增加 infra 指标释义说明 (7405bfe7) · 提交 · aiops-live-benchmark / AIOpsChallenge2025-Sample

README.md

+78 −36

原始行号	差异行号	差异行
		@@ -163,42 +163,84 @@ APM 指标主要反映业务服务在一定时间窗口内的请求与响应情

		性能指标反映系统的底层资源使用情况，包括 Pod、Node、TiDB 组件在内的多种对象。下表列出了常见指标编码及其含义（以 Parquet 文件中的 kpi_key 为主）：

		\| 对象类型 \| 指标编码 \| 指标名称 \|
		\| -------- \| ------------------------------- \| ----------------------------- \|
		\| pod \| pod_cpu_usage \| CPU 使用率 \|
		\| pod \| pod_processes \| 进程数 \|
		\| pod \| pod_memory_working_set_bytes \| 内存使用大小 \|
		\| pod \| pod_fs_writes_bytes \| 写入字节的累积计数 \|
		\| pod \| pod_fs_reads_bytes \| 累计读取字节数 \|
		\| pod \| pod_network_receive_bytes \| 接收字节的累积计数 \|
		\| pod \| pod_network_transmit_bytes \| 传输字节的累积计数 \|
		\| pod \| pod_network_receive_packets \| 接收数据包的累积计数 \|
		\| pod \| pod_network_transmit_packets \| 传输数据包的累积计数 \|
		\| node \| node_cpu_usage_rate \| CPU 使用率 \|
		\| node \| node_memory_usage_rate \| 内存使用率 \|
		\| node \| node_filesystem_usage_rate \| 磁盘使用率 \|
		\| node \| node_memory_MemAvailable_bytes \| 空闲内存大小 \|
		\| node \| node_memory_MemTotal_bytes \| 内存总大小 \|
		\| node \| node_filesystem_size_bytes \| 存储设备总大小 \|
		\| node \| node_filesystem_free_bytes \| 存储设备空闲大小 \|
		\| node \| node_disk_read_bytes_total \| 成功读取的字节数 \|
		\| node \| node_disk_read_time_seconds_total \| 磁盘分区读取花费的秒数 \|
		\| node \| node_disk_written_bytes_total \| 成功写入的字节数 \|
		\| node \| node_disk_write_time_seconds_total \| 磁盘分区写操作花费的秒数 \|
		\| node \| node_network_receive_bytes_total \| {{device}} 接口接收速率 \|
		\| node \| node_network_receive_packets_total \| {{device}} 接口每秒接收的数据包总数 \|
		\| node \| node_network_transmit_bytes_total \| {{device}} 接口发送速率 \|
		\| node \| node_network_transmit_packets_total \| {{device}} 接口每秒发送的数据包总数 \|
		\| node \| node_sockstat_TCP_inuse \| TCP_inuse – 正在使用（正在侦听）的 TCP 套接字数量 \|
		\| tidb \| connection_count \| 连接数 \|
		\| tidb \| failed_query_ops \| 失败请求数 \|
		\| tidb \| duration_99th \| 99 分位请求延迟 \|
		\| tidb \| duration_95th \| 95 分位请求延迟 \|
		\| tidb \| duration_avg \| 平均请求延迟 \|
		\| tidb \| qps \| 请求数量 \|
		\| tidb \| slow_query \| 慢查询 \|
		\| tidb \| block_cache_size \| Block Cache 大小 \|

		\| object_type \| kpi_key \| kpi_name \| promql \|
		\|-------------\|----------------------------------\|------------------------\|-----------------------------------------------------------------------------------------------------------------------------\|
		\| pod \| pod_cpu_usage \| CPU 使用率 \| `sum(irate(container_cpu_usage_seconds_total{container=~"server\\|redis"}[1m])) by (namespace,instance,pod)` \|
		\| pod \| pod_processes \| 进程数 \| `sum(container_processes{container=~"server\\|redis"}) by (namespace,instance,pod)` \|
		\| pod \| pod_memory_working_set_bytes \| 内存使用大小 \| `sum(rate(container_memory_working_set_bytes{container=~"server\\|redis"}[1m])) by (namespace,instance,pod)` \|
		\| pod \| pod_fs_writes_bytes \| 写入字节的累积计数 \| `sum(irate(container_fs_writes_bytes_total{container=~"server\\|redis"}[1m])) by (namespace,instance,pod,device)` \|
		\| pod \| pod_fs_reads_bytes \| 累计读取字节数 \| `sum(irate(container_fs_reads_bytes_total{container=~"server\\|redis"}[1m])) by (namespace,instance,pod,device)` \|
		\| pod \| pod_network_receive_bytes \| 接收字节的累积计数 \| `sum(irate(container_network_receive_bytes_total{namespace="hipstershop",interface="eth0"}[1m])) by (namespace,instance,pod)` \|
