test;

crates/e2e_test/src/reliant/node_interact_test.rs

@@ -154,7 +154,7 @@ async fn walk_dir() -> Result<(), Box<dyn Error>> {
             match response.next().await {
                 Some(Ok(resp)) => {
                     if !resp.success {
-                        println!("{}", resp.error_info.unwrap_or("".to_string()));
+                        println!("{}", resp.error_info.unwrap_or_else(|| "".to_string()));
                     }
                     let entry = serde_json::from_str::<MetaCacheEntry>(&resp.meta_cache_entry)
                         .map_err(|_e| std::io::Error::other(format!("Unexpected response: {response:?}")))

crates/e2e_test/src/special_chars_test.rs

@@ -26,10 +26,16 @@
 
 #[cfg(test)]
 mod tests {
-    use crate::common::{RustFSTestEnvironment, init_logging};
+    use crate::common::{RustFSTestEnvironment, init_logging, local_http_client};
     use aws_sdk_s3::Client;
     use aws_sdk_s3::primitives::ByteStream;
+    use http::StatusCode;
+    use http::header::HOST;
+    use rustfs_signer::constants::UNSIGNED_PAYLOAD;
+    use rustfs_signer::sign_v4;
+    use s3s::Body;
     use serial_test::serial;
+    use std::error::Error;
     use tracing::{debug, info};
 
     /// Helper function to create an S3 client for testing
@@ -56,6 +62,30 @@ mod tests {
         }
     }
 
+    async fn signed_get(
+        url: &str,
+        access_key: &str,
+        secret_key: &str,
+    ) -> Result<reqwest::Response, Box<dyn Error + Send + Sync>> {
+        let uri = url.parse::<http::Uri>()?;
+        let authority = uri.authority().ok_or("request URL missing authority")?.to_string();
+        let request = http::Request::builder()
+            .method(http::Method::GET)
+            .uri(uri)
+            .header(HOST, authority)
+            .header("x-amz-content-sha256", UNSIGNED_PAYLOAD)
+            .body(Body::empty())?;
+
+        let signed = sign_v4(request, 0, access_key, secret_key, "", "us-east-1");
+        let client = local_http_client();
+        let mut request_builder = client.get(url);
+        for (name, value) in signed.headers() {
+            request_builder = request_builder.header(name, value);
+        }
+
+        Ok(request_builder.send().await?)
+    }
+
     /// Test PUT and GET with space character in path
     ///
     /// This reproduces Part A of the issue:
@@ -274,6 +304,73 @@ mod tests {
         info!("Test completed successfully");
     }
 
+    #[tokio::test]
+    #[serial]
+    async fn test_signed_get_missing_object_with_trailing_equals_returns_no_such_key() -> Result<(), Box<dyn Error + Send + Sync>>
+    {
+        init_logging();
+
+        let mut env = RustFSTestEnvironment::new().await?;
+        env.start_rustfs_server(vec![]).await?;
+
+        let client = create_s3_client(&env);
+        let bucket = "test-missing-equals-key";
+        create_bucket(&client, bucket).await?;
+
+        let url = format!("{}/{}/path/sitemap.xmlage=", env.url, bucket);
+        let response = signed_get(&url, &env.access_key, &env.secret_key).await?;
+
+        assert_eq!(
+            response.status(),
+            StatusCode::NOT_FOUND,
+            "missing object key ending with '=' should pass signature validation before object lookup"
+        );
+
+        let body = response.text().await?;
+        assert!(body.contains("<Code>NoSuchKey</Code>"), "expected NoSuchKey XML response, got: {body}");
+
+        env.stop_server();
+        Ok(())
+    }
+
+    #[tokio::test]
+    #[serial]
+    async fn test_signed_get_existing_object_with_trailing_equals_returns_content() -> Result<(), Box<dyn Error + Send + Sync>> {
+        init_logging();
+
+        let mut env = RustFSTestEnvironment::new().await?;
+        env.start_rustfs_server(vec![]).await?;
+
+        let client = create_s3_client(&env);
+        let bucket = "test-existing-equals-key";
+        create_bucket(&client, bucket).await?;
+
+        let key = "path/sitemap.xmlage=";
+        let content = b"object content for raw signed URL with trailing equals";
+        client
+            .put_object()
+            .bucket(bucket)
+            .key(key)
+            .body(ByteStream::from_static(content))
+            .send()
+            .await?;
+
+        let url = format!("{}/{}/{}", env.url, bucket, key);
+        let response = signed_get(&url, &env.access_key, &env.secret_key).await?;
+
+        assert_eq!(
+            response.status(),
+            StatusCode::OK,
+            "existing object key ending with '=' should pass signature validation and return content"
+        );
+
+        let body = response.bytes().await?;
+        assert_eq!(body.as_ref(), content);
+
+        env.stop_server();
+        Ok(())
+    }
+
     /// Test DELETE operation with special characters
     #[tokio::test]
     #[serial]

crates/e2e_test/src/tls_gen.rs

@@ -0,0 +1,259 @@
+// 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 anyhow::{Context, Result, bail};
+use clap::Parser;
+use rcgen::{
+    BasicConstraints, CertificateParams, CertifiedIssuer, DnType, ExtendedKeyUsagePurpose, IsCa, KeyPair, KeyUsagePurpose,
+    SanType,
+};
+use std::fs;
+use std::net::IpAddr;
+use std::path::{Path, PathBuf};
+use time::{Duration, OffsetDateTime};
+
+pub const DEFAULT_OUT_DIR: &str = "target/tls";
+pub const OUTPUT_FILES: [&str; 7] = [
+    "rustfs_cert.pem",
+    "rustfs_key.pem",
+    "ca.crt",
+    "public.crt",
+    "client_ca.crt",
+    "client_cert.pem",
+    "client_key.pem",
+];
+
+#[derive(Debug, Parser)]
+#[command(name = "tls_gen", about = "Generate a full RustFS TLS bundle for local TLS and mTLS tests.")]
+pub struct Args {
+    #[arg(long, default_value = DEFAULT_OUT_DIR)]
+    pub out_dir: PathBuf,
+    #[arg(long, default_value_t = 365)]
+    pub days: i64,
+    #[arg(long)]
+    pub force: bool,
+}
+
+pub fn run(args: Args) -> Result<PathBuf> {
+    if args.days <= 0 {
+        bail!("--days must be a positive integer");
+    }
+
+    write_bundle(&args.out_dir, args.force, args.days)?;
+    Ok(args.out_dir)
+}
+
+pub fn ensure_writable(out_dir: &Path, force: bool) -> Result<()> {
+    if force {
+        return Ok(());
+    }
+
+    let existing: Vec<_> = OUTPUT_FILES
+        .iter()
+        .map(|name| out_dir.join(name))
+        .filter(|path| path.exists())
+        .collect();
+
+    if existing.is_empty() {
+        return Ok(());
+    }
+
+    let existing_list = existing
+        .iter()
+        .map(|path| path.file_name().and_then(|name| name.to_str()).unwrap_or("<unknown>"))
+        .collect::<Vec<_>>()
+        .join(", ");
+
+    bail!(
+        "Refusing to overwrite existing files in {}: {}. Re-run with --force to replace them.",
+        out_dir.display(),
+        existing_list
+    )
+}
+
+fn write_bundle(out_dir: &Path, force: bool, days: i64) -> Result<()> {
+    fs::create_dir_all(out_dir).with_context(|| format!("failed to create output directory {}", out_dir.display()))?;
+    ensure_writable(out_dir, force)?;
+
+    let ca_key = generate_private_key()?;
+    let ca = build_ca_certificate(ca_key, days)?;
+
+    let server_key = generate_private_key()?;
+    let server_cert = build_leaf_certificate(
+        &server_key,
+        "localhost",
+        &[SanType::DnsName("localhost".try_into()?)],
+        &[
+            SanType::IpAddress(IpAddr::V4("127.0.0.1".parse()?)),
+            SanType::IpAddress(IpAddr::V6("::1".parse()?)),
+        ],
+        ExtendedKeyUsagePurpose::ServerAuth,
+        &ca,
+        days,
+    )?;
+
+    let client_key = generate_private_key()?;
+    let client_cert = build_leaf_certificate(
+        &client_key,
+        "rustfs-test-client",
+        &[SanType::DnsName("rustfs-test-client".try_into()?)],
+        &[],
+        ExtendedKeyUsagePurpose::ClientAuth,
+        &ca,
+        days,
+    )?;
+
+    let ca_pem = ca.pem();
+    let bundle = [
+        ("rustfs_cert.pem", server_cert.pem()),
+        ("rustfs_key.pem", server_key.serialize_pem()),
+        ("ca.crt", ca_pem.clone()),
+        ("public.crt", ca_pem.clone()),
+        ("client_ca.crt", ca_pem),
+        ("client_cert.pem", client_cert.pem()),
+        ("client_key.pem", client_key.serialize_pem()),
+    ];
+
+    for (name, content) in bundle {
+        fs::write(out_dir.join(name), content).with_context(|| format!("failed to write {}", out_dir.join(name).display()))?;
+    }
+
+    Ok(())
+}
+
+fn build_ca_certificate(signing_key: KeyPair, days: i64) -> Result<CertifiedIssuer<'static, KeyPair>> {
+    let mut params = base_params("RustFS Test CA", days)?;
+    params.is_ca = IsCa::Ca(BasicConstraints::Unconstrained);
+    params.key_usages = vec![KeyUsagePurpose::KeyCertSign, KeyUsagePurpose::CrlSign];
+
+    CertifiedIssuer::self_signed(params, signing_key).context("failed to create CA certificate")
+}
+
+fn build_leaf_certificate(
+    signing_key: &KeyPair,
+    common_name: &str,
+    dns_names: &[SanType],
+    ip_addresses: &[SanType],
+    usage: ExtendedKeyUsagePurpose,
+    issuer: &CertifiedIssuer<'_, KeyPair>,
+    days: i64,
+) -> Result<rcgen::Certificate> {
+    let mut params = base_params(common_name, days)?;
+    params.is_ca = IsCa::ExplicitNoCa;
+    params.key_usages = vec![KeyUsagePurpose::DigitalSignature, KeyUsagePurpose::KeyEncipherment];
+    params.extended_key_usages = vec![usage];
+    params.use_authority_key_identifier_extension = true;
+    params.subject_alt_names.extend_from_slice(dns_names);
+    params.subject_alt_names.extend_from_slice(ip_addresses);
+
+    params
+        .signed_by(signing_key, issuer)
+        .with_context(|| format!("failed to create leaf certificate for {common_name}"))
+}
+
+fn base_params(common_name: &str, days: i64) -> Result<CertificateParams> {
+    let mut params = CertificateParams::default();
+    let issued_at = OffsetDateTime::now_utc() - Duration::minutes(5);
+    params.not_before = issued_at;
+    params.not_after = issued_at + Duration::days(days);
+    params.distinguished_name.push(DnType::CountryName, "US");
+    params.distinguished_name.push(DnType::OrganizationName, "RustFS");
+    params.distinguished_name.push(DnType::CommonName, common_name);
+    Ok(params)
+}
+
+fn generate_private_key() -> Result<KeyPair> {
+    KeyPair::generate().context("failed to generate private key")
+}
+
+#[cfg(test)]
+mod tests {
+    use super::{Args, OUTPUT_FILES, ensure_writable, run};
+    use std::fs;
+    use std::path::{Path, PathBuf};
+    use std::time::{SystemTime, UNIX_EPOCH};
+
+    fn unique_temp_dir() -> PathBuf {
+        let suffix = SystemTime::now()
+            .duration_since(UNIX_EPOCH)
+            .expect("system time must be after unix epoch")
+            .as_nanos();
+        std::env::temp_dir().join(format!("rustfs-tls-gen-{suffix}"))
+    }
+
+    struct TempDir {
+        path: PathBuf,
+    }
+
+    impl TempDir {
+        fn new() -> Self {
+            let path = unique_temp_dir();
+            fs::create_dir_all(&path).expect("temporary directory should be created");
+            Self { path }
+        }
+
+        fn path(&self) -> &Path {
+            &self.path
+        }
+    }
+
+    impl Drop for TempDir {
+        fn drop(&mut self) {
+            let _ = fs::remove_dir_all(&self.path);
+        }
+    }
+
+    #[test]
+    fn run_writes_full_bundle() {
+        let temp_dir = TempDir::new();
+
+        let out_dir = run(Args {
+            out_dir: temp_dir.path().join("tls"),
+            days: 365,
+            force: false,
+        })
+        .expect("bundle generation should succeed");
+
+        for name in OUTPUT_FILES {
+            let content = fs::read(out_dir.join(name)).unwrap_or_else(|error| panic!("{name} should exist: {error}"));
+            assert!(!content.is_empty(), "{name} should not be empty");
+        }
+    }
+
+    #[test]
+    fn ensure_writable_rejects_existing_files_without_force() {
+        let temp_dir = TempDir::new();
+        let existing = temp_dir.path().join(OUTPUT_FILES[0]);
+        fs::write(&existing, "existing").expect("existing file should be created");
+
+        let error = ensure_writable(temp_dir.path(), false).expect_err("existing files must be rejected");
+        let message = format!("{error:#}");
+
+        assert!(message.contains("Refusing to overwrite existing files"));
+        assert!(message.contains(OUTPUT_FILES[0]));
+    }
+
+    #[test]
+    fn run_rejects_non_positive_days() {
+        let temp_dir = TempDir::new();
+        let error = run(Args {
+            out_dir: temp_dir.path().join("tls"),
+            days: 0,
+            force: false,
+        })
+        .expect_err("non-positive days must fail");
+
+        assert_eq!(format!("{error:#}"), "--days must be a positive integer");
+    }
+}

crates/ecstore/Cargo.toml

@@ -55,6 +55,7 @@ flatbuffers.workspace = true
 futures.workspace = true
 futures-util.workspace = true
 tracing.workspace = true
+tracing-opentelemetry.workspace = true
 serde.workspace = true
 time.workspace = true
 bytesize.workspace = true
@@ -62,6 +63,7 @@ serde_json.workspace = true
 quick-xml = { workspace = true, features = ["serialize", "async-tokio"] }
 s3s.workspace = true
 http.workspace = true
+opentelemetry.workspace = true
 http-body = { workspace = true }
 http-body-util.workspace = true
 url.workspace = true
@@ -86,7 +88,7 @@ hyper.workspace = true
 hyper-util.workspace = true
 hyper-rustls.workspace = true
 rustls.workspace = true
-tokio = { workspace = true, features = ["io-util", "sync", "signal"] }
+tokio = { workspace = true, features = ["io-util", "sync", "signal","io-uring"] }
 tonic.workspace = true
 xxhash-rust = { workspace = true, features = ["xxh64", "xxh3"] }
 tower.workspace = true
@@ -98,6 +100,7 @@ pin-project-lite.workspace = true
 md-5.workspace = true
 memmap2 = { workspace = true }
 libc.workspace = true
+rustix = { workspace = true }
 rustfs-madmin.workspace = true
 rustfs-workers.workspace = true
 reqwest = { workspace = true }
@@ -124,9 +127,10 @@ metrics = { workspace = true }
 [dev-dependencies]
 tokio = { workspace = true, features = ["rt-multi-thread", "macros"] }
 criterion = { workspace = true, features = ["html_reports"] }
-temp-env = { workspace = true }
+temp-env = { workspace = true, features = ["async_closure"] }
 tracing-subscriber = { workspace = true }
 serial_test = { workspace = true }
+opentelemetry_sdk = { workspace = true }
 
 [build-dependencies]
 shadow-rs = { workspace = true, features = ["build", "metadata"] }

crates/ecstore/src/admin_server_info.rs

@@ -162,8 +162,9 @@ pub async fn get_local_server_property() -> ServerProperties {
 
     let mut props = ServerProperties {
         endpoint: addr,
-        uptime: SystemTime::now()
-            .duration_since(*GLOBAL_BOOT_TIME.get().unwrap())
+        uptime: GLOBAL_BOOT_TIME
+            .get()
+            .and_then(|boot_time| SystemTime::now().duration_since(*boot_time).ok())
             .unwrap_or_default()
             .as_secs(),
         network,

crates/ecstore/src/batch_processor.rs

@@ -87,25 +87,70 @@ impl AsyncBatchProcessor {
         T: Send + 'static,
         F: Future<Output = Result<T>> + Send + 'static,
     {
-        let results = self.execute_batch(tasks).await;
-        let mut successes = Vec::new();
+        if required_successes == 0 {
+            return Ok(Vec::new());
+        }
 
-        for value in results.into_iter().flatten() {
-            successes.push(value);
-            if successes.len() >= required_successes {
-                return Ok(successes);
+        if tasks.is_empty() {
+            return Err(Error::other(format!(
+                "Insufficient successful results: got 0, needed {required_successes}"
+            )));
+        }
+
+        let semaphore = Arc::new(tokio::sync::Semaphore::new(self.max_concurrent));
+        let mut join_set = JoinSet::new();
+        let mut successes = Vec::new();
+        let mut pending_tasks = tasks.len();
+        let mut first_error = None;
+
+        for task in tasks {
+            let sem = semaphore.clone();
+            join_set.spawn(async move {
+                let _permit = sem.acquire().await.map_err(|_| Error::other("Semaphore error"))?;
+                task.await
+            });
+        }
+
+        while let Some(join_result) = join_set.join_next().await {
+            pending_tasks = pending_tasks.saturating_sub(1);
+
+            match join_result {
+                Ok(Ok(value)) => {
+                    successes.push(value);
+                    if successes.len() >= required_successes {
+                        return Ok(successes);
+                    }
+                }
+                Ok(Err(err)) => {
+                    if first_error.is_none() {
+                        first_error = Some(err);
+                    }
+                }
+                Err(join_error) => {
+                    if first_error.is_none() {
+                        first_error = Some(Error::other(format!("Task panicked in quorum batch processor: {join_error}")));
+                    }
+                }
+            }
+
+            if successes.len() + pending_tasks < required_successes {
+                return Err(first_error.unwrap_or_else(|| {
+                    Error::other(format!(
+                        "Insufficient successful results: got {}, needed {}",
+                        successes.len(),
+                        required_successes
+                    ))
+                }));
             }
         }
 
-        if successes.len() >= required_successes {
-            Ok(successes)
-        } else {
-            Err(Error::other(format!(
+        Err(first_error.unwrap_or_else(|| {
+            Error::other(format!(
                 "Insufficient successful results: got {}, needed {}",
                 successes.len(),
                 required_successes
-            )))
-        }
+            ))
+        }))
     }
 }
 
