混合部署拓扑

本文介绍 TiDB 集群的 TiKV 和 TiDB 混合部署拓扑以及主要参数。常见的场景为,部署机为多路 CPU 处理器,内存也充足,为提高物理机资源利用率,可单机多实例部署,即 TiDB、TiKV 通过 numa 绑核,隔离 CPU 资源。PD 和 Prometheus 混合部署,但两者的数据目录需要使用独立的文件系统。

拓扑信息

实例个数物理机配置IP配置
TiDB632 VCore 64GB10.0.1.1
10.0.1.2
10.0.1.3
配置 numa 绑核操作
PD316 VCore 32 GB10.0.1.4
10.0.1.5
10.0.1.6
配置 location_labels 参数
TiKV632 VCore 64GB10.0.1.7
10.0.1.8
10.0.1.9
1. 区分实例级别的 port、status_port;
2. 配置全局参数 readpool、storage 以及 raftstore;
3. 配置实例级别 host 维度的 labels;
4. 配置 numa 绑核操作
Monitoring & Grafana14 VCore 8GB * 1 500GB (ssd)10.0.1.10默认配置

拓扑模版

简单混部配置模板
# # Global variables are applied to all deployments and used as the default value of # # the deployments if a specific deployment value is missing. global: user: "tidb" ssh_port: 22 deploy_dir: "/tidb-deploy" data_dir: "/tidb-data" server_configs: tikv: readpool.unified.max-thread-count: <The value refers to the calculation formula result of the multi-instance topology document.> readpool.storage.use-unified-pool: false readpool.coprocessor.use-unified-pool: true storage.block-cache.capacity: "<The value refers to the calculation formula result of the multi-instance topology document.>" raftstore.capacity: "<The value refers to the calculation formula result of the multi-instance topology document.>" pd: replication.location-labels: ["host"] pd_servers: - host: 10.0.1.4 - host: 10.0.1.5 - host: 10.0.1.6 tidb_servers: - host: 10.0.1.1 port: 4000 status_port: 10080 numa_node: "0" - host: 10.0.1.1 port: 4001 status_port: 10081 numa_node: "1" - host: 10.0.1.2 port: 4000 status_port: 10080 numa_node: "0" - host: 10.0.1.2 port: 4001 status_port: 10081 numa_node: "1" - host: 10.0.1.3 port: 4000 status_port: 10080 numa_node: "0" - host: 10.0.1.3 port: 4001 status_port: 10081 numa_node: "1" tikv_servers: - host: 10.0.1.7 port: 20160 status_port: 20180 numa_node: "0" config: server.labels: { host: "tikv1" } - host: 10.0.1.7 port: 20161 status_port: 20181 numa_node: "1" config: server.labels: { host: "tikv1" } - host: 10.0.1.8 port: 20160 status_port: 20180 numa_node: "0" config: server.labels: { host: "tikv2" } - host: 10.0.1.8 port: 20161 status_port: 20181 numa_node: "1" config: server.labels: { host: "tikv2" } - host: 10.0.1.9 port: 20160 status_port: 20180 numa_node: "0" config: server.labels: { host: "tikv3" } - host: 10.0.1.9 port: 20161 status_port: 20181 numa_node: "1" config: server.labels: { host: "tikv3" } monitoring_servers: - host: 10.0.1.10 grafana_servers: - host: 10.0.1.10 alertmanager_servers: - host: 10.0.1.10
详细混部配置模板
# # Global variables are applied to all deployments and used as the default value of # # the deployments if a specific deployment value is missing. global: user: "tidb" ssh_port: 22 deploy_dir: "/tidb-deploy" data_dir: "/tidb-data" monitored: node_exporter_port: 9100 blackbox_exporter_port: 9115 deploy_dir: "/tidb-deploy/monitored-9100" data_dir: "/tidb-data/monitored-9100" log_dir: "/tidb-deploy/monitored-9100/log" server_configs: tidb: log.slow-threshold: 300 tikv: readpool.unified.max-thread-count: <The value refers to the calculation formula result of the multi-instance topology document.> readpool.storage.use-unified-pool: false readpool.coprocessor.use-unified-pool: true storage.block-cache.capacity: "<The value refers to the calculation formula result of the multi-instance topology document.>" raftstore.capacity: "<The value refers to the calculation formula result of the multi-instance topology document.>" pd: replication.location-labels: ["host"] schedule.leader-schedule-limit: 4 schedule.region-schedule-limit: 2048 schedule.replica-schedule-limit: 64 pd_servers: - host: 10.0.1.4 - host: 10.0.1.5 - host: 10.0.1.6 tidb_servers: - host: 10.0.1.1 port: 4000 status_port: 10080 deploy_dir: "/tidb-deploy/tidb-4000" log_dir: "/tidb-deploy/tidb-4000/log" numa_node: "0" - host: 10.0.1.1 port: 4001 status_port: 10081 deploy_dir: "/tidb-deploy/tidb-4001" log_dir: "/tidb-deploy/tidb-4001/log" numa_node: "1" - host: 10.0.1.2 port: 4000 status_port: 10080 deploy_dir: "/tidb-deploy/tidb-4000" log_dir: "/tidb-deploy/tidb-4000/log" numa_node: "0" - host: 10.0.1.2 port: 4001 status_port: 10081 deploy_dir: "/tidb-deploy/tidb-4001" log_dir: "/tidb-deploy/tidb-4001/log" numa_node: "1" - host: 10.0.1.3 port: 4000 status_port: 10080 deploy_dir: "/tidb-deploy/tidb-4000" log_dir: "/tidb-deploy/tidb-4000/log" numa_node: "0" - host: 10.0.1.3 port: 4001 status_port: 10081 deploy_dir: "/tidb-deploy/tidb-4001" log_dir: "/tidb-deploy/tidb-4001/log" numa_node: "1" tikv_servers: - host: 10.0.1.7 port: 20160 status_port: 20180 deploy_dir: "/tidb-deploy/tikv-20160" data_dir: "/tidb-data/tikv-20160" log_dir: "/tidb-deploy/tikv-20160/log" numa_node: "0" config: server.labels: { host: "tikv1" } - host: 10.0.1.7 port: 20161 status_port: 20181 deploy_dir: "/tidb-deploy/tikv-20161" data_dir: "/tidb-data/tikv-20161" log_dir: "/tidb-deploy/tikv-20161/log" numa_node: "1" config: server.labels: { host: "tikv1" } - host: 10.0.1.8 port: 20160 status_port: 20180 deploy_dir: "/tidb-deploy/tikv-20160" data_dir: "/tidb-data/tikv-20160" log_dir: "/tidb-deploy/tikv-20160/log" numa_node: "0" config: server.labels: { host: "tikv2" } - host: 10.0.1.8 port: 20161 status_port: 20181 deploy_dir: "/tidb-deploy/tikv-20161" data_dir: "/tidb-data/tikv-20161" log_dir: "/tidb-deploy/tikv-20161/log" numa_node: "1" config: server.labels: { host: "tikv2" } - host: 10.0.1.9 port: 20160 status_port: 20180 deploy_dir: "/tidb-deploy/tikv-20160" data_dir: "/tidb-data/tikv-20160" log_dir: "/tidb-deploy/tikv-20160/log" numa_node: "0" config: server.labels: { host: "tikv3" } - host: 10.0.1.9 port: 20161 status_port: 20181 deploy_dir: "/tidb-deploy/tikv-20161" data_dir: "/tidb-data/tikv-20161" log_dir: "/tidb-deploy/tikv-20161/log" numa_node: "1" config: server.labels: { host: "tikv3" } monitoring_servers: - host: 10.0.1.10 # ssh_port: 22 # port: 9090 # deploy_dir: "/tidb-deploy/prometheus-8249" # data_dir: "/tidb-data/prometheus-8249" # log_dir: "/tidb-deploy/prometheus-8249/log" grafana_servers: - host: 10.0.1.10 # port: 3000 # deploy_dir: /tidb-deploy/grafana-3000 alertmanager_servers: - host: 10.0.1.10 # ssh_port: 22 # web_port: 9093 # cluster_port: 9094 # deploy_dir: "/tidb-deploy/alertmanager-9093" # data_dir: "/tidb-data/alertmanager-9093" # log_dir: "/tidb-deploy/alertmanager-9093/log"