		\| pod \| pod_network_transmit_bytes \| 传输字节的累积计数 \| `sum(irate(container_network_transmit_bytes_total{namespace="hipstershop",interface="eth0"}[1m])) by (namespace,instance,pod)`\|
		\| pod \| pod_network_receive_packets \| 接收数据包的累积计数 \| `sum(irate(container_network_receive_packets_total{namespace="hipstershop",interface="eth0"}[1m])) by (namespace,instance,pod)`\|
		\| pod \| pod_network_transmit_packets \| 传输数据包的累积计数 \| `sum(irate(container_network_transmit_packets_total{namespace="hipstershop",interface="eth0"}[1m])) by (namespace,instance,pod)`\|
		\| node \| node_cpu_usage_rate \| CPU使用率 \| `100 - (avg(irate(node_cpu_seconds_total{job="kubernetes-nodes",mode="idle"}[1m])) by (kubernetes_node,instance) * 100)` \|
		\| node \| node_memory_usage_rate \| 内存使用率 \| `100 - (node_memory_MemFree_bytes{job="kubernetes-nodes"} + node_memory_Cached_bytes{job="kubernetes-nodes"} + node_memory_Buffers_bytes{job="kubernetes-nodes"}) / node_memory_MemTotal_bytes{job="kubernetes-nodes"} * 100` \|
		\| node \| node_filesystem_usage_rate \| 磁盘使用率 \| `100 - (node_filesystem_free_bytes{job="kubernetes-nodes",fstype=~"ext4\|xfs"} / node_filesystem_size_bytes{job="kubernetes-nodes",fstype=~"ext4\|xfs"} * 100)` \|
		\| node \| node_memory_MemAvailable_bytes \| 空闲内存大小 \| `node_memory_MemAvailable_bytes{job="kubernetes-nodes"}` \|
		\| node \| node_memory_MemTotal_bytes \| 内存总大小 \| `node_memory_MemTotal_bytes{job="kubernetes-nodes"}` \|
		\| node \| node_filesystem_size_bytes \| 空闲磁盘大小 \| `node_filesystem_size_bytes{job="kubernetes-nodes",mountpoint=~"/rootfs\|/"}` \|
		\| node \| node_filesystem_free_bytes \| 磁盘总大小 \| `node_filesystem_free_bytes{job="kubernetes-nodes",mountpoint=~"/rootfs\|/"}`
		\| node \| node_disk_read_bytes_total \| 成功读取的字节数 \| `sum(irate(node_disk_read_bytes_total{job="kubernetes-nodes",device="vda"}[1m])) by (kubernetes_node,instance,device)` \|
		\| node \| node_disk_read_time_seconds_total \| Read time ms 每个磁盘分区读花费的毫秒数 \| `sum(irate(node_disk_read_time_seconds_total{job="kubernetes-nodes",device="vda"}[1m])) by (kubernetes_node,instance,device)` \|
		\| node \| node_disk_written_bytes_total \| 成功写入的字节数 \| `sum(irate(node_disk_written_bytes_total{job="kubernetes-nodes",device="vda"}[1m])) by (kubernetes_node,instance,device)` \|
		\| node \| node_disk_write_time_seconds_total \| Write time ms 每个磁盘分区写操作花费的毫秒数 \| `sum(irate(node_disk_write_time_seconds_total{job="kubernetes-nodes",device="vda"}[1m])) by (kubernetes_node,instance,device)` \|
		\| node \| node_network_receive_bytes_total \| `{{device}}` – Receive 各个网络接口接收速率 \| `sum(irate(node_network_receive_bytes_total{job="kubernetes-nodes"}[1m])) by (kubernetes_node,instance)` \|
		\| node \| node_network_receive_packets_total \| `{{device}}` – Receive 各个接口每秒接收的数据包总数 \| `sum(irate(node_network_receive_packets_total{job="kubernetes-nodes"}[1m])) by (kubernetes_node,instance)` \|
		\| node \| node_network_transmit_bytes_total \| `{{device}}` – Transmit 各个网络接口发送速率 \| `sum(irate(node_network_transmit_bytes_total{job="kubernetes-nodes"}[1m])) by (kubernetes_node,instance)` \|
		\| node \| node_network_transmit_packets_total \| `{{device}}` – Transmit 各个接口每秒发送的数据包总数 \| `sum(irate(node_network_transmit_packets_total{job="kubernetes-nodes"}[1m])) by (kubernetes_node,instance)` \|
		\| node \| node_sockstat_TCP_inuse \| TCP_inuse – 正在使用（正在侦听）的 TCP 套接字数量 \| `sum(irate(node_sockstat_TCP_inuse{job="kubernetes-nodes"}[1m])) by (kubernetes_node,instance)` \|