@@ -228,4 +273,52 @@ mod tests {
         let successes = results.unwrap();
         assert!(successes.len() >= 2);
     }
+
+    #[tokio::test]
+    async fn test_batch_processor_quorum_returns_before_slow_tail() {
+        let processor = AsyncBatchProcessor::new(4);
+        let started = std::time::Instant::now();
+
+        let tasks: Vec<_> = [(10_u64, Ok(1_i32)), (15, Ok(2)), (250, Ok(3))]
+            .into_iter()
+            .map(|(delay_ms, outcome)| async move {
+                tokio::time::sleep(Duration::from_millis(delay_ms)).await;
+                outcome
+            })
+            .collect();
+
+        let results = processor
+            .execute_batch_with_quorum(tasks, 2)
+            .await
+            .expect("quorum should succeed");
+        assert_eq!(results.len(), 2);
+        assert!(started.elapsed() < Duration::from_millis(100));
+    }
+
+    #[tokio::test]
+    async fn test_batch_processor_quorum_fails_once_quorum_becomes_impossible() {
+        let processor = AsyncBatchProcessor::new(4);
+        let started = std::time::Instant::now();
+
+        let tasks: Vec<_> = vec![
+            (10_u64, Ok(1_i32)),
+            (15, Err(Error::other("first failure"))),
+            (20, Err(Error::other("second failure"))),
+            (250, Ok(4)),
+        ]
+        .into_iter()
+        .map(|(delay_ms, outcome)| async move {
+            tokio::time::sleep(Duration::from_millis(delay_ms)).await;
+            outcome
+        })
+        .collect();
+
+        let err = processor
+            .execute_batch_with_quorum(tasks, 3)
+            .await
+            .expect_err("quorum should fail once it becomes impossible");
+
+        assert!(err.to_string().contains("first failure"));
+        assert!(started.elapsed() < Duration::from_millis(120));
+    }
 }

crates/ecstore/src/bucket/bucket_target_sys.rs

@@ -395,8 +395,27 @@ impl BucketTargetSys {
         }
     }
 
-    pub async fn set_target(&self, bucket: &str, target: &BucketTarget, update: bool) -> Result<(), BucketTargetError> {
-        if !target.target_type.is_valid() && !update {
+    pub async fn set_target(
+        &self,
+        bucket: &str,
+        target: &BucketTarget,
+        update: bool,
+    ) -> Result<BucketTargets, BucketTargetError> {
+        self.validate_target(bucket, target).await?;
+
+        let mut bucket_targets = match self.list_bucket_targets(bucket).await {
+            Ok(targets) => targets,
+            Err(BucketTargetError::BucketRemoteTargetNotFound { .. }) => BucketTargets::default(),
+            Err(err) => return Err(err),
+        };
+
+        Self::upsert_target_entry(&mut bucket_targets.targets, target, update)?;
+
+        Ok(bucket_targets)
+    }
+
+    pub async fn validate_target(&self, bucket: &str, target: &BucketTarget) -> Result<(), BucketTargetError> {
+        if !target.target_type.is_valid() {
             return Err(BucketTargetError::BucketRemoteArnTypeInvalid {
                 bucket: bucket.to_string(),
             });
@@ -450,52 +469,44 @@ impl BucketTargetSys {
             }
         }
 
-        {
-            let mut targets_map = self.targets_map.write().await;
-            let bucket_targets = targets_map.entry(bucket.to_string()).or_insert_with(Vec::new);
-            let mut found = false;
+        Ok(())
+    }
 
-            for (idx, existing_target) in bucket_targets.iter().enumerate() {
-                if existing_target.target_type.to_string() == target.target_type.to_string() {
-                    if existing_target.arn == target.arn {
-                        if !update {
-                            return Err(BucketTargetError::BucketRemoteAlreadyExists {
-                                bucket: existing_target.target_bucket.clone(),
-                            });
-                        }
-                        bucket_targets[idx] = target.clone();
-                        found = true;
-                        break;
-                    }
-                    if existing_target.endpoint == target.endpoint {
+    fn upsert_target_entry(
+        bucket_targets: &mut Vec<BucketTarget>,
+        target: &BucketTarget,
+        update: bool,
+    ) -> Result<(), BucketTargetError> {
+        let mut found = false;
+
+        for (idx, existing_target) in bucket_targets.iter().enumerate() {
+            if existing_target.target_type.to_string() == target.target_type.to_string() {
+                if existing_target.arn == target.arn {
+                    if !update {
                         return Err(BucketTargetError::BucketRemoteAlreadyExists {
                             bucket: existing_target.target_bucket.clone(),
                         });
                     }
+                    bucket_targets[idx] = target.clone();
+                    found = true;
+                    break;
+                }
+                if existing_target.endpoint == target.endpoint {
+                    return Err(BucketTargetError::BucketRemoteAlreadyExists {
+                        bucket: existing_target.target_bucket.clone(),
+                    });
                 }
             }
-
-            if !found && !update {
-                bucket_targets.push(target.clone());
-            }
         }
 
-        {
-            let mut arn_remotes_map = self.arn_remotes_map.write().await;
-            arn_remotes_map.insert(
-                target.arn.clone(),
-                ArnTarget {
-                    client: Some(Arc::new(target_client)),
-                    last_refresh: OffsetDateTime::now_utc(),
-                },
-            );
+        if !found && !update {
+            bucket_targets.push(target.clone());
         }
 
-        self.update_bandwidth_limit(bucket, &target.arn, target.bandwidth_limit);
         Ok(())
     }
 
-    pub async fn remove_target(&self, bucket: &str, arn_str: &str) -> Result<(), BucketTargetError> {
+    pub async fn remove_target(&self, bucket: &str, arn_str: &str) -> Result<BucketTargets, BucketTargetError> {
         if arn_str.is_empty() {
             return Err(BucketTargetError::BucketRemoteArnInvalid {
                 bucket: bucket.to_string(),
@@ -524,33 +535,16 @@ impl BucketTargetSys {
             }
         }
 
-        {
-            let mut targets_map = self.targets_map.write().await;
+        let targets = self.list_bucket_targets(bucket).await?;
+        let new_targets: Vec<BucketTarget> = targets.targets.iter().filter(|t| t.arn != arn_str).cloned().collect();
 
-            let Some(targets) = targets_map.get(bucket) else {
-                return Err(BucketTargetError::BucketRemoteTargetNotFound {
-                    bucket: bucket.to_string(),
-                });
-            };
-
-            let new_targets: Vec<BucketTarget> = targets.iter().filter(|t| t.arn != arn_str).cloned().collect();
-
-            if new_targets.len() == targets.len() {
-                return Err(BucketTargetError::BucketRemoteTargetNotFound {
-                    bucket: bucket.to_string(),
-                });
-            }
-
-            targets_map.insert(bucket.to_string(), new_targets);
+        if new_targets.len() == targets.targets.len() {
+            return Err(BucketTargetError::BucketRemoteTargetNotFound {
+                bucket: bucket.to_string(),
+            });
         }
 
-        {
-            self.arn_remotes_map.write().await.remove(arn_str);
-        }
-
-        self.update_bandwidth_limit(bucket, arn_str, 0);
-
-        Ok(())
+        Ok(BucketTargets { targets: new_targets })
     }
 
     pub async fn mark_refresh_in_progress(&self, bucket: &str, arn: &str) {
@@ -603,7 +597,7 @@ impl BucketTargetSys {
 
         if let Some(last_refresh) = last_refresh {
             let now = OffsetDateTime::now_utc();
-            if now - last_refresh > Duration::from_secs(60 * 5) {
+            if now - last_refresh < Duration::from_secs(60 * 5) {
                 return None;
             }
         }
@@ -619,6 +613,16 @@ impl BucketTargetSys {
             }
         };
 
+        let cli = self
+            .arn_remotes_map
+            .read()
+            .await
+            .get(arn)
+            .and_then(|target| target.client.clone());
+        if cli.is_some() {
+            return cli;
+        }
+
         self.inc_arn_errs(bucket, arn).await;
         None
     }
@@ -786,7 +790,10 @@ impl BucketTargetSys {
                         && tgt
                             .credentials
                             .as_ref()
-                            .map(|c| c.access_key == target.credentials.as_ref().unwrap_or(&Credentials::default()).access_key)
+                            .map(|c| {
+                                let default_creds = Credentials::default();
+                                c.access_key == target.credentials.as_ref().unwrap_or(&default_creds).access_key
+                            })
                             .unwrap_or(false)
                     {
                         return (tgt.arn.clone(), true);
@@ -894,6 +901,41 @@ pub struct RemoveObjectOptions {
     pub replication_validity_check: bool,
 }
 
+fn build_remove_object_headers(version_id: Option<&str>, opts: &RemoveObjectOptions) -> HeaderMap {
+    let mut headers = HeaderMap::new();
+    if opts.force_delete {
+        insert_header(&mut headers, SUFFIX_FORCE_DELETE, "true");
+    }
+    if opts.governance_bypass {
+        headers.insert(AMZ_OBJECT_LOCK_BYPASS_GOVERNANCE, "true".parse().unwrap());
+    }
+
+    if opts.replication_delete_marker {
+        insert_header(&mut headers, SUFFIX_SOURCE_DELETEMARKER, "true");
+    }
+
+    if let Some(t) = opts.replication_mtime {
+        insert_header(&mut headers, SUFFIX_SOURCE_MTIME, t.format(&Rfc3339).unwrap_or_default());
+    }
+
+    if !opts.replication_status.is_empty() {
+        headers.insert(AMZ_BUCKET_REPLICATION_STATUS, opts.replication_status.as_str().parse().unwrap());
+    }
+
+    if let Some(version_id) = version_id {
+        insert_header(&mut headers, SUFFIX_SOURCE_VERSION_ID, version_id);
+    }
+
+    if opts.replication_request {
+        insert_header(&mut headers, SUFFIX_SOURCE_REPLICATION_REQUEST, "true");
+    }
+    if opts.replication_validity_check {
+        insert_header(&mut headers, SUFFIX_SOURCE_REPLICATION_CHECK, "true");
+    }
+
+    headers
+}
+
 #[derive(Debug, Clone)]
 pub struct AdvancedPutOptions {
     pub source_version_id: String,
@@ -1424,39 +1466,15 @@ impl TargetClient {
         version_id: Option<String>,
         opts: RemoveObjectOptions,
     ) -> Result<(), S3ClientError> {
-        let mut headers = HeaderMap::new();
-        if opts.force_delete {
-            insert_header(&mut headers, SUFFIX_FORCE_DELETE, "true");
-        }
-        if opts.governance_bypass {
-            headers.insert(AMZ_OBJECT_LOCK_BYPASS_GOVERNANCE, "true".parse().unwrap());
-        }
-
-        if opts.replication_delete_marker {
-            insert_header(&mut headers, SUFFIX_SOURCE_DELETEMARKER, "true");
-        }
-
-        if let Some(t) = opts.replication_mtime {
-            insert_header(&mut headers, SUFFIX_SOURCE_MTIME, t.format(&Rfc3339).unwrap_or_default());
-        }
-
-        if !opts.replication_status.is_empty() {
-            headers.insert(AMZ_BUCKET_REPLICATION_STATUS, opts.replication_status.as_str().parse().unwrap());
-        }
-
-        if opts.replication_request {
-            insert_header(&mut headers, SUFFIX_SOURCE_REPLICATION_REQUEST, "true");
-        }
-        if opts.replication_validity_check {
-            insert_header(&mut headers, SUFFIX_SOURCE_REPLICATION_CHECK, "true");
-        }
+        let headers = build_remove_object_headers(version_id.as_deref(), &opts);
+        let api_version_id = if opts.replication_request { None } else { version_id };
 
         match self
             .client
             .delete_object()
             .bucket(bucket)
             .key(object)
-            .set_version_id(version_id)
+            .set_version_id(api_version_id)
             .customize()
             .map_request(move |mut req| {
                 for (k, v) in headers.clone().into_iter() {
@@ -1550,3 +1568,53 @@ impl From<std::io::Error> for BucketTargetError {
 }
 
 impl Error for BucketTargetError {}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn build_remove_object_headers_includes_internal_version_id_for_replication_delete() {
+        let version_id = Uuid::new_v4().to_string();
+        let headers = build_remove_object_headers(
+            Some(version_id.as_str()),
+            &RemoveObjectOptions {
+                force_delete: false,
+                governance_bypass: false,
+                replication_delete_marker: true,
+                replication_mtime: None,
+                replication_status: ReplicationStatusType::Replica,
+                replication_request: true,
+                replication_validity_check: false,
+            },
+        );
+
+        assert_eq!(
+            rustfs_utils::http::get_header(&headers, SUFFIX_SOURCE_VERSION_ID).as_deref(),
+            Some(version_id.as_str()),
+            "replication delete requests must preserve the version id in internal headers"
+        );
+    }
+
+    #[test]
+    fn build_remove_object_headers_omits_delete_marker_flag_for_marker_version_purge() {
+        let version_id = Uuid::new_v4().to_string();
+        let headers = build_remove_object_headers(
+            Some(version_id.as_str()),
+            &RemoveObjectOptions {
+                force_delete: false,
+                governance_bypass: false,
+                replication_delete_marker: false,
+                replication_mtime: None,
+                replication_status: ReplicationStatusType::Replica,
+                replication_request: true,
+                replication_validity_check: false,
+            },
+        );
+
+        assert!(
+            rustfs_utils::http::get_header(&headers, SUFFIX_SOURCE_DELETEMARKER).is_none(),
+            "delete-marker version purges must not masquerade as delete-marker creations"
+        );
+    }
+}

crates/ecstore/src/bucket/lifecycle/mod.rs

@@ -14,8 +14,9 @@
 
 pub mod bucket_lifecycle_audit;
 pub mod bucket_lifecycle_ops;
+pub mod core;
 pub mod evaluator;
-pub mod lifecycle;
+pub use self::core as lifecycle;
 pub mod rule;
 pub mod tier_last_day_stats;
 pub mod tier_sweeper;

crates/ecstore/src/bucket/lifecycle/tier_sweeper.rs

@@ -1,4 +1,3 @@
-#![allow(unused_imports)]
 // Copyright 2024 RustFS Team
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
@@ -12,6 +11,7 @@
 // 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.
+#![allow(unused_imports)]
 #![allow(unused_variables)]
 #![allow(unused_mut)]
 #![allow(unused_assignments)]
@@ -20,15 +20,144 @@
 
 use crate::bucket::lifecycle::bucket_lifecycle_ops::{ExpiryOp, GLOBAL_ExpiryState, TransitionedObject};
 use crate::bucket::lifecycle::lifecycle::{self, ObjectOpts};
+use crate::client::signer_error::error_chain_contains_signer_header_marker;
 use crate::global::GLOBAL_TierConfigMgr;
+use rustfs_utils::get_env_usize;
 use sha2::{Digest, Sha256};
 use std::any::Any;
+use std::collections::VecDeque;
 use std::io::Write;
+use std::sync::LazyLock;
+use std::sync::atomic::{AtomicUsize, Ordering};
+use std::time::{Duration, Instant};
+use tokio::sync::{Mutex, Semaphore};
+use tracing::warn;
 use uuid::Uuid;
 use xxhash_rust::xxh64;
 
 static XXHASH_SEED: u64 = 0;
 
+const ENV_REMOTE_DELETE_MAX_CONCURRENCY: &str = "RUSTFS_REMOTE_DELETE_MAX_CONCURRENCY";
+const ENV_REMOTE_DELETE_BREAKER_THRESHOLD: &str = "RUSTFS_REMOTE_DELETE_BREAKER_THRESHOLD";
+const ENV_REMOTE_DELETE_BREAKER_WINDOW_SECS: &str = "RUSTFS_REMOTE_DELETE_BREAKER_WINDOW_SECS";
+const DEFAULT_REMOTE_DELETE_BREAKER_THRESHOLD: usize = 50;
+const DEFAULT_REMOTE_DELETE_BREAKER_WINDOW_SECS: usize = 30;
+const METRIC_DELETE_REMOTE_FAILED_TOTAL: &str = "rustfs_delete_remote_failed_total";
+const METRIC_DELETE_REMOTE_BREAKER_TOTAL: &str = "rustfs_delete_remote_breaker_total";
+const METRIC_DELETE_REMOTE_INFLIGHT: &str = "rustfs_delete_remote_inflight";
+
+static REMOTE_DELETE_INFLIGHT: AtomicUsize = AtomicUsize::new(0);
+
+static REMOTE_DELETE_LIMITER: LazyLock<Semaphore> = LazyLock::new(|| {
+    let default_limit = std::cmp::min(num_cpus::get(), 16).max(1);
+    let concurrency = get_env_usize(ENV_REMOTE_DELETE_MAX_CONCURRENCY, default_limit).max(1);
+    Semaphore::new(concurrency)
+});
+
+static REMOTE_DELETE_BREAKER: LazyLock<Mutex<RemoteDeleteBreaker>> = LazyLock::new(|| {
+    Mutex::new(RemoteDeleteBreaker::new(
+        get_env_usize(ENV_REMOTE_DELETE_BREAKER_THRESHOLD, DEFAULT_REMOTE_DELETE_BREAKER_THRESHOLD).max(1),
+        Duration::from_secs(
+            get_env_usize(ENV_REMOTE_DELETE_BREAKER_WINDOW_SECS, DEFAULT_REMOTE_DELETE_BREAKER_WINDOW_SECS) as u64,
+        ),
+    ))
+});
+
+#[derive(Debug)]
+struct RemoteDeleteBreaker {
+    threshold: usize,
+    window: Duration,
+    failures: VecDeque<Instant>,
+}
+
+impl RemoteDeleteBreaker {
+    fn new(threshold: usize, window: Duration) -> Self {
+        Self {
+            threshold: threshold.max(1),
+            window: window.max(Duration::from_secs(1)),
+            failures: VecDeque::new(),
+        }
+    }
+
+    fn should_short_circuit(&mut self, now: Instant) -> bool {
+        self.prune(now);
+        self.failures.len() >= self.threshold
+    }
+
+    fn record_signer_failure(&mut self, now: Instant) -> bool {
+        self.prune(now);
+        let was_open = self.failures.len() >= self.threshold;
+        self.failures.push_back(now);
+        !was_open && self.failures.len() >= self.threshold
+    }
+
+    fn prune(&mut self, now: Instant) {
+        while let Some(ts) = self.failures.front().copied() {
+            if now.duration_since(ts) > self.window {
+                self.failures.pop_front();
+            } else {
+                break;
+            }
+        }
+    }
+}
+
+struct RemoteDeleteInflightGuard;
+
+impl RemoteDeleteInflightGuard {
+    fn new() -> Self {
+        let inflight = REMOTE_DELETE_INFLIGHT.fetch_add(1, Ordering::Relaxed) + 1;
+        metrics::gauge!(METRIC_DELETE_REMOTE_INFLIGHT).set(inflight as f64);
+        Self
+    }
+}
+
+impl Drop for RemoteDeleteInflightGuard {
+    fn drop(&mut self) {
+        let inflight = REMOTE_DELETE_INFLIGHT.fetch_sub(1, Ordering::Relaxed) - 1;
+        metrics::gauge!(METRIC_DELETE_REMOTE_INFLIGHT).set(inflight as f64);
+    }
+}
+
+fn is_signer_header_error(err: &std::io::Error) -> bool {
+    if err.kind() != std::io::ErrorKind::InvalidInput {
+        return false;
+    }
+
+    if let Some(source) = err.get_ref() {
+        if error_chain_contains_signer_header_marker(source) {
+            return true;
+        }
+    }
+
+    let message = err.to_string().to_ascii_lowercase();
+    message.contains("invalid utf-8 header value")
+        || message.contains("invalidheadervalue")
+        || (message.contains("sign v4") && message.contains("header value"))
+}
+
+async fn remote_delete_breaker_is_open(now: Instant) -> bool {
+    let mut breaker = REMOTE_DELETE_BREAKER.lock().await;
+    breaker.should_short_circuit(now)
+}
+
+async fn record_remote_delete_failure(err: &std::io::Error, now: Instant) {
+    metrics::counter!(METRIC_DELETE_REMOTE_FAILED_TOTAL).increment(1);
+
+    if !is_signer_header_error(err) {
+        return;
+    }
+
+    let mut breaker = REMOTE_DELETE_BREAKER.lock().await;
+    if breaker.record_signer_failure(now) {
+        warn!(
+            threshold = breaker.threshold,
+            window_secs = breaker.window.as_secs(),
+            "remote tier delete breaker opened by signer/header failures"
+        );
+    }
+}
+
 #[derive(Default)]
 #[allow(dead_code)]
 struct ObjSweeper {
@@ -109,10 +238,19 @@ impl ObjSweeper {
     }
 
     pub async fn sweep(&self) {
-        let je = self.should_remove_remote_object();
-        if !je.is_none() {
-            let mut expiry_state = GLOBAL_ExpiryState.write().await;
-            expiry_state.enqueue_tier_journal_entry(&je.expect("err!"));
+        let Some(je) = self.should_remove_remote_object() else {
+            return;
+        };
+        let hash = je.op_hash();
+        // Grab the sender under a short read lock, then release the lock so we
+        // don't hold it across the async send.
+        let wrkr = GLOBAL_ExpiryState.read().await.get_worker_ch(hash);
+        let Some(wrkr) = wrkr else {
+            GLOBAL_ExpiryState.write().await.increment_missed_tier_journal_tasks();
+            return;
+        };
+        if wrkr.send(Some(Box::new(je))).await.is_err() {
+            GLOBAL_ExpiryState.write().await.increment_missed_tier_journal_tasks();
         }
     }
 }
@@ -139,12 +277,31 @@ impl ExpiryOp for Jentry {
 }
 
 pub async fn delete_object_from_remote_tier(obj_name: &str, rv_id: &str, tier_name: &str) -> Result<(), std::io::Error> {
+    if remote_delete_breaker_is_open(Instant::now()).await {
+        metrics::counter!(METRIC_DELETE_REMOTE_BREAKER_TOTAL).increment(1);
+        return Err(std::io::Error::other("remote tier delete breaker is open due to signer/header failures"));
+    }
+
+    let _permit = REMOTE_DELETE_LIMITER
+        .acquire()
+        .await
+        .map_err(|_| std::io::Error::other("remote tier delete limiter is closed"))?;
+    let _inflight = RemoteDeleteInflightGuard::new();
+
     let mut config_mgr = GLOBAL_TierConfigMgr.write().await;
     let w = match config_mgr.get_driver(tier_name).await {
         Ok(w) => w,
-        Err(e) => return Err(std::io::Error::other(e)),
+        Err(e) => {
+            let err = std::io::Error::other(e);
+            record_remote_delete_failure(&err, Instant::now()).await;
+            return Err(err);
+        }
     };
-    w.remove(obj_name, rv_id).await
+    let result = w.remove(obj_name, rv_id).await;
+    if let Err(err) = &result {
+        record_remote_delete_failure(err, Instant::now()).await;
+    }
+    result
 }
 
 pub fn transitioned_delete_journal_entry(
@@ -180,4 +337,44 @@ pub fn transitioned_force_delete_journal_entry(transitioned: &TransitionedObject
 }
 