以上 TiDB 集群拓扑文件中,详细的配置项说明见通过 TiUP 部署 TiDB 集群的拓扑文件配置

混合部署的关键参数介绍

本节介绍单机多实例的关键参数,主要用于 TiDB、TiKV 的单机多实例部署场景。你需要按照提供的计算公式,将结果填写至上一步的配置模板中。

  • TiKV 进行配置优化

    • readpool 线程池自适应,配置 readpool.unified.max-thread-count 参数可以使 readpool.storagereadpool.coprocessor 共用统一线程池,同时要分别设置自适应开关。

      • 开启 readpool.storagereadpool.coprocessor

        readpool.storage.use-unified-pool: true readpool.coprocessor.use-unified-pool: true
      • 计算公式如下:

        readpool.unified.max-thread-count = cores * 0.8 / TiKV 数量
    • storage CF (all RocksDB column families) 内存自适应,配置 storage.block-cache.capacity 参数即可实现 CF 之间自动平衡内存使用。

      • 计算公式如下:

        storage.block-cache.capacity = (MEM_TOTAL * 0.5 / TiKV 实例数量)
    • 如果多个 TiKV 实例部署在同一块物理磁盘上,需要在 tikv 配置中添加 capacity 参数:

      raftstore.capacity = 磁盘总容量 / TiKV 实例数量
  • label 调度配置

    由于采用单机多实例部署 TiKV,为了避免物理机宕机导致 Region Group 默认 3 副本的 2 副本丢失,导致集群不可用的问题,可以通过 label 来实现 PD 智能调度,保证同台机器的多 TiKV 实例不会出现 Region Group 只有 2 副本的情况。

    • TiKV 配置

      相同物理机配置相同的 host 级别 label 信息:

      config: server.labels: host: tikv1
    • PD 配置

      PD 需要配置 labels 类型来识别并调度 Region:

      pd: replication.location-labels: ["host"]
  • numa_node 绑核

    • 在实例参数模块配置对应的 numa_node 参数,并添加对应的物理 CPU 的核数;

    • numa 绑核使用前,确认已经安装 numactl 工具,以及物理机对应的物理机 CPU 的信息后,再进行参数配置;

    • numa_node 这个配置参数与 numactl --membind 配置对应。