		\| tidb \| connection_count \| 连接数 \| `sum(tidb_server_connections) by (namespace,instance)` \|
		\| tidb \| failed_query_ops \| 失败请求数 \| `sum(increase(tidb_server_execute_error_total[1m])) by (namespace,type,instance)` \|
		\| tidb \| duration_99th \| 99 分位请求延迟 \| `histogram_quantile(0.99, sum(rate(tidb_server_handle_query_duration_seconds_bucket{sql_type!="internal"}[1m])) by (namespace,instance,le))` \|
		\| tidb \| duration_95th \| 95 分位请求延迟 \| `histogram_quantile(0.95, sum(rate(tidb_server_handle_query_duration_seconds_bucket{sql_type!="internal"}[1m])) by (namespace,instance,le))` \|
		\| tidb \| duration_avg \| 平均请求延迟 \| `sum(rate(tidb_server_handle_query_duration_seconds_sum{sql_type!="internal"}[2m])) by (namespace,instance) / sum(rate(tidb_server_handle_query_duration_seconds_count{sql_type!="internal"}[2m])) by (namespace,instance)` \|
		\| tidb \| qps \| 请求数量 \| `sum(rate(tidb_executor_statement_total[1m])) by (namespace,instance,type)` \|
		\| tidb \| slow_query \| 慢查询 \| `histogram_quantile(0.90, sum(rate(tidb_server_slow_query_process_duration_seconds_bucket{sql_type="general"}[1m])) by (namespace,instance,le,sql_type))` \|
		\| tidb \| block_cache_size \| block_cache_size \| `avg(tikv_engine_block_cache_size_bytes{db="kv"}) by (namespace,instance,cf)` \|
		\| tidb \| uptime \| 服务存活时长 \| `max(time() - process_start_time_seconds{cluster="tidb",component="tidb"}) by (namespace,instance)` \|
		\| tidb \| cpu_usage \| CPU 使用率 \| `sum(rate(process_cpu_seconds_total{cluster="tidb",component="tidb"}[1m])) by (namespace,instance)` \|
		\| tidb \| memory_usage \| 内存使用量 \| `avg(process_resident_memory_bytes{cluster="tidb",component="tidb"}) by (namespace,instance)` \|
		\| tidb \| statement_avg_queries \| 多语句平均查询数 \| `sum(rate(tidb_server_multi_query_num_sum[1m])) / sum(rate(tidb_server_multi_query_num_count[1m]))` \|
		\| tidb \| transaction_retry \| 事务重试次数 \| `sum(rate(tidb_session_retry_num_sum[1m])) by (namespace,instance)` \|
		\| tidb \| ddl_job_count \| DDL 作业数 \| `sum(tidb_ddl_running_job_count) by (namespace,instance,state)` \|
		\| tidb \| top_sql_cpu \| TopSQL CPU 消耗 \| `sum(rate(tidb_topsql_report_duration_seconds_sum[1m])) by (namespace,instance,sql_type)` \|
		\| tidb \| server_is_up \| 服务存活节点数 \| `count(up{cluster="tidb",component="tidb"} == 1) by (namespace)` \|
		\| tikv \| store_size \| 已用存储容量 \| `sum(tikv_store_size_bytes{type="used"}) by (namespace,instance)` \|
		\| tikv \| available_size \| 可用存储容量 \| `sum(tikv_store_size_bytes{type="available"}) by (namespace,instance)` \|
		\| tikv \| capacity_size \| 总存储容量 \| `sum(tikv_store_size_bytes{type="capacity"}) by (namespace,instance)` \|
		\| tikv \| cpu_usage \| CPU 使用率 \| `sum(rate(process_cpu_seconds_total{cluster="tidb",component="tikv"}[1m])) by (namespace,instance)` \|
		\| tikv \| memory_usage \| 内存使用量 \| `avg(process_resident_memory_bytes{cluster="tidb",component="tikv"}) by (namespace,instance)` \|
		\| tikv \| io_util \| IO 利用率 \| `rate(node_disk_io_time_seconds_total[1m])` \|
		\| tikv \| write_wal_mbps \| 写入速率 (Mbps) \| `sum(rate(tikv_engine_flow_bytes{type="wal_file_bytes"}[1m])) by (namespace,instance)` \|
		\| tikv \| read_mbps \| 读取速率 (Mbps) \| `sum(rate(tikv_engine_flow_bytes{type=~"bytes_read\|iter_bytes_read"}[1m])) by (namespace,instance)` \|
		\| tikv \| qps \| QPS \| `sum(rate(tikv_storage_command_total[1m])) by (namespace,instance,type)` \|