 #[cfg(test)]
-mod test {}
+mod test {
+    use crate::client::signer_error::invalid_utf8_header_error;
+
+    use super::{RemoteDeleteBreaker, is_signer_header_error};
+    use std::io::{Error, ErrorKind};
+    use std::time::{Duration, Instant};
+
+    #[test]
+    fn signer_header_error_detection_matches_utf8_failures() {
+        let err = Error::new(
+            ErrorKind::InvalidInput,
+            "failed to sign v4 request: invalid UTF-8 header value for `x-amz-meta-invalid`",
+        );
+        assert!(is_signer_header_error(&err));
+    }
+
+    #[test]
+    fn signer_header_error_detection_rejects_unrelated_errors() {
+        let err = Error::other("dial tcp: i/o timeout");
+        assert!(!is_signer_header_error(&err));
+    }
+
+    #[test]
+    fn signer_header_error_detection_matches_structured_marker() {
+        let err = invalid_utf8_header_error("failed to sign v4 request", "x-amz-meta-invalid");
+        assert!(is_signer_header_error(&err));
+    }
+
+    #[test]
+    fn breaker_opens_at_threshold_and_recovers_after_window() {
+        let mut breaker = RemoteDeleteBreaker::new(3, Duration::from_secs(30));
+        let start = Instant::now();
+
+        assert!(!breaker.should_short_circuit(start));
+        assert!(!breaker.record_signer_failure(start));
+        assert!(!breaker.record_signer_failure(start + Duration::from_secs(1)));
+        assert!(breaker.record_signer_failure(start + Duration::from_secs(2)));
+        assert!(breaker.should_short_circuit(start + Duration::from_secs(3)));
+        assert!(!breaker.should_short_circuit(start + Duration::from_secs(40)));
+    }
+}

crates/ecstore/src/bucket/replication/config.rs

@@ -152,7 +152,16 @@ impl ReplicationConfigurationExt for ReplicationConfiguration {
             }
 
             if obj.op_type == ReplicationType::Delete {
+                if !rule.metadata_replicate(obj) {
+                    return false;
+                }
+
                 if obj.version_id.is_some() {
+                    if obj.delete_marker {
+                        return rule.delete_marker_replication.clone().is_some_and(|d| {
+                            d.status == Some(DeleteMarkerReplicationStatus::from_static(DeleteMarkerReplicationStatus::ENABLED))
+                        });
+                    }
                     return rule
                         .delete_replication
                         .clone()
@@ -215,19 +224,98 @@ impl ReplicationConfigurationExt for ReplicationConfiguration {
                 continue;
             }
 
-            if !self.role.is_empty() {
-                arns.push(self.role.clone()); // Use the legacy RoleArn when present
-                return arns;
-            }
-
-            if !targets_map.contains(&rule.destination.bucket) {
+            if !rule.destination.bucket.is_empty() && !targets_map.contains(&rule.destination.bucket) {
                 targets_map.insert(rule.destination.bucket.clone());
             }
         }
 
+        if targets_map.is_empty() && !self.role.is_empty() {
+            arns.push(self.role.clone());
+            return arns;
+        }
+
         for arn in targets_map {
             arns.push(arn);
         }
         arns
     }
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use s3s::dto::{DeleteMarkerReplication, Destination, ExistingObjectReplication, ReplicationRule};
+
+    fn replication_rule(id: &str, arn: &str) -> ReplicationRule {
+        ReplicationRule {
+            delete_marker_replication: Some(DeleteMarkerReplication::default()),
+            delete_replication: None,
+            destination: Destination {
+                bucket: arn.to_string(),
+                ..Default::default()
+            },
+            existing_object_replication: Some(ExistingObjectReplication {
+                status: ExistingObjectReplicationStatus::from_static(ExistingObjectReplicationStatus::ENABLED),
+            }),
+            filter: None,
+            id: Some(id.to_string()),
+            prefix: Some(String::new()),
+            priority: Some(1),
+            source_selection_criteria: None,
+            status: ReplicationRuleStatus::from_static(ReplicationRuleStatus::ENABLED),
+        }
+    }
+
+    #[test]
+    fn filter_target_arns_keeps_multiple_destinations_when_role_is_present() {
+        let config = ReplicationConfiguration {
+            role: "arn:legacy:target".to_string(),
+            rules: vec![
+                replication_rule("rule-1", "arn:target:a"),
+                replication_rule("rule-2", "arn:target:b"),
+            ],
+        };
+
+        let arns = config.filter_target_arns(&ObjectOpts {
+            name: "object".to_string(),
+            op_type: ReplicationType::Object,
+            ..Default::default()
+        });
+
+        assert_eq!(arns.len(), 2);
+        assert!(arns.iter().any(|arn| arn == "arn:target:a"));
+        assert!(arns.iter().any(|arn| arn == "arn:target:b"));
+    }
+
+    #[test]
+    fn filter_target_arns_falls_back_to_role_when_destination_is_empty() {
+        let config = ReplicationConfiguration {
+            role: "arn:legacy:target".to_string(),
+            rules: vec![ReplicationRule {
+                delete_marker_replication: Some(DeleteMarkerReplication::default()),
+                delete_replication: None,
+                destination: Destination {
+                    bucket: String::new(),
+                    ..Default::default()
+                },
+                existing_object_replication: Some(ExistingObjectReplication {
+                    status: ExistingObjectReplicationStatus::from_static(ExistingObjectReplicationStatus::ENABLED),
+                }),
+                filter: None,
+                id: Some("rule-1".to_string()),
+                prefix: Some(String::new()),
+                priority: Some(1),
+                source_selection_criteria: None,
+                status: ReplicationRuleStatus::from_static(ReplicationRuleStatus::ENABLED),
+            }],
+        };
+
+        let arns = config.filter_target_arns(&ObjectOpts {
+            name: "object".to_string(),
+            op_type: ReplicationType::Object,
+            ..Default::default()
+        });
+
+        assert_eq!(arns, vec!["arn:legacy:target".to_string()]);
+    }
+}

crates/ecstore/src/bucket/replication/replication_pool.rs

@@ -824,10 +824,7 @@ impl<S: StorageAPI> ReplicationPool<S> {
         let cancel_token = CancellationToken::new();
         resyncer.register_cancel_token(&opts, cancel_token.clone()).await;
         tokio::spawn(async move {
-            resyncer
-                .clone()
-                .resync_bucket(cancel_token, storage, false, opts.clone())
-                .await;
+            Box::pin(resyncer.clone().resync_bucket(cancel_token, storage, false, opts.clone())).await;
             resyncer.clear_cancel_token(&opts).await;
         });
 
@@ -914,7 +911,7 @@ impl<S: StorageAPI> ReplicationPool<S> {
                         };
                         tokio::spawn(async move {
                             resync.register_cancel_token(&opts, ctx.clone()).await;
-                            resync.clone().resync_bucket(ctx, storage, true, opts.clone()).await;
+                            Box::pin(resync.clone().resync_bucket(ctx, storage, true, opts.clone())).await;
                             resync.clear_cancel_token(&opts).await;
                         });
                     }
@@ -959,6 +956,9 @@ pub type DynReplicationPool = dyn ReplicationPoolTrait + Send + Sync;
 /// Trait that abstracts the replication pool operations
 #[async_trait::async_trait]
 pub trait ReplicationPoolTrait: std::fmt::Debug {
+    fn active_workers(&self) -> i32;
+    fn active_mrf_workers(&self) -> i32;
+    fn active_lrg_workers(&self) -> i32;
     async fn queue_replica_task(&self, ri: ReplicateObjectInfo);
     async fn queue_replica_delete_task(&self, ri: DeletedObjectReplicationInfo);
     async fn resize(&self, priority: ReplicationPriority, max_workers: usize, max_l_workers: usize);
@@ -975,6 +975,18 @@ pub trait ReplicationPoolTrait: std::fmt::Debug {
 // Implement the trait for ReplicationPool
 #[async_trait::async_trait]
 impl<S: StorageAPI> ReplicationPoolTrait for ReplicationPool<S> {
+    fn active_workers(&self) -> i32 {
+        ReplicationPool::<S>::active_workers(self)
+    }
+
+    fn active_mrf_workers(&self) -> i32 {
+        ReplicationPool::<S>::active_mrf_workers(self)
+    }
+
+    fn active_lrg_workers(&self) -> i32 {
+        ReplicationPool::<S>::active_lrg_workers(self)
+    }
+
     async fn queue_replica_task(&self, ri: ReplicateObjectInfo) {
         self.queue_replica_task(ri).await;
     }

crates/ecstore/src/bucket/replication/replication_resyncer.rs

@@ -34,8 +34,8 @@ use crate::global::get_global_bucket_monitor;
 use crate::set_disk::get_lock_acquire_timeout;
 use crate::store_api::{DeletedObject, HTTPRangeSpec, ObjectInfo, ObjectOptions, ObjectToDelete, WalkOptions};
 use crate::{StorageAPI, new_object_layer_fn};
-use aws_sdk_s3::error::SdkError;
-use aws_sdk_s3::operation::head_object::HeadObjectOutput;
+use aws_sdk_s3::error::{ProvideErrorMetadata, SdkError};
+use aws_sdk_s3::operation::head_object::{HeadObjectError, HeadObjectOutput};
 use aws_sdk_s3::primitives::ByteStream;
 use aws_sdk_s3::types::{CompletedPart, ObjectLockLegalHoldStatus};
 use aws_smithy_types::body::SdkBody;
@@ -85,7 +85,8 @@ use tokio::task::JoinSet;
 use tokio::time::Duration as TokioDuration;
 use tokio_util::io::ReaderStream;
 use tokio_util::sync::CancellationToken;
-use tracing::{error, info, instrument, warn};
+use tracing::{debug, error, info, instrument, warn};
+use uuid::Uuid;
 
 pub(crate) const REPLICATION_DIR: &str = ".replication";
 pub(crate) const RESYNC_FILE_NAME: &str = "resync.bin";
@@ -114,6 +115,41 @@ fn resync_state_accepts_update(state: &TargetReplicationResyncStatus, opts: &Res
     state.resync_id.is_empty() || opts.resync_id.is_empty() || state.resync_id == opts.resync_id
 }
 
+fn should_count_head_proxy_failure(is_not_found: bool, code: Option<&str>, raw_status: Option<u16>) -> bool {
+    if is_not_found || matches!(code, Some("MethodNotAllowed" | "405")) {
+        return false;
+    }
+    !matches!(raw_status, Some(404 | 405))
+}
+
+fn is_head_proxy_failure(err: &SdkError<HeadObjectError>) -> bool {
+    let (is_not_found, code) = err
+        .as_service_error()
+        .map(|service_err| (service_err.is_not_found(), service_err.code()))
+        .unwrap_or((false, None));
+    let raw_status = err.raw_response().map(|resp| resp.status().as_u16());
+    should_count_head_proxy_failure(is_not_found, code, raw_status)
+}
+
+async fn record_proxy_request(bucket: &str, api: &str, is_err: bool) {
+    if let Some(stats) = GLOBAL_REPLICATION_STATS.get() {
+        stats.inc_proxy(bucket, api, is_err).await;
+    }
+}
+
+async fn head_object_with_proxy_stats(
+    source_bucket: &str,
+    target_client: &TargetClient,
+    target_bucket: &str,
+    object: &str,
+    version_id: Option<String>,
+) -> std::result::Result<HeadObjectOutput, SdkError<HeadObjectError>> {
+    let result = target_client.head_object(target_bucket, object, version_id).await;
+    let is_err = result.as_ref().err().is_some_and(is_head_proxy_failure);
+    record_proxy_request(source_bucket, "HeadObject", is_err).await;
+    result
+}
+
 #[derive(Debug, Clone, Default)]
 pub struct ResyncOpts {
     pub bucket: String,
@@ -747,10 +783,15 @@ impl ReplicationResyncer {
 
                     let reset_id = target_client.reset_id.clone();
 
-                    let (size, err) = if let Err(err) = target_client
-                        .head_object(&target_client.bucket, &roi.name, roi.version_id.map(|v| v.to_string()))
-                        .await
-                    {
+                    let head_result = head_object_with_proxy_stats(
+                        &bucket_name,
+                        target_client.as_ref(),
+                        &target_client.bucket,
+                        &roi.name,
+                        roi.version_id.map(|v| v.to_string()),
+                    )
+                    .await;
+                    let (size, err) = if let Err(err) = head_result {
                         if roi.delete_marker {
                             st.replicated_count += 1;
                         } else {
@@ -1271,15 +1312,7 @@ pub async fn check_replicate_delete(
         return ReplicateDecision::default();
     }
 
-    let opts = ObjectOpts {
-        name: dobj.object_name.clone(),
-        ssec: is_ssec_encrypted(&oi.user_defined),
-        user_tags: oi.user_tags.clone(),
-        delete_marker: oi.delete_marker,
-        version_id: dobj.version_id,
-        op_type: ReplicationType::Delete,
-        ..Default::default()
-    };
+    let opts = delete_replication_object_opts(dobj, oi);
 
     let tgt_arns = rcfg.filter_target_arns(&opts);
     let mut dsc = ReplicateDecision::new();
@@ -1331,6 +1364,19 @@ pub async fn check_replicate_delete(
     dsc
 }
 
+fn delete_replication_object_opts(dobj: &ObjectToDelete, oi: &ObjectInfo) -> ObjectOpts {
+    ObjectOpts {
+        name: dobj.object_name.clone(),
+        ssec: is_ssec_encrypted(&oi.user_defined),
+        user_tags: oi.user_tags.clone(),
+        delete_marker: oi.delete_marker,
+        version_id: dobj.version_id,
+        op_type: ReplicationType::Delete,
+        replica: oi.replication_status == ReplicationStatusType::Replica,
+        ..Default::default()
+    }
+}
+
 /// Check if the user-defined metadata contains SSEC encryption headers
 fn is_ssec_encrypted(user_defined: &HashMap<String, String>) -> bool {
     user_defined.contains_key(SSEC_ALGORITHM_HEADER)
@@ -1497,6 +1543,54 @@ pub async fn replicate_delete<S: StorageAPI>(dobj: DeletedObjectReplicationInfo,
         }
     };
 
+    if dobj.delete_object.delete_marker
+        && let Some(delete_marker_version_id) = dobj.delete_object.delete_marker_version_id
+    {
+        let source_marker_state = storage
+            .get_object_info(
+                &bucket,
+                &dobj.delete_object.object_name,
+                &ObjectOptions {
+                    version_id: Some(delete_marker_version_id.to_string()),
+                    versioned: BucketVersioningSys::prefix_enabled(&bucket, &dobj.delete_object.object_name).await,
+                    version_suspended: BucketVersioningSys::prefix_suspended(&bucket, &dobj.delete_object.object_name).await,
+                    ..Default::default()
+                },
+            )
+            .await;
+
+        match source_marker_state {
+            Ok(info) if info.delete_marker && info.version_id == Some(delete_marker_version_id) => {}
+            Ok(_) => {
+                warn!(
+                    bucket,
+                    object = dobj.delete_object.object_name,
+                    version_id = %delete_marker_version_id,
+                    "skipping stale delete-marker replication because source version is no longer a delete marker"
+                );
+                return;
+            }
+            Err(err) if is_err_object_not_found(&err) || is_err_version_not_found(&err) => {
+                warn!(
+                    bucket,
+                    object = dobj.delete_object.object_name,
+                    version_id = %delete_marker_version_id,
+                    "skipping stale delete-marker replication because source version no longer exists"
+                );
+                return;
+            }
+            Err(err) => {
+                warn!(
+                    bucket,
+                    object = dobj.delete_object.object_name,
+                    version_id = %delete_marker_version_id,
+                    error = %err,
+                    "failed to verify source delete-marker state before replication"
+                );
+            }
+        }
+    }
+
     let dsc = match parse_replicate_decision(
         &bucket,
         &dobj
@@ -1529,7 +1623,6 @@ pub async fn replicate_delete<S: StorageAPI>(dobj: DeletedObjectReplicationInfo,
             return;
         }
     };
-
     let ns_lock = match storage
         .new_ns_lock(&bucket, format!("/[replicate]/{}", dobj.delete_object.object_name).as_str())
         .await
@@ -1653,7 +1746,33 @@ pub async fn replicate_delete<S: StorageAPI>(dobj: DeletedObjectReplicationInfo,
         }
     }
 
-    let (replication_status, prev_status) = if dobj.delete_object.version_id.is_none() {
+    let is_version_purge = is_version_delete_replication(&dobj.delete_object);
+
+    if should_retry_delete_marker_purge(&dobj.delete_object) {
+        let bucket_clone = bucket.clone();
+        let dobj_clone = dobj.clone();
+        let dsc_clone = dsc.clone();
+        let storage_clone = storage.clone();
+        tokio::spawn(async move {
+            for _ in 0..5 {
+                if let Some(delete_marker_version_id) = dobj_clone.delete_object.delete_marker_version_id
+                    && source_delete_marker_missing(
+                        &*storage_clone,
+                        &bucket_clone,
+                        &dobj_clone.delete_object.object_name,
+                        delete_marker_version_id,
+                    )
+                    .await
+                {
+                    replicate_delete_marker_purge_to_targets(&bucket_clone, &dobj_clone, &dsc_clone).await;
+                    break;
+                }
+                tokio::time::sleep(TokioDuration::from_secs(1)).await;
+            }
+        });
+    }
+
+    let (replication_status, prev_status) = if !is_version_purge {
         (
             rinfos.replication_status(),
             dobj.delete_object
@@ -1741,6 +1860,65 @@ pub async fn replicate_delete<S: StorageAPI>(dobj: DeletedObjectReplicationInfo,
     }
 }
 
+async fn source_delete_marker_missing<S: StorageAPI>(
+    storage: &S,
+    bucket: &str,
+    object_name: &str,
+    delete_marker_version_id: Uuid,
+) -> bool {
+    match storage
+        .get_object_info(
+            bucket,
+            object_name,
+            &ObjectOptions {
+                version_id: Some(delete_marker_version_id.to_string()),
+                versioned: BucketVersioningSys::prefix_enabled(bucket, object_name).await,
+                version_suspended: BucketVersioningSys::prefix_suspended(bucket, object_name).await,
+                ..Default::default()
+            },
+        )
+        .await
+    {
+        Ok(info) => !info.delete_marker || info.version_id != Some(delete_marker_version_id),
+        Err(err) => is_err_object_not_found(&err) || is_err_version_not_found(&err),
+    }
+}
+
+async fn replicate_delete_marker_purge_to_targets(bucket: &str, dobj: &DeletedObjectReplicationInfo, dsc: &ReplicateDecision) {
+    let Some(delete_marker_version_id) = dobj.delete_object.delete_marker_version_id else {
+        return;
+    };
+
+    for tgt_entry in dsc.targets_map.values() {
+        if !tgt_entry.replicate {
+            continue;
+        }
+        if !dobj.target_arn.is_empty() && dobj.target_arn != tgt_entry.arn {
+            continue;
+        }
+        let Some(tgt_client) = BucketTargetSys::get().get_remote_target_client(bucket, &tgt_entry.arn).await else {
+            continue;
+        };
+
+        let _ = tgt_client
+            .remove_object(
+                &tgt_client.bucket,
+                &dobj.delete_object.object_name,
+                Some(delete_marker_version_id.to_string()),
+                RemoveObjectOptions {
+                    force_delete: false,
+                    governance_bypass: false,
+                    replication_delete_marker: false,
+                    replication_mtime: dobj.delete_object.delete_marker_mtime,
+                    replication_status: ReplicationStatusType::Replica,
+                    replication_request: true,
+                    replication_validity_check: false,
+                },
+            )
+            .await;
+    }
+}
+
 async fn replicate_force_delete_to_targets<S: StorageAPI>(dobj: &DeletedObjectReplicationInfo, storage: Arc<S>) {
     let bucket = &dobj.bucket;
     let object_name = &dobj.delete_object.object_name;
@@ -1924,6 +2102,18 @@ async fn replicate_force_delete_to_targets<S: StorageAPI>(dobj: &DeletedObjectRe
     }
 }
 
+fn is_version_delete_replication(dobj: &DeletedObject) -> bool {
+    dobj.version_id.is_some() || (dobj.delete_marker_version_id.is_some() && !dobj.delete_marker)
+}
+
+fn should_retry_delete_marker_purge(dobj: &DeletedObject) -> bool {
+    dobj.delete_marker_version_id.is_some()
+}
+
+fn is_retryable_delete_replication_head_error(is_not_found: bool, code: Option<&str>) -> bool {
+    !is_not_found && !matches!(code, Some("MethodNotAllowed" | "405"))
+}
+
 async fn replicate_delete_to_target(dobj: &DeletedObjectReplicationInfo, tgt_client: Arc<TargetClient>) -> ReplicatedTargetInfo {
     let version_id = if let Some(version_id) = &dobj.delete_object.delete_marker_version_id {
         version_id.to_owned()
@@ -1941,7 +2131,8 @@ async fn replicate_delete_to_target(dobj: &DeletedObjectReplicationInfo, tgt_cli
     rinfo.endpoint = tgt_client.endpoint.clone();
     rinfo.secure = tgt_client.secure;
 
-    if dobj.delete_object.version_id.is_none()
+    let is_version_purge = is_version_delete_replication(&dobj.delete_object);
+    if !is_version_purge
         && rinfo.prev_replication_status == ReplicationStatusType::Completed
         && dobj.op_type != ReplicationType::ExistingObject
     {
@@ -1949,12 +2140,12 @@ async fn replicate_delete_to_target(dobj: &DeletedObjectReplicationInfo, tgt_cli
         return rinfo;
     }
 
-    if dobj.delete_object.version_id.is_some() && rinfo.version_purge_status == VersionPurgeStatusType::Complete {
+    if is_version_purge && rinfo.version_purge_status == VersionPurgeStatusType::Complete {
         return rinfo;
     }
 
     if BucketTargetSys::get().is_offline(&tgt_client.to_url()).await {
-        if dobj.delete_object.version_id.is_none() {
+        if !is_version_purge {
             rinfo.replication_status = ReplicationStatusType::Failed;
         } else {
             rinfo.version_purge_status = VersionPurgeStatusType::Failed;
@@ -1968,18 +2159,34 @@ async fn replicate_delete_to_target(dobj: &DeletedObjectReplicationInfo, tgt_cli
         Some(version_id.to_string())
     };
 
-    if dobj.delete_object.delete_marker_version_id.is_some()
-        && let Err(e) = tgt_client
-            .head_object(&tgt_client.bucket, &dobj.delete_object.object_name, version_id.clone())
-            .await
-        && let SdkError::ServiceError(service_err) = &e
-        && !service_err.err().is_not_found()
-    {
-        rinfo.replication_status = ReplicationStatusType::Failed;
-        rinfo.error = Some(e.to_string());
-
-        return rinfo;
-    };
+    if dobj.delete_object.delete_marker && dobj.delete_object.delete_marker_version_id.is_some() {
+        match head_object_with_proxy_stats(
+            &dobj.bucket,
+            tgt_client.as_ref(),
+            &tgt_client.bucket,
+            &dobj.delete_object.object_name,
+            version_id.clone(),
+        )
+        .await
+        {
+            Ok(_) => {}
+            Err(e) => {
+                let non_retryable = matches!(
+                    &e,
+                    SdkError::ServiceError(service_err)
+                        if is_retryable_delete_replication_head_error(
+                            service_err.err().is_not_found(),
+                            service_err.err().code(),
+                        )
+                );
+                if non_retryable {
+                    rinfo.replication_status = ReplicationStatusType::Failed;
+                    rinfo.error = Some(e.to_string());
+                    return rinfo;
+                }
+            }
+        }
+    }
 
     match tgt_client
         .remove_object(
@@ -1989,7 +2196,7 @@ async fn replicate_delete_to_target(dobj: &DeletedObjectReplicationInfo, tgt_cli
             RemoveObjectOptions {
                 force_delete: false,
                 governance_bypass: false,
-                replication_delete_marker: dobj.delete_object.delete_marker_version_id.is_some(),
+                replication_delete_marker: dobj.delete_object.delete_marker,
                 replication_mtime: dobj.delete_object.delete_marker_mtime,
                 replication_status: ReplicationStatusType::Replica,
                 replication_request: true,
@@ -1999,15 +2206,32 @@ async fn replicate_delete_to_target(dobj: &DeletedObjectReplicationInfo, tgt_cli
         .await
     {
         Ok(_) => {
-            if dobj.delete_object.version_id.is_none() {
+            debug!(
+                bucket = tgt_client.bucket,
+                object = dobj.delete_object.object_name,
+                version_id = ?version_id,
+                delete_marker = dobj.delete_object.delete_marker,
+                is_version_purge,
+                "replicate_delete_to_target succeeded"
+            );
+            if !is_version_purge {
                 rinfo.replication_status = ReplicationStatusType::Completed;
             } else {
                 rinfo.version_purge_status = VersionPurgeStatusType::Complete;
             }
         }
         Err(e) => {
+            warn!(
+                bucket = tgt_client.bucket,
+                object = dobj.delete_object.object_name,
+                version_id = ?version_id,
+                delete_marker = dobj.delete_object.delete_marker,
+                is_version_purge,
+                error = %e,
+                "replicate_delete_to_target failed"
+            );
             rinfo.error = Some(e.to_string());
-            if dobj.delete_object.version_id.is_none() {
+            if !is_version_purge {
                 rinfo.replication_status = ReplicationStatusType::Failed;
             } else {
                 rinfo.version_purge_status = VersionPurgeStatusType::Failed;
@@ -2292,9 +2516,14 @@ impl ReplicateObjectInfoExt for ReplicateObjectInfo {
         }
 
         let mut replication_action = replication_action;
-        match tgt_client
-            .head_object(&tgt_client.bucket, &object, self.version_id.map(|v| v.to_string()))
-            .await
+        match head_object_with_proxy_stats(
+            &bucket,
+            tgt_client.as_ref(),
+            &tgt_client.bucket,
+            &object,
+            self.version_id.map(|v| v.to_string()),
+        )
+        .await
         {
             Ok(oi) => {
                 replication_action = get_replication_action(&object_info, &oi, self.op_type);
@@ -2349,9 +2578,10 @@ impl ReplicateObjectInfoExt for ReplicateObjectInfo {
             }
         };
 
+        let has_tagging_replication = !put_opts.user_tags.is_empty();
         if let Some(err) = if is_multipart {
             drop(gr);
-            replicate_object_with_multipart(MultipartReplicationContext {
+            let result = replicate_object_with_multipart(MultipartReplicationContext {
                 storage: storage.clone(),
                 cli: tgt_client.clone(),
                 src_bucket: &bucket,
@@ -2362,16 +2592,24 @@ impl ReplicateObjectInfoExt for ReplicateObjectInfo {
                 arn: &rinfo.arn,
                 put_opts,
             })
-            .await
-            .err()
+            .await;
+            record_proxy_request(&bucket, "PutObject", result.is_err()).await;
+            if has_tagging_replication {
+                record_proxy_request(&bucket, "PutObjectTagging", result.is_err()).await;
+            }
+            result.err()
         } else {
             gr.stream = wrap_with_bandwidth_monitor(gr.stream, &put_opts, &bucket, &rinfo.arn);
             let byte_stream = async_read_to_bytestream(gr.stream);
-            tgt_client
+            let result = tgt_client
                 .put_object(&tgt_client.bucket, &object, size, byte_stream, &put_opts)
                 .await
-                .map_err(|e| std::io::Error::other(e.to_string()))
-                .err()
+                .map_err(|e| std::io::Error::other(e.to_string()));
+            record_proxy_request(&bucket, "PutObject", result.is_err()).await;
+            if has_tagging_replication {
+                record_proxy_request(&bucket, "PutObjectTagging", result.is_err()).await;
+            }
+            result.err()
         } {
             rinfo.replication_status = ReplicationStatusType::Failed;
             rinfo.error = Some(err.to_string());
@@ -2511,9 +2749,14 @@ impl ReplicateObjectInfoExt for ReplicateObjectInfo {
             warn!("failed to set replication tagging directive header: {err}");
         }
 
-        match tgt_client
-            .head_object(&tgt_client.bucket, &object, self.version_id.map(|v| v.to_string()))
-            .await
+        match head_object_with_proxy_stats(
+            &bucket,
+            tgt_client.as_ref(),
+            &tgt_client.bucket,
+            &object,
+            self.version_id.map(|v| v.to_string()),
+        )
+        .await
         {
             Ok(oi) => {
                 replication_action = get_replication_action(&object_info, &oi, self.op_type);
@@ -2631,9 +2874,10 @@ impl ReplicateObjectInfoExt for ReplicateObjectInfo {
                 }
             };
 
+            let has_tagging_replication = !put_opts.user_tags.is_empty();
             if let Some(err) = if is_multipart {
                 drop(gr);
-                replicate_object_with_multipart(MultipartReplicationContext {
+                let result = replicate_object_with_multipart(MultipartReplicationContext {
                     storage: storage.clone(),
                     cli: tgt_client.clone(),
                     src_bucket: &bucket,
@@ -2644,16 +2888,24 @@ impl ReplicateObjectInfoExt for ReplicateObjectInfo {
                     arn: &rinfo.arn,
                     put_opts,
                 })
-                .await
-                .err()
+                .await;
+                record_proxy_request(&bucket, "PutObject", result.is_err()).await;
+                if has_tagging_replication {
+                    record_proxy_request(&bucket, "PutObjectTagging", result.is_err()).await;
+                }
+                result.err()
             } else {
                 gr.stream = wrap_with_bandwidth_monitor(gr.stream, &put_opts, &bucket, &rinfo.arn);
                 let byte_stream = async_read_to_bytestream(gr.stream);
-                tgt_client
+                let result = tgt_client
                     .put_object(&tgt_client.bucket, &object, size, byte_stream, &put_opts)
                     .await
-                    .map_err(|e| std::io::Error::other(e.to_string()))
-                    .err()
+                    .map_err(|e| std::io::Error::other(e.to_string()));
+                record_proxy_request(&bucket, "PutObject", result.is_err()).await;
+                if has_tagging_replication {
+                    record_proxy_request(&bucket, "PutObjectTagging", result.is_err()).await;
+                }
+                result.err()
             } {
                 rinfo.replication_status = ReplicationStatusType::Failed;
                 rinfo.error = Some(err.to_string());
@@ -3450,6 +3702,153 @@ mod tests {
         );
     }
 
+    #[test]
+    fn test_delete_replication_object_opts_marks_replica_deletes() {
+        let dobj = ObjectToDelete {
+            object_name: "obj".to_string(),
+            version_id: Some(Uuid::new_v4()),
+            ..Default::default()
+        };
+        let oi = ObjectInfo {
+            bucket: "b".to_string(),
+            name: "obj".to_string(),
+            replication_status: ReplicationStatusType::Replica,
+            ..Default::default()
+        };
+
+        let opts = delete_replication_object_opts(&dobj, &oi);
+
+        assert!(
+            opts.replica,
+            "replica deletes must preserve replica status for downstream ReplicaModifications rules"
+        );
+        assert_eq!(opts.version_id, dobj.version_id);
+        assert_eq!(opts.name, dobj.object_name);
+        assert_eq!(opts.op_type, ReplicationType::Delete);
+    }
+
+    #[test]
+    fn test_delete_replication_object_opts_keeps_non_replica_deletes_local() {
+        let dobj = ObjectToDelete {
+            object_name: "obj".to_string(),
+            ..Default::default()
+        };
+        let oi = ObjectInfo {
+            bucket: "b".to_string(),
+            name: "obj".to_string(),
+            replication_status: ReplicationStatusType::Completed,
+            ..Default::default()
+        };
+
+        let opts = delete_replication_object_opts(&dobj, &oi);
+
+        assert!(!opts.replica, "source-originated deletes should not be treated as replica modifications");
+    }
+
+    #[test]
+    fn test_is_version_delete_replication_for_delete_marker_version_purge() {
+        let dobj = DeletedObject {
+            delete_marker: false,
+            delete_marker_version_id: Some(Uuid::new_v4()),
+            ..Default::default()
+        };
+
+        assert!(
+            is_version_delete_replication(&dobj),
+            "delete-marker version purges must be tracked as version purge replication, not delete-marker creation replication"
+        );
+    }
+
+    #[test]
+    fn test_is_version_delete_replication_for_delete_marker_creation() {
+        let dobj = DeletedObject {
+            delete_marker: true,
+            delete_marker_version_id: Some(Uuid::new_v4()),
+            ..Default::default()
+        };
+
+        assert!(
+            !is_version_delete_replication(&dobj),
+            "delete-marker creation should remain on the delete-marker replication path"
+        );
+    }
+
+    #[test]
+    fn test_should_retry_delete_marker_purge_for_version_purge() {
+        let dobj = DeletedObject {
+            delete_marker: false,
+            delete_marker_version_id: Some(Uuid::new_v4()),
+            ..Default::default()
+        };
+
+        assert!(
+            should_retry_delete_marker_purge(&dobj),
+            "delete-marker version purge should schedule delayed target cleanup in case the target marker arrives late"
+        );
+    }
+
+    #[test]
+    fn test_should_retry_delete_marker_purge_for_delete_marker_creation() {
+        let dobj = DeletedObject {
+            delete_marker: true,
+            delete_marker_version_id: Some(Uuid::new_v4()),
+            ..Default::default()
+        };
+
+        assert!(
+            should_retry_delete_marker_purge(&dobj),
+            "delete-marker creation should keep the late-arrival cleanup path so downstream purges can catch up"
+        );
+    }
+
+    #[test]
+    fn test_is_retryable_delete_replication_head_error_allows_delete_marker_head_responses() {
+        assert!(
+            !is_retryable_delete_replication_head_error(false, Some("405")),
+            "numeric 405 responses should not block delete-marker purge replication"
+        );
+        assert!(
+            !is_retryable_delete_replication_head_error(false, Some("MethodNotAllowed")),
+            "MethodNotAllowed responses should not block delete-marker purge replication"
+        );
+        assert!(
+            !is_retryable_delete_replication_head_error(true, Some("NoSuchKey")),
+            "not-found responses should not block delete-marker purge replication"
+        );
+        assert!(
+            is_retryable_delete_replication_head_error(false, Some("AccessDenied")),
+            "unexpected head errors should still fail fast"
+        );
+    }
+
+    #[test]
+    fn test_should_count_head_proxy_failure_ignores_not_found_and_405() {
+        assert!(
+            !should_count_head_proxy_failure(true, Some("NoSuchKey"), Some(404)),
+            "not-found heads are expected when the object has not reached the target yet"
+        );
+        assert!(
+            !should_count_head_proxy_failure(false, Some("MethodNotAllowed"), Some(405)),
+            "405 delete-marker probing responses should not be counted as proxy failures"
+        );
+        assert!(
+            !should_count_head_proxy_failure(false, Some("405"), Some(405)),
+            "numeric 405 codes must align with MethodNotAllowed semantics"
+        );
+    }
+
+    #[test]
+    fn test_should_count_head_proxy_failure_counts_unexpected_errors() {
+        assert!(
+            should_count_head_proxy_failure(false, Some("AccessDenied"), Some(403)),
+            "non-NotFound and non-405 service errors should be counted as failures"
+        );
+        assert!(
+            should_count_head_proxy_failure(false, None, Some(500)),
+            "raw 5xx head responses should be counted as proxy failures"
+        );
+    }
+
     #[tokio::test]
     async fn test_get_heal_replicate_object_info_failed_object_returns_heal_roi() {
         let oi = ObjectInfo {

crates/ecstore/src/bucket/replication/replication_state.rs

@@ -12,18 +12,22 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+use crate::bucket::replication::get_global_replication_pool;
 use crate::error::Error;
 use crate::global::get_global_bucket_monitor;
 use rustfs_filemeta::{ReplicatedTargetInfo, ReplicationStatusType, ReplicationType};
 use serde::{Deserialize, Serialize};
-use std::collections::HashMap;
+use std::collections::{HashMap, VecDeque};
 use std::sync::Arc;
 use std::sync::atomic::{AtomicI64, Ordering};
 use std::sync::atomic::{AtomicU64, Ordering as AtomicOrdering};
-use std::time::{Duration, SystemTime};
+use std::time::{Duration, Instant, SystemTime};
 use tokio::sync::{Mutex, RwLock};
 use tokio::time::interval;
 
+const ROLLING_WINDOW: Duration = Duration::from_secs(60);
+const FAILURE_LAST_HOUR_WINDOW: Duration = Duration::from_secs(60 * 60);
+
 /// Exponential Moving Average with thread-safe interior mutability
 #[derive(Debug)]
 pub struct ExponentialMovingAverage {
@@ -328,6 +332,13 @@ pub struct InQueueStats {
     pub now_count: AtomicI64,
 }
 