		\| tikv \| grpc_qps \| 各类 gRPC 请求 QPS \| `sum(rate(tikv_grpc_msg_duration_seconds_count{type=~"coprocessor\|kv_get\|kv_batch_get\|kv_prewrite\|kv_commit"}[1m])) by (namespace,instance,type)` \|
		\| tikv \| threadpool_readpool_cpu \| StorageReadPool 线程池 CPU \| `sum(rate(tikv_thread_cpu_seconds_total{name=~"unified_read_pool_.*"}[1m])) by (namespace,instance,name)` \|
		\| tikv \| raft_propose_wait \| RaftPropose 等待延迟 P99 \| `histogram_quantile(0.99, sum(rate(tikv_raftstore_request_wait_time_duration_secs_bucket[1m])) by (le,namespace,instance))` \|
		\| tikv \| raft_apply_wait \| RaftApply 等待延迟 P99 \| `histogram_quantile(0.99, sum(rate(tikv_raftstore_apply_wait_time_duration_secs_bucket[1m])) by (le,namespace,instance))` \|
		\| tikv \| region_pending \| PendingRegion 数量 \| `sum(tikv_raftstore_read_index_pending_duration_count) by (namespace,instance)` \|
		\| tikv \| rocksdb_write_stall \| RocksDB 写阻塞次数 \| `sum(tikv_engine_write_stall) by (namespace,instance)` \|
		\| tikv \| snapshot_apply_count \| SnapshotApply 次数 \| `sum(rate(tikv_raftstore_apply_duration_secs_count[1m])) by (namespace,instance)` \|
		\| tikv \| server_is_up \| 服务存活节点数 \| `count(up{cluster="tidb",component="tikv"} == 1) by (namespace)` \|
		\| tikv \| cpu_usage \| CPU 使用率 \| `sum(rate(process_cpu_seconds_total{cluster="tidb",component="tikv"}[1m])) by (namespace,instance)` \|
		\| pd \| storage_capacity \| 集群总容量 \| `sum(pd_cluster_status{type="storage_capacity"}) by (namespace)` \|
		\| pd \| storage_size \| 已用容量 \| `sum(pd_cluster_status{type="storage_size"}) by (namespace)` \|
		\| pd \| storage_used_ratio \| 已用容量比 \| `sum(pd_cluster_status{type="storage_size"}) / sum(pd_cluster_status{type="storage_capacity"})` \|
		\| pd \| store_up_count \| 健康 Store 数量 \| `sum(pd_cluster_status{type="store_up_count"}) by (namespace)` \|
		\| pd \| store_down_count \| Down Store 数量 \| `sum(pd_cluster_status{type="store_down_count"}) by (namespace)` \|
		\| pd \| store_unhealth_count \| Unhealth Store 数量 \| `sum(pd_cluster_status{type="store_unhealth_count"}) by (namespace)` \|
		\| pd \| store_low_space_count \| 低空间 Store 数量 \| `sum(pd_cluster_status{type="store_low_space_count"}) by (namespace)` \|
		\| pd \| store_slow_count \| 慢 Store 数量 \| `sum(pd_cluster_status{type="store_slow_count"}) by (namespace)` \|
		\| pd \| abnormal_region_count \| 异常 Region 数量 \| `sum(pd_regions_status{type=~"pending-peer-region-count\\|down-peer-region-count\\|offline-peer-region-count\\|miss-peer-region-count\\|learner-peer-region-count"}) by (namespace,type)` \|
		\| pd \| region_health \| Region 健康状态 \| `max(pd_regions_status) by (namespace,type)` \|
		\| pd \| leader_primary \| PDLeader-Primary 节点 \| `service_member_role{cluster="tidb",component="pd",service="PD"}` \|
		\| pd \| region_count \| Region 总数 \| `sum(pd_cluster_status{type="region_count"}) by (namespace)` \|
		\| pd \| leader_count \| Leader 总数 \| `sum(pd_cluster_status{type="leader_count"}) by (namespace)` \|
		\| pd \| learner_count \| Learner 总数 \| `sum(pd_cluster_status{type="learner_count"}) by (namespace)` \|
		\| pd \| witness_count \| Witness 总数 \| `sum(pd_cluster_status{type="witness_count"}) by (namespace)` \|
		\| pd \| cpu_usage \| CPU 使用率 \| `sum(rate(process_cpu_seconds_total{cluster="tidb",component="pd"}[1m])) by (namespace,instance)` \|
		\| pd \| memory_usage \| 内存使用量 \| `avg(process_resident_memory_bytes{cluster="tidb",component="pd"}) by (namespace,instance)` \|