+#[derive(Debug, Clone)]
+struct QueueSample {
+    observed_at: Instant,
+    bytes: i64,
+    count: i64,
+}
+
 impl Clone for InQueueStats {
     fn clone(&self) -> Self {
         Self {
@@ -359,9 +370,60 @@ pub struct InQueueMetric {
     pub curr: InQueueStats,
     pub avg: InQueueStats,
     pub max: InQueueStats,
+    pub last_minute: InQueueStats,
+    #[serde(skip)]
+    samples: VecDeque<QueueSample>,
 }
 
 impl InQueueMetric {
+    fn observe(&mut self, observed_at: Instant) {
+        let bytes = self.curr.now_bytes.load(Ordering::Relaxed);
+        let count = self.curr.now_count.load(Ordering::Relaxed);
+
+        self.curr.bytes = bytes;
+        self.curr.count = count;
+        self.samples.push_back(QueueSample {
+            observed_at,
+            bytes,
+            count,
+        });
+
+        while self
+            .samples
+            .front()
+            .is_some_and(|sample| observed_at.duration_since(sample.observed_at) > ROLLING_WINDOW)
+        {
+            self.samples.pop_front();
+        }
+
+        if self.samples.is_empty() {
+            self.avg = InQueueStats::default();
+            self.max = InQueueStats::default();
+            self.last_minute = InQueueStats::default();
+            return;
+        }
+
+        let sample_count = self.samples.len() as i64;
+        let total_bytes = self.samples.iter().map(|sample| sample.bytes).sum::<i64>();
+        let total_count = self.samples.iter().map(|sample| sample.count).sum::<i64>();
+        let max_bytes = self.samples.iter().map(|sample| sample.bytes).max().unwrap_or(0);
+        let max_count = self.samples.iter().map(|sample| sample.count).max().unwrap_or(0);
+
+        self.avg.bytes = total_bytes / sample_count;
+        self.avg.count = total_count / sample_count;
+        self.max.bytes = max_bytes;
+        self.max.count = max_count;
+        self.last_minute.bytes = self.avg.bytes;
+        self.last_minute.count = self.avg.count;
+    }
+
+    fn snapshot(&self) -> Self {
+        let mut snapshot = self.clone();
+        snapshot.curr.bytes = snapshot.curr.now_bytes.load(Ordering::Relaxed);
+        snapshot.curr.count = snapshot.curr.now_count.load(Ordering::Relaxed);
+        snapshot
+    }
+
     pub fn merge(&self, other: &InQueueMetric) -> Self {
         Self {
             curr: InQueueStats {
@@ -384,6 +446,12 @@ impl InQueueMetric {
                 count: self.max.count.max(other.max.count),
                 ..Default::default()
             },
+            last_minute: InQueueStats {
+                bytes: self.last_minute.bytes + other.last_minute.bytes,
+                count: self.last_minute.count + other.last_minute.count,
+                ..Default::default()
+            },
+            samples: VecDeque::new(),
         }
     }
 }
@@ -391,8 +459,8 @@ impl InQueueMetric {
 /// Queue cache
 #[derive(Debug, Default)]
 pub struct QueueCache {
-    pub bucket_stats: HashMap<String, InQueueStats>,
-    pub sr_queue_stats: InQueueStats,
+    pub bucket_stats: HashMap<String, InQueueMetric>,
+    pub sr_queue_stats: InQueueMetric,
 }
 
 impl QueueCache {
@@ -401,36 +469,19 @@ impl QueueCache {
     }
 
     pub fn update(&mut self) {
-        // Update queue statistics cache
-        // In actual implementation, this would get latest statistics from queue system
+        let observed_at = Instant::now();
+        self.sr_queue_stats.observe(observed_at);
+        for stats in self.bucket_stats.values_mut() {
+            stats.observe(observed_at);
+        }
     }
 
     pub fn get_bucket_stats(&self, bucket: &str) -> InQueueMetric {
-        if let Some(bucket_stat) = self.bucket_stats.get(bucket) {
-            InQueueMetric {
-                curr: InQueueStats {
-                    bytes: bucket_stat.now_bytes.load(Ordering::Relaxed),
-                    count: bucket_stat.now_count.load(Ordering::Relaxed),
-                    ..Default::default()
-                },
-                avg: InQueueStats::default(), // simplified implementation
-                max: InQueueStats::default(), // simplified implementation
-            }
-        } else {
-            InQueueMetric::default()
-        }
+        self.bucket_stats.get(bucket).map(InQueueMetric::snapshot).unwrap_or_default()
     }
 
     pub fn get_site_stats(&self) -> InQueueMetric {
-        InQueueMetric {
-            curr: InQueueStats {
-                bytes: self.sr_queue_stats.now_bytes.load(Ordering::Relaxed),
-                count: self.sr_queue_stats.now_count.load(Ordering::Relaxed),
-                ..Default::default()
-            },
-            avg: InQueueStats::default(), // simplified implementation
-            max: InQueueStats::default(), // simplified implementation
-        }
+        self.sr_queue_stats.snapshot()
     }
 }
 
@@ -438,8 +489,14 @@ impl QueueCache {
 pub struct ProxyMetric {
     pub get_total: i64,
     pub get_failed: i64,
+    pub get_tag_total: i64,
+    pub get_tag_failed: i64,
     pub put_total: i64,
     pub put_failed: i64,
+    pub put_tag_total: i64,
+    pub put_tag_failed: i64,
+    pub delete_tag_total: i64,
+    pub delete_tag_failed: i64,
     pub head_total: i64,
     pub head_failed: i64,
 }
@@ -448,8 +505,14 @@ impl ProxyMetric {
     pub fn add(&mut self, other: &ProxyMetric) {
         self.get_total += other.get_total;
         self.get_failed += other.get_failed;
+        self.get_tag_total += other.get_tag_total;
+        self.get_tag_failed += other.get_tag_failed;
         self.put_total += other.put_total;
         self.put_failed += other.put_failed;
+        self.put_tag_total += other.put_tag_total;
+        self.put_tag_failed += other.put_tag_failed;
+        self.delete_tag_total += other.delete_tag_total;
+        self.delete_tag_failed += other.delete_tag_failed;
         self.head_total += other.head_total;
         self.head_failed += other.head_failed;
     }
@@ -476,18 +539,36 @@ impl ProxyStatsCache {
                     metric.get_failed += 1;
                 }
             }
+            "GetObjectTagging" => {
+                metric.get_tag_total += 1;
+                if is_err {
+                    metric.get_tag_failed += 1;
+                }
+            }
             "PutObject" => {
                 metric.put_total += 1;
                 if is_err {
                     metric.put_failed += 1;
                 }
             }
+            "PutObjectTagging" => {
+                metric.put_tag_total += 1;
+                if is_err {
+                    metric.put_tag_failed += 1;
+                }
+            }
             "HeadObject" => {
                 metric.head_total += 1;
                 if is_err {
                     metric.head_failed += 1;
                 }
             }
+            "DeleteObjectTagging" => {
+                metric.delete_tag_total += 1;
+                if is_err {
+                    metric.delete_tag_failed += 1;
+                }
+            }
             _ => {}
         }
     }
@@ -505,11 +586,19 @@ impl ProxyStatsCache {
     }
 }
 
+#[derive(Debug, Clone)]
+struct FailureSample {
+    observed_at: Instant,
+    size: i64,
+}
+
 /// Failure statistics
 #[derive(Debug, Clone, Default, Serialize, Deserialize)]
 pub struct FailStats {
     pub count: i64,
     pub size: i64,
+    #[serde(skip)]
+    recent: VecDeque<FailureSample>,
 }
 
 impl FailStats {
@@ -518,14 +607,42 @@ impl FailStats {
     }
 
     pub fn add_size(&mut self, size: i64, _err: Option<&Error>) {
+        let observed_at = Instant::now();
         self.count += 1;
         self.size += size;
+        self.recent.push_back(FailureSample { observed_at, size });
+        self.prune(observed_at);
+    }
+
+    fn prune(&mut self, observed_at: Instant) {
+        while self
+            .recent
+            .front()
+            .is_some_and(|sample| observed_at.duration_since(sample.observed_at) > FAILURE_LAST_HOUR_WINDOW)
+        {
+            self.recent.pop_front();
+        }
+    }
+
+    pub fn recent_since(&self, window: Duration) -> FailedMetric {
+        let now = Instant::now();
+        let mut count = 0i64;
+        let mut size = 0i64;
+        for sample in self.recent.iter().rev() {
+            if now.duration_since(sample.observed_at) > window {
+                break;
+            }
+            count += 1;
+            size += sample.size;
+        }
+        FailedMetric { count, size }
     }
 
     pub fn merge(&self, other: &FailStats) -> Self {
         Self {
             count: self.count + other.count,
             size: self.size + other.size,
+            recent: VecDeque::new(),
         }
     }
 
@@ -674,6 +791,14 @@ pub struct ActiveWorkerStat {
     pub curr: i32,
     pub max: i32,
     pub avg: f64,
+    #[serde(skip)]
+    samples: VecDeque<WorkerSample>,
+}
+
+#[derive(Debug, Clone)]
+struct WorkerSample {
+    observed_at: Instant,
+    workers: i32,
 }
 
 impl ActiveWorkerStat {
@@ -685,9 +810,31 @@ impl ActiveWorkerStat {
         self.clone()
     }
 
-    pub fn update(&mut self) {
-        // Simulate worker statistics update logic
-        // In actual implementation, this would get current active count from worker pool
+    pub fn update(&mut self, curr: i32) {
+        let observed_at = Instant::now();
+        self.curr = curr;
+        self.samples.push_back(WorkerSample {
+            observed_at,
+            workers: curr,
+        });
+
+        while self
+            .samples
+            .front()
+            .is_some_and(|sample| observed_at.duration_since(sample.observed_at) > ROLLING_WINDOW)
+        {
+            self.samples.pop_front();
+        }
+
+        if self.samples.is_empty() {
+            self.max = curr;
+            self.avg = curr as f64;
+            return;
+        }
+
+        self.max = self.samples.iter().map(|sample| sample.workers).max().unwrap_or(curr);
+        let total = self.samples.iter().map(|sample| sample.workers as i64).sum::<i64>();
+        self.avg = total as f64 / self.samples.len() as f64;
     }
 }
 
@@ -740,8 +887,11 @@ impl ReplicationStats {
             let mut interval = interval(Duration::from_secs(2));
             loop {
                 interval.tick().await;
+                let current = get_global_replication_pool()
+                    .map(|pool| pool.active_workers() + pool.active_lrg_workers() + pool.active_mrf_workers())
+                    .unwrap_or(0);
                 let mut workers = workers_clone.lock().await;
-                workers.update();
+                workers.update(current);
             }
         });
 
@@ -832,61 +982,55 @@ impl ReplicationStats {
         let mut rs = ReplStat::new();
 
         match status {
-            ReplicationStatusType::Pending => {
-                if ri.op_type.is_data_replication() && prev_status != status {
-                    rs.set(
-                        ri.arn.clone(),
-                        ri.size,
-                        Duration::default(),
-                        status,
-                        ri.op_type,
-                        ri.endpoint.clone(),
-                        ri.secure,
-                        ri.error.as_ref().map(|e| crate::error::Error::other(e.clone())),
-                    );
-                }
+            ReplicationStatusType::Pending if ri.op_type.is_data_replication() && prev_status != status => {
+                rs.set(
+                    ri.arn.clone(),
+                    ri.size,
+                    Duration::default(),
+                    status,
+                    ri.op_type,
+                    ri.endpoint.clone(),
+                    ri.secure,
+                    ri.error.as_ref().map(|e| crate::error::Error::other(e.clone())),
+                );
             }
-            ReplicationStatusType::Completed => {
-                if ri.op_type.is_data_replication() {
-                    rs.set(
-                        ri.arn.clone(),
-                        ri.size,
-                        ri.duration,
-                        status,
-                        ri.op_type,
-                        ri.endpoint.clone(),
-                        ri.secure,
-                        ri.error.as_ref().map(|e| crate::error::Error::other(e.clone())),
-                    );
-                }
+            ReplicationStatusType::Completed if ri.op_type.is_data_replication() => {
+                rs.set(
+                    ri.arn.clone(),
+                    ri.size,
+                    ri.duration,
+                    status,
+                    ri.op_type,
+                    ri.endpoint.clone(),
+                    ri.secure,
+                    ri.error.as_ref().map(|e| crate::error::Error::other(e.clone())),
+                );
             }
-            ReplicationStatusType::Failed => {
-                if ri.op_type.is_data_replication() && prev_status == ReplicationStatusType::Pending {
-                    rs.set(
-                        ri.arn.clone(),
-                        ri.size,
-                        ri.duration,
-                        status,
-                        ri.op_type,
-                        ri.endpoint.clone(),
-                        ri.secure,
-                        ri.error.as_ref().map(|e| crate::error::Error::other(e.clone())),
-                    );
-                }
+            ReplicationStatusType::Failed
+                if ri.op_type.is_data_replication() && prev_status == ReplicationStatusType::Pending =>
+            {
+                rs.set(
+                    ri.arn.clone(),
+                    ri.size,
+                    ri.duration,
+                    status,
+                    ri.op_type,
+                    ri.endpoint.clone(),
+                    ri.secure,
+                    ri.error.as_ref().map(|e| crate::error::Error::other(e.clone())),
+                );
             }
-            ReplicationStatusType::Replica => {
-                if ri.op_type == ReplicationType::Object {
-                    rs.set(
-                        ri.arn.clone(),
-                        ri.size,
-                        Duration::default(),
-                        status,
-                        ri.op_type,
-                        String::new(),
-                        false,
-                        ri.error.as_ref().map(|e| crate::error::Error::other(e.clone())),
-                    );
-                }
+            ReplicationStatusType::Replica if ri.op_type == ReplicationType::Object => {
+                rs.set(
+                    ri.arn.clone(),
+                    ri.size,
+                    Duration::default(),
+                    status,
+                    ri.op_type,
+                    String::new(),
+                    false,
+                    ri.error.as_ref().map(|e| crate::error::Error::other(e.clone())),
+                );
             }
             _ => {}
         }
@@ -931,14 +1075,26 @@ impl ReplicationStats {
     /// Get replication metrics for all buckets
     pub async fn get_all(&self) -> HashMap<String, BucketReplicationStats> {
         let cache = self.cache.read().await;
-        let mut result = HashMap::new();
+        let mut result = HashMap::with_capacity(cache.len());
 
         for (bucket, stats) in cache.iter() {
-            let mut cloned_stats = stats.clone_stats();
-            // Add queue statistics
+            result.insert(bucket.clone(), stats.clone_stats());
+        }
+        drop(cache);
+
+        {
             let q_cache = self.q_cache.lock().await;
-            cloned_stats.q_stat = q_cache.get_bucket_stats(bucket);
-            result.insert(bucket.clone(), cloned_stats);
+            for (bucket, queue_stats) in &q_cache.bucket_stats {
+                let bucket_stats = result.entry(bucket.clone()).or_insert_with(BucketReplicationStats::new);
+                bucket_stats.q_stat = queue_stats.snapshot();
+            }
+        }
+
+        {
+            let p_cache = self.p_cache.lock().await;
+            for bucket in p_cache.bucket_stats.keys() {
+                result.entry(bucket.clone()).or_insert_with(BucketReplicationStats::new);
+            }
         }
 
         result
@@ -1120,12 +1276,12 @@ impl ReplicationStats {
         let stats = q_cache
             .bucket_stats
             .entry(bucket.to_string())
-            .or_insert_with(InQueueStats::default);
-        stats.now_bytes.fetch_add(size, Ordering::Relaxed);
-        stats.now_count.fetch_add(1, Ordering::Relaxed);
+            .or_insert_with(InQueueMetric::default);
+        stats.curr.now_bytes.fetch_add(size, Ordering::Relaxed);
+        stats.curr.now_count.fetch_add(1, Ordering::Relaxed);
 
-        q_cache.sr_queue_stats.now_bytes.fetch_add(size, Ordering::Relaxed);
-        q_cache.sr_queue_stats.now_count.fetch_add(1, Ordering::Relaxed);
+        q_cache.sr_queue_stats.curr.now_bytes.fetch_add(size, Ordering::Relaxed);
+        q_cache.sr_queue_stats.curr.now_count.fetch_add(1, Ordering::Relaxed);
     }
 
     /// Decrease queue statistics
@@ -1134,12 +1290,12 @@ impl ReplicationStats {
         let stats = q_cache
             .bucket_stats
             .entry(bucket.to_string())
-            .or_insert_with(InQueueStats::default);
-        stats.now_bytes.fetch_sub(size, Ordering::Relaxed);
-        stats.now_count.fetch_sub(1, Ordering::Relaxed);
+            .or_insert_with(InQueueMetric::default);
+        stats.curr.now_bytes.fetch_sub(size, Ordering::Relaxed);
+        stats.curr.now_count.fetch_sub(1, Ordering::Relaxed);
 
-        q_cache.sr_queue_stats.now_bytes.fetch_sub(size, Ordering::Relaxed);
-        q_cache.sr_queue_stats.now_count.fetch_sub(1, Ordering::Relaxed);
+        q_cache.sr_queue_stats.curr.now_bytes.fetch_sub(size, Ordering::Relaxed);
+        q_cache.sr_queue_stats.curr.now_count.fetch_sub(1, Ordering::Relaxed);
     }
 
     /// Increase proxy metrics
@@ -1172,6 +1328,51 @@ mod tests {
         assert_eq!(workers.curr, 0);
     }
 
+    #[test]
+    fn test_in_queue_metric_observe_updates_rolling_stats() {
+        let mut metric = InQueueMetric::default();
+        metric.curr.now_bytes.store(128, Ordering::Relaxed);
+        metric.curr.now_count.store(4, Ordering::Relaxed);
+        metric.observe(Instant::now());
+
+        metric.curr.now_bytes.store(256, Ordering::Relaxed);
+        metric.curr.now_count.store(6, Ordering::Relaxed);
+        metric.observe(Instant::now());
+
+        assert_eq!(metric.curr.bytes, 256);
+        assert_eq!(metric.curr.count, 6);
+        assert_eq!(metric.max.bytes, 256);
+        assert_eq!(metric.max.count, 6);
+        assert_eq!(metric.last_minute.bytes, 192);
+        assert_eq!(metric.last_minute.count, 5);
+    }
+
+    #[test]
+    fn test_fail_stats_recent_since_tracks_windows() {
+        let mut stats = FailStats::default();
+        stats.add_size(64, None);
+        stats.add_size(32, None);
+
+        let last_minute = stats.recent_since(Duration::from_secs(60));
+        let last_hour = stats.recent_since(Duration::from_secs(60 * 60));
+        assert_eq!(last_minute.count, 2);
+        assert_eq!(last_minute.size, 96);
+        assert_eq!(last_hour.count, 2);
+        assert_eq!(last_hour.size, 96);
+    }
+
+    #[test]
+    fn test_active_worker_stat_update_tracks_rolling_avg_and_max() {
+        let mut stats = ActiveWorkerStat::default();
+        stats.update(2);
+        stats.update(6);
+        stats.update(4);
+
+        assert_eq!(stats.curr, 4);
+        assert_eq!(stats.max, 6);
+        assert_eq!(stats.avg, 4.0);
+    }
+
     #[tokio::test]
     async fn test_delete_bucket_stats() {
         let stats = ReplicationStats::new();
@@ -1224,6 +1425,15 @@ mod tests {
         assert_eq!(stat.replicated_count, 1);
     }
 
+    #[tokio::test]
+    async fn test_get_all_includes_proxy_only_bucket() {
+        let stats = ReplicationStats::new();
+        stats.inc_proxy("proxy-only-bucket", "HeadObject", false).await;
+
+        let all = stats.get_all().await;
+        assert!(all.contains_key("proxy-only-bucket"));
+    }
+
     #[test]
     fn test_sr_stats() {
         let sr_stats = SRStats::new();

crates/ecstore/src/cache_value/metacache_set.rs

@@ -12,12 +12,16 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+use crate::disk::disk_store::get_drive_walkdir_stall_timeout;
 use crate::disk::error::DiskError;
 use crate::disk::{self, DiskAPI, DiskStore, WalkDirOptions};
 use futures::future::join_all;
+use metrics::counter;
 use rustfs_filemeta::{MetaCacheEntries, MetaCacheEntry, MetacacheReader, is_io_eof};
-use std::{future::Future, pin::Pin};
+use std::{future::Future, pin::Pin, time::Duration};
+use tokio::io::AsyncRead;
 use tokio::spawn;
+use tokio::time::timeout;
 use tokio_util::sync::CancellationToken;
 use tracing::{error, info, warn};
 
@@ -26,6 +30,21 @@ pub type PartialFn =
     Box<dyn Fn(MetaCacheEntries, &[Option<DiskError>]) -> Pin<Box<dyn Future<Output = ()> + Send>> + Send + 'static>;
 type FinishedFn = Box<dyn Fn(&[Option<DiskError>]) -> Pin<Box<dyn Future<Output = ()> + Send>> + Send + 'static>;
 
+#[derive(Debug)]
+enum PeekOutcome {
+    Ready(Option<MetaCacheEntry>),
+    Error(rustfs_filemeta::Error),
+    TimedOut,
+}
+
+async fn peek_with_timeout<R: AsyncRead + Unpin>(reader: &mut MetacacheReader<R>, timeout_duration: Duration) -> PeekOutcome {
+    match timeout(timeout_duration, reader.peek()).await {
+        Ok(Ok(entry)) => PeekOutcome::Ready(entry),
+        Ok(Err(err)) => PeekOutcome::Error(err),
+        Err(_) => PeekOutcome::TimedOut,
+    }
+}
+
 #[derive(Default)]
 pub struct ListPathRawOptions {
     pub disks: Vec<Option<DiskStore>>,
@@ -160,6 +179,7 @@ pub async fn list_path_raw(rx: CancellationToken, opts: ListPathRawOptions) -> d
     }
 
     let revjob = spawn(async move {
+        let peek_timeout = get_drive_walkdir_stall_timeout();
         let mut errs: Vec<Option<DiskError>> = Vec::with_capacity(readers.len());
         for _ in 0..readers.len() {
             errs.push(None);
@@ -191,8 +211,8 @@ pub async fn list_path_raw(rx: CancellationToken, opts: ListPathRawOptions) -> d
                     continue;
                 }
 
-                let entry = match r.peek().await {
-                    Ok(res) => {
+                let entry = match peek_with_timeout(r, peek_timeout).await {
+                    PeekOutcome::Ready(res) => {
                         if let Some(entry) = res {
                             // info!("read entry disk: {}, name: {}", i, entry.name);
                             entry
@@ -203,7 +223,7 @@ pub async fn list_path_raw(rx: CancellationToken, opts: ListPathRawOptions) -> d
                             continue;
                         }
                     }
-                    Err(err) => {
+                    PeekOutcome::Error(err) => {
                         if err == rustfs_filemeta::Error::Unexpected {
                             at_eof += 1;
                             // warn!("list_path_raw: peek err eof, disk: {}", i);
@@ -236,6 +256,31 @@ pub async fn list_path_raw(rx: CancellationToken, opts: ListPathRawOptions) -> d
                             continue;
                         }
                     }
+                    PeekOutcome::TimedOut => {
+                        has_err += 1;
+                        errs[i] = Some(DiskError::Timeout);
+                        let endpoint = opts
+                            .disks
+                            .get(i)
+                            .and_then(|disk| disk.as_ref().map(|disk| disk.endpoint().to_string()))
+                            .unwrap_or_else(|| "missing".to_string());
+                        counter!(
+                            "rustfs_list_path_raw_stall_total",
+                            "drive" => endpoint.clone()
+                        )
+                        .increment(1);
+                        warn!(
+                            drive = %endpoint,
+                            bucket = %opts.bucket,
+                            path = %opts.path,
+                            timeout_ms = peek_timeout.as_millis(),
+                            "list_path_raw reader peek timed out; excluding drive from current merge"
+                        );
+                        let (detached_rd, write_half) = tokio::io::duplex(1);
+                        drop(write_half);
+                        *r = MetacacheReader::new(detached_rd);
+                        continue;
+                    }
                 };
 
                 // warn!("list_path_raw: loop entry: {:?}, disk: {}", &entry.name, i);
@@ -364,3 +409,42 @@ pub async fn list_path_raw(rx: CancellationToken, opts: ListPathRawOptions) -> d
     // warn!("list_path_raw: done");
     Ok(())
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use rustfs_filemeta::MetacacheWriter;
+
+    #[tokio::test]
+    async fn peek_with_timeout_times_out_on_silent_reader() {
+        let (_writer, reader) = tokio::io::duplex(64);
+        let mut reader = MetacacheReader::new(reader);
+
+        let outcome = peek_with_timeout(&mut reader, Duration::from_millis(20)).await;
+        assert!(matches!(outcome, PeekOutcome::TimedOut));
+    }
+
+    #[tokio::test]
+    async fn peek_with_timeout_reads_entry_before_deadline() {
+        let (reader, writer) = tokio::io::duplex(256);
+        let mut metacache_reader = MetacacheReader::new(reader);
+
+        tokio::spawn(async move {
+            let mut writer = MetacacheWriter::new(writer);
+            let entry = MetaCacheEntry {
+                name: "bucket/object".to_string(),
+                metadata: vec![1, 2, 3],
+                cached: None,
+                reusable: false,
+            };
+            writer.write(&[entry]).await.expect("entry should be written");
+            writer.close().await.expect("writer should close");
+        });
+
+        let outcome = peek_with_timeout(&mut metacache_reader, Duration::from_secs(1)).await;
+        match outcome {
+            PeekOutcome::Ready(Some(entry)) => assert_eq!(entry.name, "bucket/object"),
+            other => panic!("expected ready entry, got {other:?}"),
+        }
+    }
+}

crates/ecstore/src/client/api_error_response.rs

@@ -100,7 +100,7 @@ pub fn http_resp_to_error_response(
     bucket_name: &str,
     object_name: &str,
 ) -> ErrorResponse {
-    let err_body = String::from_utf8(b).unwrap();
+    let err_body = String::from_utf8_lossy(&b).to_string();
     if h.is_empty() || resp_status.is_client_error() || resp_status.is_server_error() {
         return ErrorResponse {
             status_code: resp_status,
@@ -178,36 +178,46 @@ pub fn http_resp_to_error_response(
                 };
             }
         }
-    } else {
-        err_resp = err_resp_.unwrap();
+    } else if let Ok(parsed_resp) = err_resp_ {
+        err_resp = parsed_resp;
     }
     err_resp.status_code = resp_status;
     if let Some(server_name) = h.get("Server") {
-        err_resp.server = server_name.to_str().expect("err").to_string();
+        if let Ok(server_str) = server_name.to_str() {
+            err_resp.server = server_str.to_string();
+        }
     }
 
-    let code = h.get("x-minio-error-code");
-    if code.is_some() {
-        err_resp.code = S3ErrorCode::Custom(code.expect("err").to_str().expect("err").into());
+    if let Some(code) = h.get("x-minio-error-code") {
+        if let Ok(code_str) = code.to_str() {
+            err_resp.code = S3ErrorCode::Custom(code_str.into());
+        }
     }
-    let desc = h.get("x-minio-error-desc");
-    if desc.is_some() {
-        err_resp.message = desc.expect("err").to_str().expect("err").trim_matches('"').to_string();
+    if let Some(desc) = h.get("x-minio-error-desc") {
+        if let Ok(desc_str) = desc.to_str() {
+            err_resp.message = desc_str.trim_matches('"').to_string();
+        }
     }
 
     if err_resp.request_id == "" {
         if let Some(x_amz_request_id) = h.get("x-amz-request-id") {
-            err_resp.request_id = x_amz_request_id.to_str().expect("err").to_string();
+            if let Ok(request_id_str) = x_amz_request_id.to_str() {
+                err_resp.request_id = request_id_str.to_string();
+            }
         }
     }
     if err_resp.host_id == "" {
         if let Some(x_amz_id_2) = h.get("x-amz-id-2") {
-            err_resp.host_id = x_amz_id_2.to_str().expect("err").to_string();
+            if let Ok(host_id_str) = x_amz_id_2.to_str() {
+                err_resp.host_id = host_id_str.to_string();
+            }
         }
     }
     if err_resp.region == "" {
         if let Some(x_amz_bucket_region) = h.get("x-amz-bucket-region") {
-            err_resp.region = x_amz_bucket_region.to_str().expect("err").to_string();
+            if let Ok(region_str) = x_amz_bucket_region.to_str() {
+                err_resp.region = region_str.to_string();
+            }
         }
     }
     if err_resp.code == S3ErrorCode::InvalidLocationConstraint/*InvalidRegion*/ && err_resp.region != "" {

crates/ecstore/src/client/api_get_object.rs

@@ -1,4 +1,3 @@
-#![allow(clippy::map_entry)]
 // Copyright 2024 RustFS Team
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
@@ -12,6 +11,7 @@
 // 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.
+#![allow(clippy::map_entry)]
 #![allow(unused_imports)]
 #![allow(unused_variables)]
 #![allow(unused_mut)]

crates/ecstore/src/client/api_put_object.rs

@@ -137,21 +137,21 @@ impl Default for PutObjectOptions {
 impl PutObjectOptions {
     fn set_match_etag(&mut self, etag: &str) {
         if etag == "*" {
-            self.custom_header
-                .insert("If-Match", HeaderValue::from_str("*").expect("err"));
+            self.custom_header.insert("If-Match", HeaderValue::from_static("*"));
         } else {
-            self.custom_header
-                .insert("If-Match", HeaderValue::from_str(&format!("\"{}\"", etag)).expect("err"));
+            if let Ok(etag_value) = HeaderValue::from_str(&format!("\"{}\"", etag)) {
+                self.custom_header.insert("If-Match", etag_value);
+            }
         }
     }
 
     fn set_match_etag_except(&mut self, etag: &str) {
         if etag == "*" {
-            self.custom_header
-                .insert("If-None-Match", HeaderValue::from_str("*").expect("err"));
+            self.custom_header.insert("If-None-Match", HeaderValue::from_static("*"));
         } else {
-            self.custom_header
-                .insert("If-None-Match", HeaderValue::from_str(&format!("\"{etag}\"")).expect("err"));
+            if let Ok(etag_value) = HeaderValue::from_str(&format!("\"{etag}\"")) {
+                self.custom_header.insert("If-None-Match", etag_value);
+            }
         }
     }
 
@@ -162,59 +162,75 @@ impl PutObjectOptions {
         if content_type == "" {
             content_type = "application/octet-stream".to_string();
         }
-        header.insert("Content-Type", HeaderValue::from_str(&content_type).expect("err"));
+        if let Ok(content_type_value) = HeaderValue::from_str(&content_type) {
+            header.insert("Content-Type", content_type_value);
+        }
 
         if self.content_encoding != "" {
-            header.insert("Content-Encoding", HeaderValue::from_str(&self.content_encoding).expect("err"));
+            if let Ok(encoding_value) = HeaderValue::from_str(&self.content_encoding) {
+                header.insert("Content-Encoding", encoding_value);
+            }
         }
         if self.content_disposition != "" {
-            header.insert("Content-Disposition", HeaderValue::from_str(&self.content_disposition).expect("err"));
+            if let Ok(disposition_value) = HeaderValue::from_str(&self.content_disposition) {
+                header.insert("Content-Disposition", disposition_value);
+            }
         }
         if self.content_language != "" {
-            header.insert("Content-Language", HeaderValue::from_str(&self.content_language).expect("err"));
+            if let Ok(language_value) = HeaderValue::from_str(&self.content_language) {
+                header.insert("Content-Language", language_value);
+            }
         }
         if self.cache_control != "" {
-            header.insert("Cache-Control", HeaderValue::from_str(&self.cache_control).expect("err"));
+            if let Ok(cache_value) = HeaderValue::from_str(&self.cache_control) {
+                header.insert("Cache-Control", cache_value);
+            }
         }
 
         if self.expires.unix_timestamp() != 0 {
-            header.insert(
-                "Expires",
-                HeaderValue::from_str(&self.expires.format(ISO8601_DATEFORMAT).unwrap()).expect("err"),
-            ); //rustfs invalid header
+            if let Ok(expires_str) = self.expires.format(ISO8601_DATEFORMAT) {
+                if let Ok(expires_value) = HeaderValue::from_str(&expires_str) {
+                    header.insert("Expires", expires_value);
+                }
+            }
         }
 
         if self.mode.as_str() != "" {
-            header.insert(X_AMZ_OBJECT_LOCK_MODE, HeaderValue::from_str(self.mode.as_str()).expect("err"));
+            if let Ok(mode_value) = HeaderValue::from_str(self.mode.as_str()) {
+                header.insert(X_AMZ_OBJECT_LOCK_MODE, mode_value);
+            }
         }
 
         if self.retain_until_date.unix_timestamp() != 0 {
-            header.insert(
-                X_AMZ_OBJECT_LOCK_RETAIN_UNTIL_DATE,
-                HeaderValue::from_str(&self.retain_until_date.format(ISO8601_DATEFORMAT).unwrap()).expect("err"),
-            );
+            if let Ok(retain_str) = self.retain_until_date.format(ISO8601_DATEFORMAT) {
+                if let Ok(retain_value) = HeaderValue::from_str(&retain_str) {
+                    header.insert(X_AMZ_OBJECT_LOCK_RETAIN_UNTIL_DATE, retain_value);
+                }
+            }
         }
 
         if self.legalhold.as_str() != "" {
-            header.insert(X_AMZ_OBJECT_LOCK_LEGAL_HOLD, HeaderValue::from_str(self.legalhold.as_str()).expect("err"));
+            if let Ok(legalhold_value) = HeaderValue::from_str(self.legalhold.as_str()) {
+                header.insert(X_AMZ_OBJECT_LOCK_LEGAL_HOLD, legalhold_value);
+            }
         }
 
         if self.storage_class != "" {
-            header.insert(X_AMZ_STORAGE_CLASS, HeaderValue::from_str(&self.storage_class).expect("err"));
+            if let Ok(storage_class_value) = HeaderValue::from_str(&self.storage_class) {
+                header.insert(X_AMZ_STORAGE_CLASS, storage_class_value);
+            }
         }
 
         if self.website_redirect_location != "" {
-            header.insert(
-                X_AMZ_WEBSITE_REDIRECT_LOCATION,
-                HeaderValue::from_str(&self.website_redirect_location).expect("err"),
-            );
+            if let Ok(redirect_value) = HeaderValue::from_str(&self.website_redirect_location) {
+                header.insert(X_AMZ_WEBSITE_REDIRECT_LOCATION, redirect_value);
+            }
         }
 
         if !self.internal.replication_status.as_str().is_empty() {
-            header.insert(
-                X_AMZ_REPLICATION_STATUS,
-                HeaderValue::from_str(self.internal.replication_status.as_str()).expect("err"),
-            );
+            if let Ok(replication_status_value) = HeaderValue::from_str(self.internal.replication_status.as_str()) {
+                header.insert(X_AMZ_REPLICATION_STATUS, replication_status_value);
+            }
         }
 
         for (k, v) in &self.user_metadata {
@@ -360,17 +376,21 @@ impl TransitionClient {
 
             let mut md5_base64: String = "".to_string();
             if opts.send_content_md5 {
-                let mut md5_hasher = self.md5_hasher.lock().unwrap();
-                let hash = md5_hasher.as_mut().expect("err");
-                let hash = hash.hash_encode(&buf[..length]);
-                md5_base64 = base64_encode(hash.as_ref());
+                if let Some(mut md5_hasher) = self.md5_hasher.lock().unwrap().as_mut() {
+                    let hash = md5_hasher.hash_encode(&buf[..length]);
+                    md5_base64 = base64_encode(hash.as_ref());
+                }
             } else {
                 let mut crc = opts.auto_checksum.hasher()?;
                 crc.update(&buf[..length]);
                 let csum = crc.finalize();
 
                 if let Ok(header_name) = HeaderName::from_bytes(opts.auto_checksum.key().as_bytes()) {
-                    custom_header.insert(header_name, base64_encode(csum.as_ref()).parse().expect("err"));
+                    if let Ok(header_value) = base64_encode(csum.as_ref()).parse() {
+                        custom_header.insert(header_name, header_value);
+                    } else {
+                        warn!("Failed to parse checksum value");
+                    }
                 } else {
                     warn!("Invalid header name: {}", opts.auto_checksum.key());
                 }

crates/ecstore/src/client/api_put_object_multipart.rs

@@ -127,7 +127,11 @@ impl TransitionClient {
                 let csum = crc.finalize();
 
                 if let Ok(header_name) = HeaderName::from_bytes(opts.auto_checksum.key().as_bytes()) {
-                    custom_header.insert(header_name, base64_encode(csum.as_ref()).parse().expect("err"));
+                    if let Ok(header_value) = base64_encode(csum.as_ref()).parse() {
+                        custom_header.insert(header_name, header_value);
+                    } else {
+                        warn!("Failed to parse checksum value");
+                    }
                 } else {
                     warn!("Invalid header name: {}", opts.auto_checksum.key());
                 }
@@ -309,27 +313,27 @@ impl TransitionClient {
         let h = resp.headers();
         let mut obj_part = ObjectPart {
             checksum_crc32: if let Some(h_checksum_crc32) = h.get(ChecksumMode::ChecksumCRC32.key()) {
-                h_checksum_crc32.to_str().expect("err").to_string()
+                h_checksum_crc32.to_str().unwrap_or("").to_string()
             } else {
                 "".to_string()
             },
             checksum_crc32c: if let Some(h_checksum_crc32c) = h.get(ChecksumMode::ChecksumCRC32C.key()) {
-                h_checksum_crc32c.to_str().expect("err").to_string()
+                h_checksum_crc32c.to_str().unwrap_or("").to_string()
             } else {
                 "".to_string()
             },
             checksum_sha1: if let Some(h_checksum_sha1) = h.get(ChecksumMode::ChecksumSHA1.key()) {
-                h_checksum_sha1.to_str().expect("err").to_string()
+                h_checksum_sha1.to_str().unwrap_or("").to_string()
             } else {
                 "".to_string()
             },
             checksum_sha256: if let Some(h_checksum_sha256) = h.get(ChecksumMode::ChecksumSHA256.key()) {
-                h_checksum_sha256.to_str().expect("err").to_string()
+                h_checksum_sha256.to_str().unwrap_or("").to_string()
             } else {
                 "".to_string()
             },
             checksum_crc64nvme: if let Some(h_checksum_crc64nvme) = h.get(ChecksumMode::ChecksumCRC64NVME.key()) {
-                h_checksum_crc64nvme.to_str().expect("err").to_string()
+                h_checksum_crc64nvme.to_str().unwrap_or("").to_string()
             } else {
                 "".to_string()
             },
@@ -338,7 +342,7 @@ impl TransitionClient {
         obj_part.size = p.size;
         obj_part.part_num = p.part_number;
         obj_part.etag = if let Some(h_etag) = h.get("ETag") {
-            h_etag.to_str().expect("err").trim_matches('"').to_string()
+            h_etag.to_str().unwrap_or("").trim_matches('"').to_string()
         } else {
             "".to_string()
         };
@@ -398,7 +402,7 @@ impl TransitionClient {
             key: complete_multipart_upload_result.key,
             etag: trim_etag(&complete_multipart_upload_result.etag),
             version_id: if let Some(h_x_amz_version_id) = h.get(X_AMZ_VERSION_ID) {
-                h_x_amz_version_id.to_str().expect("err").to_string()
+                h_x_amz_version_id.to_str().unwrap_or("").to_string()
             } else {
                 "".to_string()
             },

crates/ecstore/src/client/api_put_object_streaming.rs

@@ -21,6 +21,7 @@
 use bytes::Bytes;
 use futures::future::join_all;
 use http::{HeaderMap, HeaderName, HeaderValue, StatusCode};
+use std::io::Error;
 use std::sync::RwLock;
 use std::{collections::HashMap, sync::Arc};
 use time::{OffsetDateTime, format_description};
@@ -152,7 +153,10 @@ impl TransitionClient {
 