		字段释义（infra 样例数据文件）

原始行号	差异行号	差异行
		@@ -163,42 +163,84 @@ APM 指标主要反映业务服务在一定时间窗口内的请求与响应情

		性能指标反映系统的底层资源使用情况，包括 Pod、Node、TiDB 组件在内的多种对象。下表列出了常见指标编码及其含义（以 Parquet 文件中的 kpi_key 为主）：

		\| 对象类型 \| 指标编码 \| 指标名称 \|
		\| -------- \| ------------------------------- \| ----------------------------- \|
		\| pod \| pod_cpu_usage \| CPU 使用率 \|
		\| pod \| pod_processes \| 进程数 \|
		\| pod \| pod_memory_working_set_bytes \| 内存使用大小 \|
		\| pod \| pod_fs_writes_bytes \| 写入字节的累积计数 \|
		\| pod \| pod_fs_reads_bytes \| 累计读取字节数 \|
		\| pod \| pod_network_receive_bytes \| 接收字节的累积计数 \|
		\| pod \| pod_network_transmit_bytes \| 传输字节的累积计数 \|
		\| pod \| pod_network_receive_packets \| 接收数据包的累积计数 \|
		\| pod \| pod_network_transmit_packets \| 传输数据包的累积计数 \|
		\| node \| node_cpu_usage_rate \| CPU 使用率 \|
		\| node \| node_memory_usage_rate \| 内存使用率 \|
		\| node \| node_filesystem_usage_rate \| 磁盘使用率 \|
		\| node \| node_memory_MemAvailable_bytes \| 空闲内存大小 \|
		\| node \| node_memory_MemTotal_bytes \| 内存总大小 \|
		\| node \| node_filesystem_size_bytes \| 存储设备总大小 \|
		\| node \| node_filesystem_free_bytes \| 存储设备空闲大小 \|
		\| node \| node_disk_read_bytes_total \| 成功读取的字节数 \|
		\| node \| node_disk_read_time_seconds_total \| 磁盘分区读取花费的秒数 \|
		\| node \| node_disk_written_bytes_total \| 成功写入的字节数 \|
		\| node \| node_disk_write_time_seconds_total \| 磁盘分区写操作花费的秒数 \|
		\| node \| node_network_receive_bytes_total \| {{device}} 接口接收速率 \|
		\| node \| node_network_receive_packets_total \| {{device}} 接口每秒接收的数据包总数 \|
		\| node \| node_network_transmit_bytes_total \| {{device}} 接口发送速率 \|
		\| node \| node_network_transmit_packets_total \| {{device}} 接口每秒发送的数据包总数 \|
		\| node \| node_sockstat_TCP_inuse \| TCP_inuse – 正在使用（正在侦听）的 TCP 套接字数量 \|
		\| tidb \| connection_count \| 连接数 \|
		\| tidb \| failed_query_ops \| 失败请求数 \|
		\| tidb \| duration_99th \| 99 分位请求延迟 \|
		\| tidb \| duration_95th \| 95 分位请求延迟 \|
		\| tidb \| duration_avg \| 平均请求延迟 \|
		\| tidb \| qps \| 请求数量 \|
		\| tidb \| slow_query \| 慢查询 \|
		\| tidb \| block_cache_size \| Block Cache 大小 \|

		\| object_type \| kpi_key \| kpi_name \| promql \|
		\|-------------\|----------------------------------\|------------------------\|-----------------------------------------------------------------------------------------------------------------------------\|
		\| pod \| pod_cpu_usage \| CPU 使用率 \| `sum(irate(container_cpu_usage_seconds_total{container=~"server\\|redis"}[1m])) by (namespace,instance,pod)` \|
		\| pod \| pod_processes \| 进程数 \| `sum(container_processes{container=~"server\\|redis"}) by (namespace,instance,pod)` \|
		\| pod \| pod_memory_working_set_bytes \| 内存使用大小 \| `sum(rate(container_memory_working_set_bytes{container=~"server\\|redis"}[1m])) by (namespace,instance,pod)` \|
		\| pod \| pod_fs_writes_bytes \| 写入字节的累积计数 \| `sum(irate(container_fs_writes_bytes_total{container=~"server\\|redis"}[1m])) by (namespace,instance,pod,device)` \|
		\| pod \| pod_fs_reads_bytes \| 累计读取字节数 \| `sum(irate(container_fs_reads_bytes_total{container=~"server\\|redis"}[1m])) by (namespace,instance,pod,device)` \|
		\| pod \| pod_network_receive_bytes \| 接收字节的累积计数 \| `sum(irate(container_network_receive_bytes_total{namespace="hipstershop",interface="eth0"}[1m])) by (namespace,instance,pod)` \|
		\| pod \| pod_network_transmit_bytes \| 传输字节的累积计数 \| `sum(irate(container_network_transmit_bytes_total{namespace="hipstershop",interface="eth0"}[1m])) by (namespace,instance,pod)`\|