             if opts.send_content_md5 {
                 let mut md5_hasher = self.md5_hasher.lock().unwrap();
-                let md5_hash = md5_hasher.as_mut().expect("err");
+                let md5_hash = match md5_hasher.as_mut() {
+                    Some(hasher) => hasher,
+                    None => return Err(std::io::Error::other("MD5 hasher not initialized")),
+                };
                 let hash = md5_hash.hash_encode(&buf[..length]);
                 md5_base64 = base64_encode(hash.as_ref());
             } else {
@@ -161,7 +165,11 @@ impl TransitionClient {
                 let csum = crc.finalize();
 
                 if let Ok(header_name) = HeaderName::from_bytes(opts.auto_checksum.key().as_bytes()) {
-                    custom_header.insert(header_name, base64_encode(csum.as_ref()).parse().expect("err"));
+                    if let Ok(header_value) = base64_encode(csum.as_ref()).parse() {
+                        custom_header.insert(header_name, header_value);
+                    } else {
+                        warn!("Failed to parse checksum value");
+                    }
                 } else {
                     warn!("Invalid header name: {}", opts.auto_checksum.key());
                 }
@@ -275,11 +283,14 @@ impl TransitionClient {
         for part_number in 1..=total_parts_count {
             let mut buf = Vec::<u8>::new();
             select! {
-                buf = bufs_rx.recv() => {}
+                buf1 = bufs_rx.recv() => {
+                    if let Some(buf1) = buf1 {
+                        buf = buf1;
+                    }
+                }
                 err = err_rx.recv() => {
                     //cancel_token.cancel();
-                    //wg.Wait()
-                    return Err(err.expect("err"));
+                    return Err(err.unwrap_or_else(|| std::io::Error::other("Unknown error received from channel")));
                 }
                 else => (),
             }
@@ -309,7 +320,11 @@ impl TransitionClient {
                 let csum = crc.finalize();
 
                 if let Ok(header_name) = HeaderName::from_bytes(opts.auto_checksum.key().as_bytes()) {
-                    custom_header.insert(header_name, base64_encode(csum.as_ref()).parse().expect("err"));
+                    if let Ok(header_value) = base64_encode(csum.as_ref()).parse() {
+                        custom_header.insert(header_name, header_value);
+                    } else {
+                        warn!("Failed to parse checksum value");
+                    }
                 } else {
                     warn!("Invalid header name: {}", opts.auto_checksum.key());
                 }
@@ -319,12 +334,19 @@ impl TransitionClient {
             let clone_parts_info = parts_info.clone();
             let clone_upload_id = upload_id.clone();
             let clone_self = self.clone();
+            let err_tx_clone = err_tx.clone();
             futures.push(async move {
                 let mut md5_base64: String = "".to_string();
 
                 if opts.send_content_md5 {
                     let mut md5_hasher = clone_self.md5_hasher.lock().unwrap();
-                    let md5_hash = md5_hasher.as_mut().expect("err");
+                    let md5_hash = match md5_hasher.as_mut() {
+                        Some(hasher) => hasher,
+                        None => {
+                            //let _ = err_tx_clone.send(std::io::Error::other("MD5 hasher not initialized")).await;
+                            return Ok::<(), Error>(());
+                        }
+                    };
                     let hash = md5_hash.hash_encode(&buf[..length]);
                     md5_base64 = base64_encode(hash.as_ref());
                 }
@@ -344,12 +366,21 @@ impl TransitionClient {
                     sha256_hex: "".to_string(),
                     trailer: HeaderMap::new(),
                 };
-                let obj_part = clone_self.upload_part(&mut p).await.expect("err");
+                let obj_part = match clone_self.upload_part(&mut p).await {
+                    Ok(part) => part,
+                    Err(err) => {
+                        let _ = err_tx_clone.send(std::io::Error::other(err.to_string())).await;
+                        return Err::<(), Error>(err);
+                    }
+                };
 
-                let mut clone_parts_info = clone_parts_info.write().unwrap();
-                clone_parts_info.entry(part_number).or_insert(obj_part);
+                {
+                    let mut clone_parts_info = clone_parts_info.write().unwrap();
+                    clone_parts_info.entry(part_number).or_insert(obj_part);
+                }
 
-                clone_bufs_tx.send(buf);
+                let _ = clone_bufs_tx.send(buf).await;
+                Ok::<(), Error>(())
             });
 
             total_uploaded_size += length as i64;
@@ -359,7 +390,7 @@ impl TransitionClient {
 
         select! {
             err = err_rx.recv() => {
-                return Err(err.expect("err"));
+                return Err(err.unwrap_or_else(|| std::io::Error::other("Unknown error received from channel")));
             }
             else => (),
         }
@@ -504,9 +535,10 @@ impl TransitionClient {
         Ok(UploadInfo {
             bucket: bucket_name.to_string(),
             key: object_name.to_string(),
-            etag: trim_etag(h.get("ETag").expect("err").to_str().expect("err")),
+            etag: trim_etag(h.get("ETag").and_then(|v| v.to_str().ok()).unwrap_or("")),
+
             version_id: if let Some(h_x_amz_version_id) = h.get(X_AMZ_VERSION_ID) {
-                h_x_amz_version_id.to_str().expect("err").to_string()
+                h_x_amz_version_id.to_str().unwrap_or("").to_string()
             } else {
                 "".to_string()
             },
@@ -514,27 +546,27 @@ impl TransitionClient {
             expiration: exp_time,
             expiration_rule_id: rule_id,
             checksum_crc32: if let Some(h_checksum_crc32) = h.get(ChecksumMode::ChecksumCRC32.key()) {
-                h_checksum_crc32.to_str().expect("err").to_string()
+                h_checksum_crc32.to_str().unwrap_or("").to_string()
             } else {
                 "".to_string()
             },
             checksum_crc32c: if let Some(h_checksum_crc32c) = h.get(ChecksumMode::ChecksumCRC32C.key()) {
-                h_checksum_crc32c.to_str().expect("err").to_string()
+                h_checksum_crc32c.to_str().unwrap_or("").to_string()
             } else {
                 "".to_string()
             },
             checksum_sha1: if let Some(h_checksum_sha1) = h.get(ChecksumMode::ChecksumSHA1.key()) {
-                h_checksum_sha1.to_str().expect("err").to_string()
+                h_checksum_sha1.to_str().unwrap_or("").to_string()
             } else {
                 "".to_string()
             },
             checksum_sha256: if let Some(h_checksum_sha256) = h.get(ChecksumMode::ChecksumSHA256.key()) {
-                h_checksum_sha256.to_str().expect("err").to_string()
+                h_checksum_sha256.to_str().unwrap_or("").to_string()
             } else {
                 "".to_string()
             },
             checksum_crc64nvme: if let Some(h_checksum_crc64nvme) = h.get(ChecksumMode::ChecksumCRC64NVME.key()) {
-                h_checksum_crc64nvme.to_str().expect("err").to_string()
+                h_checksum_crc64nvme.to_str().unwrap_or("").to_string()
             } else {
                 "".to_string()
             },

crates/ecstore/src/client/api_remove.rs

@@ -25,7 +25,7 @@ use hyper::body::Bytes;
 use rustfs_utils::HashAlgorithm;
 use s3s::S3ErrorCode;
 use s3s::dto::ReplicationStatus;
-use s3s::header::X_AMZ_BYPASS_GOVERNANCE_RETENTION;
+use s3s::header::{X_AMZ_BYPASS_GOVERNANCE_RETENTION, X_AMZ_DELETE_MARKER, X_AMZ_VERSION_ID};
 use serde::Deserialize;
 use std::fmt::Display;
 use std::{
@@ -111,8 +111,9 @@ impl TransitionClient {
             .await?;
 
         {
-            let mut bucket_loc_cache = self.bucket_loc_cache.lock().unwrap();
-            bucket_loc_cache.delete(bucket_name);
+            if let Ok(mut bucket_loc_cache) = self.bucket_loc_cache.lock() {
+                bucket_loc_cache.delete(bucket_name);
+            }
         }
         Ok(())
     }
@@ -142,8 +143,9 @@ impl TransitionClient {
             .await?;
 
         {
-            let mut bucket_loc_cache = self.bucket_loc_cache.lock().unwrap();
-            bucket_loc_cache.delete(bucket_name);
+            if let Ok(mut bucket_loc_cache) = self.bucket_loc_cache.lock() {
+                bucket_loc_cache.delete(bucket_name);
+            }
         }
 
         Ok(())
@@ -168,7 +170,7 @@ impl TransitionClient {
         let mut headers = HeaderMap::new();
 
         if opts.governance_bypass {
-            headers.insert(X_AMZ_BYPASS_GOVERNANCE_RETENTION, "true".parse().expect("err")); //amzBypassGovernance
+            headers.insert(X_AMZ_BYPASS_GOVERNANCE_RETENTION, HeaderValue::from_static("true")); //amzBypassGovernance
         }
 
         let resp = self
@@ -197,13 +199,12 @@ impl TransitionClient {
         Ok(RemoveObjectResult {
             object_name: object_name.to_string(),
             object_version_id: opts.version_id,
-            delete_marker: resp.headers().get("x-amz-delete-marker").expect("err") == "true",
+            delete_marker: resp.headers().get(X_AMZ_DELETE_MARKER).map_or(false, |v| v == "true"),
             delete_marker_version_id: resp
                 .headers()
-                .get("x-amz-version-id")
-                .expect("err")
-                .to_str()
-                .expect("err")
+                .get(X_AMZ_VERSION_ID)
+                .and_then(|v| v.to_str().ok())
+                .unwrap_or_default()
                 .to_string(),
             ..Default::default()
         })
@@ -290,15 +291,15 @@ impl TransitionClient {
                             bucket_name,
                             &object.name,
                             RemoveObjectOptions {
-                                version_id: object.version_id.expect("err").to_string(),
+                                version_id: object.version_id.map(|id| id.to_string()).unwrap_or_default(),
                                 governance_bypass: opts.governance_bypass,
                                 ..Default::default()
                             },
                         )
                         .await?;
                     let remove_result_clone = remove_result.clone();
-                    if !remove_result.err.is_none() {
-                        match to_error_response(&remove_result.err.expect("err")).code {
+                    if let Some(err) = &remove_result.err {
+                        match to_error_response(err).code {
                             S3ErrorCode::InvalidArgument | S3ErrorCode::NoSuchVersion => {
                                 continue;
                             }
@@ -326,7 +327,7 @@ impl TransitionClient {
 
             let mut headers = HeaderMap::new();
             if opts.governance_bypass {
-                headers.insert(X_AMZ_BYPASS_GOVERNANCE_RETENTION, "true".parse().expect("err"));
+                headers.insert(X_AMZ_BYPASS_GOVERNANCE_RETENTION, HeaderValue::from_static("true"));
             }
 
             let remove_bytes = generate_remove_multi_objects_request(&batch);
@@ -423,23 +424,20 @@ impl TransitionClient {
                         request_id: resp
                             .headers()
                             .get("x-amz-request-id")
-                            .expect("err")
-                            .to_str()
-                            .expect("err")
+                            .and_then(|v| v.to_str().ok())
+                            .unwrap_or_default()
                             .to_string(),
                         host_id: resp
                             .headers()
                             .get("x-amz-id-2")
-                            .expect("err")
-                            .to_str()
-                            .expect("err")
+                            .and_then(|v| v.to_str().ok())
+                            .unwrap_or_default()
                             .to_string(),
                         region: resp
                             .headers()
                             .get("x-amz-bucket-region")
-                            .expect("err")
-                            .to_str()
-                            .expect("err")
+                            .and_then(|v| v.to_str().ok())
+                            .unwrap_or_default()
                             .to_string(),
                         ..Default::default()
                     };
@@ -472,10 +470,11 @@ pub struct RemoveObjectError {
 
 impl Display for RemoveObjectError {
     fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
-        if self.err.is_none() {
-            return write!(f, "unexpected remove object error result");
+        if let Some(err) = &self.err {
+            write!(f, "{}", err.to_string())
+        } else {
+            write!(f, "unexpected remove object error result")
         }
-        write!(f, "{}", self.err.as_ref().expect("err").to_string())
     }
 }
 

crates/ecstore/src/client/bucket_cache.rs

@@ -22,6 +22,7 @@ use super::constants::UNSIGNED_PAYLOAD;
 use super::credentials::SignatureType;
 use crate::client::{
     api_error_response::http_resp_to_error_response,
+    signer_error,
     transition_api::{CreateBucketConfiguration, LocationConstraint, TransitionClient},
 };
 use http::Request;
@@ -35,6 +36,10 @@ use rustfs_utils::hash::EMPTY_STRING_SHA256_HASH;
 use s3s::S3ErrorCode;
 use std::collections::HashMap;
 
+fn signer_error_to_io_error(scope: &str, error: rustfs_signer::SignV4Error) -> std::io::Error {
+    signer_error::signer_error_to_io_error(scope, error)
+}
+
 #[derive(Debug, Clone)]
 pub struct BucketLocationCache {
     items: HashMap<String, String>,
@@ -70,10 +75,10 @@ impl TransitionClient {
 
         let mut location;
         {
-            let mut bucket_loc_cache = self.bucket_loc_cache.lock().unwrap();
-            let ret = bucket_loc_cache.get(bucket_name);
-            if let Some(location) = ret {
-                return Ok(location);
+            if let Ok(bucket_loc_cache) = self.bucket_loc_cache.lock() {
+                if let Some(location) = bucket_loc_cache.get(bucket_name) {
+                    return Ok(location);
+                }
             }
             //location = ret?;
         }
@@ -83,8 +88,9 @@ impl TransitionClient {
         let mut resp = self.doit(req).await?;
         location = process_bucket_location_response(resp, bucket_name, &self.tier_type).await?;
         {
-            let mut bucket_loc_cache = self.bucket_loc_cache.lock().unwrap();
-            bucket_loc_cache.set(bucket_name, &location);
+            if let Ok(mut bucket_loc_cache) = self.bucket_loc_cache.lock() {
+                bucket_loc_cache.set(bucket_name, &location);
+            }
         }
         Ok(location)
     }
@@ -108,7 +114,11 @@ impl TransitionClient {
             url_str.push_str("://");
             url_str.push_str(bucket_name);
             url_str.push_str(".");
-            url_str.push_str(target_url.host_str().expect("err"));
+            url_str.push_str(
+                target_url
+                    .host_str()
+                    .ok_or_else(|| std::io::Error::new(std::io::ErrorKind::InvalidInput, "host is none"))?,
+            );
             url_str.push_str("/?location");
         } else {
             let mut path = bucket_name.to_string();
@@ -135,13 +145,16 @@ impl TransitionClient {
 
         let value;
         {
-            let mut creds_provider = self.creds_provider.lock().unwrap();
-            value = match creds_provider.get_with_context(Some(self.cred_context())) {
-                Ok(v) => v,
-                Err(err) => {
-                    return Err(std::io::Error::other(err));
-                }
-            };
+            if let Ok(mut creds_provider) = self.creds_provider.lock() {
+                value = match creds_provider.get_with_context(Some(self.cred_context())) {
+                    Ok(v) => v,
+                    Err(err) => {
+                        return Err(std::io::Error::other(err));
+                    }
+                };
+            } else {
+                return Err(std::io::Error::other("Failed to acquire credentials provider lock"));
+            }
         }
 
         let mut signer_type = value.signer_type.clone();
@@ -171,9 +184,15 @@ impl TransitionClient {
             content_sha256 = UNSIGNED_PAYLOAD.to_string();
         }
 
-        req.headers_mut()
-            .insert("X-Amz-Content-Sha256", content_sha256.parse().unwrap());
-        let req = rustfs_signer::sign_v4(req, 0, &access_key_id, &secret_access_key, &session_token, "us-east-1");
+        let content_sha256_value = content_sha256.parse().map_err(|err| {
+            std::io::Error::new(
+                std::io::ErrorKind::InvalidInput,
+                format!("invalid X-Amz-Content-Sha256 header value: {err}"),
+            )
+        })?;
+        req.headers_mut().insert("X-Amz-Content-Sha256", content_sha256_value);
+        let req = rustfs_signer::try_sign_v4(req, 0, &access_key_id, &secret_access_key, &session_token, "us-east-1")
+            .map_err(|err| signer_error_to_io_error("failed to sign bucket location request", err))?;
         Ok(req)
     }
 }
@@ -228,13 +247,16 @@ async fn process_bucket_location_response(
     }
     let mut location = "".to_string();
     if tier_type == "huaweicloud" {
-        let d = quick_xml::de::from_str::<CreateBucketConfiguration>(&String::from_utf8(body_vec).unwrap()).unwrap();
-        location = d.location_constraint;
+        if let Ok(body_str) = String::from_utf8(body_vec) {
+            if let Ok(d) = quick_xml::de::from_str::<CreateBucketConfiguration>(&body_str) {
+                location = d.location_constraint;
+            }
+        }
     } else {
-        if let Ok(LocationConstraint { field }) =
-            quick_xml::de::from_str::<LocationConstraint>(&String::from_utf8(body_vec).unwrap())
-        {
-            location = field;
+        if let Ok(body_str) = String::from_utf8(body_vec) {
+            if let Ok(LocationConstraint { field }) = quick_xml::de::from_str::<LocationConstraint>(&body_str) {
+                location = field;
+            }
         }
     }
     //debug!("location: {}", location);

crates/ecstore/src/client/credentials.rs

@@ -1,4 +1,3 @@
-#![allow(unused_imports)]
 // Copyright 2024 RustFS Team
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
@@ -12,6 +11,7 @@
 // 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.
+#![allow(unused_imports)]
 #![allow(unused_variables)]
 #![allow(unused_mut)]
 #![allow(unused_assignments)]
@@ -57,7 +57,9 @@ impl<P: Provider + Default> Credentials<P> {
 
     pub fn get_with_context(&mut self, mut cc: Option<CredContext>) -> Result<Value, std::io::Error> {
         if self.is_expired() {
-            let creds = self.provider.retrieve_with_cred_context(cc.expect("err"));
+            let creds = self.provider.retrieve_with_cred_context(cc.unwrap_or(CredContext {
+                endpoint: "".to_string(),
+            }));
             self.creds = creds;
             self.force_refresh = false;
         }

crates/ecstore/src/client/mod.rs

@@ -35,5 +35,6 @@ pub mod constants;
 pub mod credentials;
 pub mod object_api_utils;
 pub mod object_handlers_common;
+pub mod signer_error;
 pub mod transition_api;
 pub mod utils;

crates/ecstore/src/client/object_handlers_common.rs

@@ -12,14 +12,38 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+use std::sync::Arc;
+use tracing::warn;
+
 use crate::bucket::lifecycle::lifecycle;
+use crate::bucket::replication::{DeletedObjectReplicationInfo, check_replicate_delete, schedule_replication_delete};
 use crate::bucket::versioning::VersioningApi;
 use crate::bucket::versioning_sys::BucketVersioningSys;
 use crate::store::ECStore;
 use crate::store_api::{ObjectOperations, ObjectOptions, ObjectToDelete};
+use rustfs_filemeta::{REPLICATE_INCOMING_DELETE, ReplicationState, version_purge_statuses_map};
 use rustfs_lock::MAX_DELETE_LIST;
 
-pub async fn delete_object_versions(api: ECStore, bucket: &str, to_del: &[ObjectToDelete], _lc_event: lifecycle::Event) {
+fn lifecycle_version_delete_replication_state(
+    replicate_decision_str: String,
+    pending_status: Option<String>,
+) -> ReplicationState {
+    ReplicationState {
+        replicate_decision_str,
+        version_purge_status_internal: pending_status.clone(),
+        purge_targets: version_purge_statuses_map(pending_status.as_deref().unwrap_or_default()),
+        ..Default::default()
+    }
+}
+
+pub async fn delete_object_versions(api: &Arc<ECStore>, bucket: &str, to_del: &[ObjectToDelete], _lc_event: lifecycle::Event) {
+    let version_suspended = match BucketVersioningSys::get(bucket).await {
+        Ok(vc) => vc.suspended(),
+        Err(err) => {
+            warn!(bucket, error = ?err, "failed to get versioning config during lifecycle noncurrent version cleanup");
+            return;
+        }
+    };
     let mut remaining = to_del;
     loop {
         let mut to_del = remaining;
@@ -29,15 +53,94 @@ pub async fn delete_object_versions(api: ECStore, bucket: &str, to_del: &[Object
         } else {
             remaining = &[];
         }
-        let vc = BucketVersioningSys::get(bucket).await.expect("err!");
-        let _deleted_objs = api.delete_objects(
-            bucket,
-            to_del.to_vec(),
-            ObjectOptions {
-                //prefix_enabled_fn:  vc.prefix_enabled(""),
-                version_suspended: vc.suspended(),
+
+        let mut replication_candidates: Vec<Option<ReplicationState>> = Vec::with_capacity(to_del.len());
+        for object in to_del.iter() {
+            let version_id = object.version_id.map(|vid| vid.to_string());
+            let opts = ObjectOptions {
+                version_id: version_id.clone(),
+                versioned: true,
+                version_suspended,
                 ..Default::default()
-            },
-        );
+            };
+            let candidate = match api.get_object_info(bucket, &object.object_name, &opts).await {
+                Ok(info) => {
+                    let dsc = check_replicate_delete(bucket, object, &info, &opts, None).await;
+                    dsc.replicate_any()
+                        .then(|| lifecycle_version_delete_replication_state(dsc.to_string(), dsc.pending_status()))
+                }
+                Err(err) => {
+                    warn!(
+                        bucket,
+                        object = %object.object_name,
+                        version_id = ?version_id,
+                        error = ?err,
+                        "failed to get object info during lifecycle noncurrent version cleanup; skipping delete replication scheduling"
+                    );
+                    None
+                }
+            };
+            replication_candidates.push(candidate);
+        }
+
+        let (mut deleted_objs, errors) = api
+            .delete_objects(
+                bucket,
+                to_del.to_vec(),
+                ObjectOptions {
+                    version_suspended,
+                    ..Default::default()
+                },
+            )
+            .await;
+
+        for (i, deleted_obj) in deleted_objs.iter_mut().enumerate() {
+            if errors.get(i).and_then(|err| err.as_ref()).is_some() {
+                continue;
+            }
+            let Some(replication_state) = replication_candidates.get(i).and_then(|c| c.clone()) else {
+                continue;
+            };
+            deleted_obj.replication_state = Some(replication_state);
+            schedule_replication_delete(DeletedObjectReplicationInfo {
+                delete_object: deleted_obj.clone(),
+                bucket: bucket.to_string(),
+                event_type: REPLICATE_INCOMING_DELETE.to_string(),
+                ..Default::default()
+            })
+            .await;
+        }
+
+        for (i, err) in errors.iter().enumerate() {
+            if let Some(e) = err {
+                let obj_name = to_del.get(i).map(|o| o.object_name.as_str()).unwrap_or("<unknown>");
+                let vid = to_del
+                    .get(i)
+                    .and_then(|o| o.version_id)
+                    .map(|v| v.to_string())
+                    .unwrap_or_default();
+                warn!(bucket, object = obj_name, version_id = %vid, error = ?e, "failed to delete noncurrent version during lifecycle cleanup");
+            }
+        }
+        if remaining.is_empty() {
+            break;
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::lifecycle_version_delete_replication_state;
+
+    #[test]
+    fn lifecycle_version_delete_replication_state_tracks_pending_purge_targets() {
+        let state = lifecycle_version_delete_replication_state(
+            "arn:aws:s3:::target=true;false;arn:aws:s3:::target;".to_string(),
+            Some("arn:aws:s3:::target=PENDING;".to_string()),
+        );
+
+        assert_eq!(state.version_purge_status_internal.as_deref(), Some("arn:aws:s3:::target=PENDING;"));
+        assert!(state.purge_targets.contains_key("arn:aws:s3:::target"));
+        assert_eq!(state.replicate_decision_str, "arn:aws:s3:::target=true;false;arn:aws:s3:::target;");
     }
 }

crates/ecstore/src/client/signer_error.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.
+
+use std::error::Error as StdError;
+use std::fmt::{Display, Formatter};
+use std::io::{Error, ErrorKind};
+
+pub(crate) const SIGNER_HEADER_ERROR_MARKER: &str = "rustfs_signer_header_error";
+
+#[derive(Debug)]
+struct SignerHeaderError {
+    scope: String,
+    header_name: String,
+}
+
+impl SignerHeaderError {
+    fn new(scope: &str, header_name: &str) -> Self {
+        Self {
+            scope: scope.to_string(),
+            header_name: header_name.to_string(),
+        }
+    }
+}
+
+impl Display for SignerHeaderError {
+    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
+        write!(
+            f,
+            "{}: invalid UTF-8 header value for `{}` [{}]",
+            self.scope, self.header_name, SIGNER_HEADER_ERROR_MARKER
+        )
+    }
+}
+
+impl StdError for SignerHeaderError {}
+
+pub(crate) fn invalid_utf8_header_error(scope: &str, header_name: &str) -> Error {
+    Error::new(ErrorKind::InvalidInput, SignerHeaderError::new(scope, header_name))
+}
+
+pub(crate) fn signer_error_to_io_error(scope: &str, error: rustfs_signer::SignV4Error) -> Error {
+    match error {
+        rustfs_signer::SignV4Error::InvalidHeaderValue { name } => invalid_utf8_header_error(scope, &name),
+        other => Error::other(format!("{scope}: {other}")),
+    }
+}
+
+pub(crate) fn error_chain_contains_signer_header_marker(err: &(dyn StdError + 'static)) -> bool {
+    let mut current = Some(err);
+    while let Some(source) = current {
+        if source.downcast_ref::<SignerHeaderError>().is_some() {
+            return true;
+        }
+
+        if source.to_string().contains(SIGNER_HEADER_ERROR_MARKER) {
+            return true;
+        }
+
+        current = source.source();
+    }
+
+    false
+}
+
+#[cfg(test)]
+mod tests {
+    use super::{error_chain_contains_signer_header_marker, invalid_utf8_header_error, signer_error_to_io_error};
+
+    #[test]
+    fn invalid_utf8_header_error_is_detected_through_error_chain() {
+        let err = invalid_utf8_header_error("failed to sign request", "x-amz-meta-invalid");
+
+        assert!(error_chain_contains_signer_header_marker(&err));
+    }
+
+    #[test]
+    fn mapped_signer_header_error_is_detected_through_error_chain() {
+        let err = signer_error_to_io_error(
+            "failed to sign request",
+            rustfs_signer::SignV4Error::InvalidHeaderValue {
+                name: "x-amz-meta-invalid".to_string(),
+            },
+        );
+
+        assert!(error_chain_contains_signer_header_marker(&err));
+    }
+
+    #[test]
+    fn generic_io_errors_do_not_match_signer_header_marker() {
+        let err = std::io::Error::other("unrelated failure");
+
+        assert!(!error_chain_contains_signer_header_marker(&err));
+    }
+}

crates/ecstore/src/client/transition_api.rs

@@ -31,6 +31,7 @@ use crate::client::{
     },
     constants::{UNSIGNED_PAYLOAD, UNSIGNED_PAYLOAD_TRAILER},
     credentials::{CredContext, Credentials, SignatureType, Static},
+    signer_error,
 };
 use crate::{client::checksum::ChecksumMode, store_api::GetObjectReader};
 use futures::{Future, StreamExt};
@@ -85,6 +86,21 @@ const C_UNKNOWN: i32 = -1;
 const C_OFFLINE: i32 = 0;
 const C_ONLINE: i32 = 1;
 
+fn invalid_utf8_header_error(scope: &str, header_name: &str) -> std::io::Error {
+    signer_error::invalid_utf8_header_error(scope, header_name)
+}
+
+fn validate_header_values(headers: &HeaderMap, scope: &str) -> Result<(), std::io::Error> {
+    for (name, value) in headers {
+        value.to_str().map_err(|_| invalid_utf8_header_error(scope, name.as_str()))?;
+    }
+    Ok(())
+}
+
+fn signer_error_to_io_error(scope: &str, error: rustfs_signer::SignV4Error) -> std::io::Error {
+    signer_error::signer_error_to_io_error(scope, error)
+}
+
 //pub type ReaderImpl = Box<dyn Reader + Send + Sync + 'static>;
 pub enum ReaderImpl {
     Body(Bytes),
@@ -251,9 +267,10 @@ impl TransitionClient {
         };
 
         {
-            let mut md5_hasher = client.md5_hasher.lock().unwrap();
-            if md5_hasher.is_none() {
-                *md5_hasher = Some(HashAlgorithm::Md5);
+            if let Ok(mut md5_hasher) = client.md5_hasher.lock() {
+                if md5_hasher.is_none() {
+                    *md5_hasher = Some(HashAlgorithm::Md5);
+                }
             }
         }
         if client.sha256_hasher.is_none() {
@@ -275,25 +292,30 @@ impl TransitionClient {
     }
 
     fn trace_errors_only_off(&self) {
-        let mut trace_errors_only = self.trace_errors_only.lock().unwrap();
-        *trace_errors_only = false;
+        if let Ok(mut trace_errors_only) = self.trace_errors_only.lock() {
+            *trace_errors_only = false;
+        }
     }
 
     fn trace_off(&self) {
-        let mut is_trace_enabled = self.is_trace_enabled.lock().unwrap();
-        *is_trace_enabled = false;
-        let mut trace_errors_only = self.trace_errors_only.lock().unwrap();
-        *trace_errors_only = false;
+        if let Ok(mut is_trace_enabled) = self.is_trace_enabled.lock() {
+            *is_trace_enabled = false;
+        }
+        if let Ok(mut trace_errors_only) = self.trace_errors_only.lock() {
+            *trace_errors_only = false;
+        }
     }
 
     fn set_s3_transfer_accelerate(&self, accelerate_endpoint: &str) {
-        let mut endpoint = self.s3_accelerate_endpoint.lock().unwrap();
-        *endpoint = accelerate_endpoint.to_string();
+        if let Ok(mut endpoint) = self.s3_accelerate_endpoint.lock() {
+            *endpoint = accelerate_endpoint.to_string();
+        }
     }
 
     fn set_s3_enable_dual_stack(&self, enabled: bool) {
-        let mut dual_stack = self.s3_dual_stack_enabled.lock().unwrap();
-        *dual_stack = enabled;
+        if let Ok(mut dual_stack) = self.s3_dual_stack_enabled.lock() {
+            *dual_stack = enabled;
+        }
     }
 
     pub fn hash_materials(
@@ -352,7 +374,6 @@ impl TransitionClient {
         let resp;
         let http_client = self.http_client.clone();
         {
-            //let mut http_client = http_client.lock().unwrap();
             req_method = req.method().clone();
             req_uri = req.uri().clone();
             req_headers = req.headers().clone();
@@ -368,7 +389,10 @@ impl TransitionClient {
             return Err(std::io::Error::other(err));
         }
 
-        let resp = resp.unwrap();
+        let resp = match resp {
+            Ok(r) => r,
+            Err(_) => return Err(std::io::Error::other("Unexpected error in response")),
+        };
         debug!("http_resp: {:?}", resp);
 
         //let b = resp.body_mut().store_all_unlimited().await.unwrap().to_vec();
@@ -455,11 +479,13 @@ impl TransitionClient {
                             return Err(std::io::Error::other(err_response));
                         }
                         if metadata.bucket_name != "" {
-                            let mut bucket_loc_cache = self.bucket_loc_cache.lock().unwrap();
-                            let location = bucket_loc_cache.get(&metadata.bucket_name);
-                            if location.is_some() && location.unwrap() != err_response.region {
-                                bucket_loc_cache.set(&metadata.bucket_name, &err_response.region);
-                                //continue;
+                            if let Ok(mut bucket_loc_cache) = self.bucket_loc_cache.lock() {
+                                if let Some(location) = bucket_loc_cache.get(&metadata.bucket_name) {
+                                    if location != err_response.region {
+                                        bucket_loc_cache.set(&metadata.bucket_name, &err_response.region);
+                                        //continue;
+                                    }
+                                }
                             }
                         } else if err_response.region != metadata.bucket_location {
                             metadata.bucket_location = err_response.region.clone();
@@ -518,8 +544,11 @@ impl TransitionClient {
 
         let value;
         {
-            let mut creds_provider = self.creds_provider.lock().unwrap();
-            value = creds_provider.get_with_context(Some(self.cred_context()))?;
+            if let Ok(mut creds_provider) = self.creds_provider.lock() {
+                value = creds_provider.get_with_context(Some(self.cred_context()))?;
+            } else {
+                return Err(std::io::Error::other("Failed to acquire credentials provider lock"));
+            }
         }
 
         let mut signer_type = value.signer_type.clone();
@@ -547,15 +576,18 @@ impl TransitionClient {
                         "extra signed headers for presign with signature v2 is not supported.",
                     )));
                 }
-                let headers = req.headers_mut();
-                for (k, v) in metadata.extra_pre_sign_header.as_ref().unwrap() {
-                    headers.insert(k, v.clone());
+                if let Some(extra_headers) = metadata.extra_pre_sign_header.as_ref() {
+                    validate_header_values(extra_headers, "presign extra header")?;
+                    let headers = req.headers_mut();
+                    for (k, v) in extra_headers {
+                        headers.insert(k, v.clone());
+                    }
                 }
             }
             if signer_type == SignatureType::SignatureV2 {
                 req = rustfs_signer::pre_sign_v2(req, &access_key_id, &secret_access_key, metadata.expires, is_virtual_host);
             } else if signer_type == SignatureType::SignatureV4 {
-                req = rustfs_signer::pre_sign_v4(
+                req = rustfs_signer::try_pre_sign_v4(
                     req,
                     &access_key_id,
                     &secret_access_key,
@@ -563,25 +595,31 @@ impl TransitionClient {
                     &location,
                     metadata.expires,
                     OffsetDateTime::now_utc(),
-                );
+                )
+                .map_err(|err| signer_error_to_io_error("failed to presign v4 request", err))?;
             }
             return Ok(req);
         }
 
         self.set_user_agent(&mut req);
+        validate_header_values(&metadata.custom_header, "request custom header")?;
 
         for (k, v) in metadata.custom_header.clone() {
-            req.headers_mut().insert(k.expect("err"), v);
+            if let Some(key) = k {
+                req.headers_mut().insert(key, v);
+            }
         }
 
         //req.content_length = metadata.content_length;
         if metadata.content_length <= -1 {
-            let chunked_value = HeaderValue::from_str(&vec!["chunked"].join(",")).expect("err");
-            req.headers_mut().insert(http::header::TRANSFER_ENCODING, chunked_value);
+            req.headers_mut()
+                .insert(http::header::TRANSFER_ENCODING, HeaderValue::from_static("chunked"));
         }
 
-        if metadata.content_md5_base64.len() > 0 {
-            let md5_value = HeaderValue::from_str(&metadata.content_md5_base64).expect("err");
+        if !metadata.content_md5_base64.is_empty() {
+            let md5_value = HeaderValue::from_str(&metadata.content_md5_base64).map_err(|err| {
+                std::io::Error::new(std::io::ErrorKind::InvalidInput, format!("invalid Content-Md5 header value: {err}"))
+            })?;
             req.headers_mut().insert("Content-Md5", md5_value);
         }
 
@@ -607,17 +645,23 @@ impl TransitionClient {
             } else if metadata.trailer.len() > 0 {
                 sha_header = UNSIGNED_PAYLOAD_TRAILER.to_string();
             }
-            req.headers_mut()
-                .insert("X-Amz-Content-Sha256".parse::<HeaderName>().unwrap(), sha_header.parse().expect("err"));
+            let header_name = "X-Amz-Content-Sha256"
+                .parse::<HeaderName>()
+                .map_err(|e| std::io::Error::new(std::io::ErrorKind::InvalidInput, e))?;
+            let header_value = sha_header
+                .parse()
+                .map_err(|e| std::io::Error::new(std::io::ErrorKind::InvalidInput, e))?;
+            req.headers_mut().insert(header_name, header_value);
 
-            req = rustfs_signer::sign_v4_trailer(
+            req = rustfs_signer::try_sign_v4_trailer(
                 req,
                 &access_key_id,
                 &secret_access_key,
                 &session_token,
                 &location,
                 metadata.trailer.clone(),
-            );
+            )
+            .map_err(|err| signer_error_to_io_error("failed to sign v4 request", err))?;
         }
 
         if metadata.content_length > 0 {
@@ -636,7 +680,7 @@ impl TransitionClient {
 
     pub fn set_user_agent(&self, req: &mut Request<s3s::Body>) {
         let headers = req.headers_mut();
-        headers.insert("User-Agent", C_USER_AGENT.parse().expect("err"));
+        headers.insert("User-Agent", HeaderValue::from_static(C_USER_AGENT));
     }
 
     fn make_target_url(
@@ -648,7 +692,10 @@ impl TransitionClient {
         query_values: &HashMap<String, String>,
     ) -> Result<Url, std::io::Error> {
         let scheme = self.endpoint_url.scheme();
-        let host = self.endpoint_url.host().unwrap();
+        let host = self
+            .endpoint_url
+            .host()
+            .ok_or_else(|| std::io::Error::other("Endpoint URL has no host"))?;
         let default_port = if scheme == "https" { 443 } else { 80 };
         let port = self.endpoint_url.port().unwrap_or(default_port);
 
@@ -1155,9 +1202,10 @@ pub fn to_object_info(bucket_name: &str, object_name: &str, h: &HeaderMap) -> Re
         for (name, value) in h.iter() {
             let header_name = name.as_str().to_lowercase();
             if header_name.starts_with("x-amz-meta-") {
-                let key = header_name.strip_prefix("x-amz-meta-").unwrap().to_string();
-                if let Ok(value_str) = value.to_str() {
-                    meta.insert(key, value_str.to_string());
+                if let Some(key) = header_name.strip_prefix("x-amz-meta-") {
+                    if let Ok(value_str) = value.to_str() {
+                        meta.insert(key.to_string(), value_str.to_string());
+                    }
                 }
             }
         }
@@ -1326,7 +1374,10 @@ pub struct CreateBucketConfiguration {
 
 #[cfg(test)]
 mod tests {
-    use super::{build_tls_config, load_root_store_from_tls_path, with_rustls_init_guard};
+    use super::{
+        build_tls_config, load_root_store_from_tls_path, signer_error_to_io_error, validate_header_values, with_rustls_init_guard,
+    };
+    use http::{HeaderMap, HeaderValue};
 
     #[test]
     fn rustls_guard_converts_panics_to_io_errors() {
@@ -1376,4 +1427,29 @@ mod tests {
         });
         assert!(outcome.is_ok(), "provider install guard must not panic when a provider is already set");
     }
+
+    #[test]
+    fn validate_header_values_returns_header_name_for_non_utf8_values() {
+        let mut headers = HeaderMap::new();
+        headers.insert(
+            "x-amz-meta-invalid",
+            HeaderValue::from_bytes(&[0xFF]).expect("invalid utf8 bytes should be accepted by HeaderValue"),
+        );
+
+        let err =
+            validate_header_values(&headers, "request custom header").expect_err("invalid header value should fail validation");
+        assert!(err.to_string().contains("x-amz-meta-invalid"));
+    }
+
+    #[test]
+    fn signer_error_mapping_preserves_header_name() {
+        let err = signer_error_to_io_error(
+            "failed to sign v4 request",
+            rustfs_signer::SignV4Error::InvalidHeaderValue {
+                name: "x-amz-meta-invalid".to_string(),
+            },
+        );
+        assert_eq!(err.kind(), std::io::ErrorKind::InvalidInput);
+        assert!(err.to_string().contains("x-amz-meta-invalid"));
+    }
 }