		\| pod \| pod_network_receive_packets \| 接收数据包的累积计数 \| `sum(irate(container_network_receive_packets_total{namespace="hipstershop",interface="eth0"}[1m])) by (namespace,instance,pod)`\|
		\| pod \| pod_network_transmit_packets \| 传输数据包的累积计数 \| `sum(irate(container_network_transmit_packets_total{namespace="hipstershop",interface="eth0"}[1m])) by (namespace,instance,pod)`\|
		\| node \| node_cpu_usage_rate \| CPU使用率 \| `100 - (avg(irate(node_cpu_seconds_total{job="kubernetes-nodes",mode="idle"}[1m])) by (kubernetes_node,instance) * 100)` \|
		\| node \| node_memory_usage_rate \| 内存使用率 \| `100 - (node_memory_MemFree_bytes{job="kubernetes-nodes"} + node_memory_Cached_bytes{job="kubernetes-nodes"} + node_memory_Buffers_bytes{job="kubernetes-nodes"}) / node_memory_MemTotal_bytes{job="kubernetes-nodes"} * 100` \|
		\| node \| node_filesystem_usage_rate \| 磁盘使用率 \| `100 - (node_filesystem_free_bytes{job="kubernetes-nodes",fstype=~"ext4\|xfs"} / node_filesystem_size_bytes{job="kubernetes-nodes",fstype=~"ext4\|xfs"} * 100)` \|
		\| node \| node_memory_MemAvailable_bytes \| 空闲内存大小 \| `node_memory_MemAvailable_bytes{job="kubernetes-nodes"}` \|
		\| node \| node_memory_MemTotal_bytes \| 内存总大小 \| `node_memory_MemTotal_bytes{job="kubernetes-nodes"}` \|
		\| node \| node_filesystem_size_bytes \| 空闲磁盘大小 \| `node_filesystem_size_bytes{job="kubernetes-nodes",mountpoint=~"/rootfs\|/"}` \|
		\| node \| node_filesystem_free_bytes \| 磁盘总大小 \| `node_filesystem_free_bytes{job="kubernetes-nodes",mountpoint=~"/rootfs\|/"}`
		\| node \| node_disk_read_bytes_total \| 成功读取的字节数 \| `sum(irate(node_disk_read_bytes_total{job="kubernetes-nodes",device="vda"}[1m])) by (kubernetes_node,instance,device)` \|
		\| node \| node_disk_read_time_seconds_total \| Read time ms 每个磁盘分区读花费的毫秒数 \| `sum(irate(node_disk_read_time_seconds_total{job="kubernetes-nodes",device="vda"}[1m])) by (kubernetes_node,instance,device)` \|
		\| node \| node_disk_written_bytes_total \| 成功写入的字节数 \| `sum(irate(node_disk_written_bytes_total{job="kubernetes-nodes",device="vda"}[1m])) by (kubernetes_node,instance,device)` \|
		\| node \| node_disk_write_time_seconds_total \| Write time ms 每个磁盘分区写操作花费的毫秒数 \| `sum(irate(node_disk_write_time_seconds_total{job="kubernetes-nodes",device="vda"}[1m])) by (kubernetes_node,instance,device)` \|
		\| node \| node_network_receive_bytes_total \| `{{device}}` – Receive 各个网络接口接收速率 \| `sum(irate(node_network_receive_bytes_total{job="kubernetes-nodes"}[1m])) by (kubernetes_node,instance)` \|
		\| node \| node_network_receive_packets_total \| `{{device}}` – Receive 各个接口每秒接收的数据包总数 \| `sum(irate(node_network_receive_packets_total{job="kubernetes-nodes"}[1m])) by (kubernetes_node,instance)` \|
		\| node \| node_network_transmit_bytes_total \| `{{device}}` – Transmit 各个网络接口发送速率 \| `sum(irate(node_network_transmit_bytes_total{job="kubernetes-nodes"}[1m])) by (kubernetes_node,instance)` \|
		\| node \| node_network_transmit_packets_total \| `{{device}}` – Transmit 各个接口每秒发送的数据包总数 \| `sum(irate(node_network_transmit_packets_total{job="kubernetes-nodes"}[1m])) by (kubernetes_node,instance)` \|
		\| node \| node_sockstat_TCP_inuse \| TCP_inuse – 正在使用（正在侦听）的 TCP 套接字数量 \| `sum(irate(node_sockstat_TCP_inuse{job="kubernetes-nodes"}[1m])) by (kubernetes_node,instance)` \|
		\| tidb \| connection_count \| 连接数 \| `sum(tidb_server_connections) by (namespace,instance)` \|
		\| tidb \| failed_query_ops \| 失败请求数 \| `sum(increase(tidb_server_execute_error_total[1m])) by (namespace,type,instance)` \|
		\| tidb \| duration_99th \| 99 分位请求延迟 \| `histogram_quantile(0.99, sum(rate(tidb_server_handle_query_duration_seconds_bucket{sql_type!="internal"}[1m])) by (namespace,instance,le))` \|
		\| tidb \| duration_95th \| 95 分位请求延迟 \| `histogram_quantile(0.95, sum(rate(tidb_server_handle_query_duration_seconds_bucket{sql_type!="internal"}[1m])) by (namespace,instance,le))` \|
		\| tidb \| duration_avg \| 平均请求延迟 \| `sum(rate(tidb_server_handle_query_duration_seconds_sum{sql_type!="internal"}[2m])) by (namespace,instance) / sum(rate(tidb_server_handle_query_duration_seconds_count{sql_type!="internal"}[2m])) by (namespace,instance)` \|
		\| tidb \| qps \| 请求数量 \| `sum(rate(tidb_executor_statement_total[1m])) by (namespace,instance,type)` \|
		\| tidb \| slow_query \| 慢查询 \| `histogram_quantile(0.90, sum(rate(tidb_server_slow_query_process_duration_seconds_bucket{sql_type="general"}[1m])) by (namespace,instance,le,sql_type))` \|
		\| tidb \| block_cache_size \| block_cache_size \| `avg(tikv_engine_block_cache_size_bytes{db="kv"}) by (namespace,instance,cf)` \|
		\| tidb \| uptime \| 服务存活时长 \| `max(time() - process_start_time_seconds{cluster="tidb",component="tidb"}) by (namespace,instance)` \|
		\| tidb \| cpu_usage \| CPU 使用率 \| `sum(rate(process_cpu_seconds_total{cluster="tidb",component="tidb"}[1m])) by (namespace,instance)` \|
		\| tidb \| memory_usage \| 内存使用量 \| `avg(process_resident_memory_bytes{cluster="tidb",component="tidb"}) by (namespace,instance)` \|
		\| tidb \| statement_avg_queries \| 多语句平均查询数 \| `sum(rate(tidb_server_multi_query_num_sum[1m])) / sum(rate(tidb_server_multi_query_num_count[1m]))` \|
		\| tidb \| transaction_retry \| 事务重试次数 \| `sum(rate(tidb_session_retry_num_sum[1m])) by (namespace,instance)` \|
		\| tidb \| ddl_job_count \| DDL 作业数 \| `sum(tidb_ddl_running_job_count) by (namespace,instance,state)` \|
		\| tidb \| top_sql_cpu \| TopSQL CPU 消耗 \| `sum(rate(tidb_topsql_report_duration_seconds_sum[1m])) by (namespace,instance,sql_type)` \|
		\| tidb \| server_is_up \| 服务存活节点数 \| `count(up{cluster="tidb",component="tidb"} == 1) by (namespace)` \|
		\| tikv \| store_size \| 已用存储容量 \| `sum(tikv_store_size_bytes{type="used"}) by (namespace,instance)` \|
		\| tikv \| available_size \| 可用存储容量 \| `sum(tikv_store_size_bytes{type="available"}) by (namespace,instance)` \|
		\| tikv \| capacity_size \| 总存储容量 \| `sum(tikv_store_size_bytes{type="capacity"}) by (namespace,instance)` \|
		\| tikv \| cpu_usage \| CPU 使用率 \| `sum(rate(process_cpu_seconds_total{cluster="tidb",component="tikv"}[1m])) by (namespace,instance)` \|
		\| tikv \| memory_usage \| 内存使用量 \| `avg(process_resident_memory_bytes{cluster="tidb",component="tikv"}) by (namespace,instance)` \|
		\| tikv \| io_util \| IO 利用率 \| `rate(node_disk_io_time_seconds_total[1m])` \|
		\| tikv \| write_wal_mbps \| 写入速率 (Mbps) \| `sum(rate(tikv_engine_flow_bytes{type="wal_file_bytes"}[1m])) by (namespace,instance)` \|
		\| tikv \| read_mbps \| 读取速率 (Mbps) \| `sum(rate(tikv_engine_flow_bytes{type=~"bytes_read\|iter_bytes_read"}[1m])) by (namespace,instance)` \|
		\| tikv \| qps \| QPS \| `sum(rate(tikv_storage_command_total[1m])) by (namespace,instance,type)` \|
		\| tikv \| grpc_qps \| 各类 gRPC 请求 QPS \| `sum(rate(tikv_grpc_msg_duration_seconds_count{type=~"coprocessor\|kv_get\|kv_batch_get\|kv_prewrite\|kv_commit"}[1m])) by (namespace,instance,type)` \|
		\| tikv \| threadpool_readpool_cpu \| StorageReadPool 线程池 CPU \| `sum(rate(tikv_thread_cpu_seconds_total{name=~"unified_read_pool_.*"}[1m])) by (namespace,instance,name)` \|
		\| tikv \| raft_propose_wait \| RaftPropose 等待延迟 P99 \| `histogram_quantile(0.99, sum(rate(tikv_raftstore_request_wait_time_duration_secs_bucket[1m])) by (le,namespace,instance))` \|
		\| tikv \| raft_apply_wait \| RaftApply 等待延迟 P99 \| `histogram_quantile(0.99, sum(rate(tikv_raftstore_apply_wait_time_duration_secs_bucket[1m])) by (le,namespace,instance))` \|
		\| tikv \| region_pending \| PendingRegion 数量 \| `sum(tikv_raftstore_read_index_pending_duration_count) by (namespace,instance)` \|
		\| tikv \| rocksdb_write_stall \| RocksDB 写阻塞次数 \| `sum(tikv_engine_write_stall) by (namespace,instance)` \|
		\| tikv \| snapshot_apply_count \| SnapshotApply 次数 \| `sum(rate(tikv_raftstore_apply_duration_secs_count[1m])) by (namespace,instance)` \|
		\| tikv \| server_is_up \| 服务存活节点数 \| `count(up{cluster="tidb",component="tikv"} == 1) by (namespace)` \|
		\| tikv \| cpu_usage \| CPU 使用率 \| `sum(rate(process_cpu_seconds_total{cluster="tidb",component="tikv"}[1m])) by (namespace,instance)` \|
		\| pd \| storage_capacity \| 集群总容量 \| `sum(pd_cluster_status{type="storage_capacity"}) by (namespace)` \|
		\| pd \| storage_size \| 已用容量 \| `sum(pd_cluster_status{type="storage_size"}) by (namespace)` \|
		\| pd \| storage_used_ratio \| 已用容量比 \| `sum(pd_cluster_status{type="storage_size"}) / sum(pd_cluster_status{type="storage_capacity"})` \|
		\| pd \| store_up_count \| 健康 Store 数量 \| `sum(pd_cluster_status{type="store_up_count"}) by (namespace)` \|
		\| pd \| store_down_count \| Down Store 数量 \| `sum(pd_cluster_status{type="store_down_count"}) by (namespace)` \|
		\| pd \| store_unhealth_count \| Unhealth Store 数量 \| `sum(pd_cluster_status{type="store_unhealth_count"}) by (namespace)` \|
		\| pd \| store_low_space_count \| 低空间 Store 数量 \| `sum(pd_cluster_status{type="store_low_space_count"}) by (namespace)` \|
		\| pd \| store_slow_count \| 慢 Store 数量 \| `sum(pd_cluster_status{type="store_slow_count"}) by (namespace)` \|
		\| pd \| abnormal_region_count \| 异常 Region 数量 \| `sum(pd_regions_status{type=~"pending-peer-region-count\\|down-peer-region-count\\|offline-peer-region-count\\|miss-peer-region-count\\|learner-peer-region-count"}) by (namespace,type)` \|
		\| pd \| region_health \| Region 健康状态 \| `max(pd_regions_status) by (namespace,type)` \|
		\| pd \| leader_primary \| PDLeader-Primary 节点 \| `service_member_role{cluster="tidb",component="pd",service="PD"}` \|
		\| pd \| region_count \| Region 总数 \| `sum(pd_cluster_status{type="region_count"}) by (namespace)` \|
		\| pd \| leader_count \| Leader 总数 \| `sum(pd_cluster_status{type="leader_count"}) by (namespace)` \|
		\| pd \| learner_count \| Learner 总数 \| `sum(pd_cluster_status{type="learner_count"}) by (namespace)` \|
		\| pd \| witness_count \| Witness 总数 \| `sum(pd_cluster_status{type="witness_count"}) by (namespace)` \|
		\| pd \| cpu_usage \| CPU 使用率 \| `sum(rate(process_cpu_seconds_total{cluster="tidb",component="pd"}[1m])) by (namespace,instance)` \|
		\| pd \| memory_usage \| 内存使用量 \| `avg(process_resident_memory_bytes{cluster="tidb",component="pd"}) by (namespace,instance)` \|

		字段释义（infra 样例数据文件）