Tikv集群有6万多key，每个key每10秒更新一次，但是Tikv集群的数据量一直增加，同时Region一直增加

兴文-逍遥子 · 2023 年5 月 24 日 02:37

【 TiDB 使用环境】测试
【 TiDB 版本】v7.0.0
【复现路径】
tiup部署拓扑文件如下：

global:
  user: "tikv"
  ssh_port: 22
  deploy_dir: "/data/data1/tikv-deploy"
  data_dir: "/data/data1/tikv-data"

pd_servers:
  - host: 192.168.0.3
    config:
      schedule.max-merge-region-size: 50
      schedule.max-merge-region-keys: 400000
      schedule.merge-schedule-limit: 16
  - host: 192.168.0.4
    config:
      schedule.max-merge-region-size: 50
      schedule.max-merge-region-keys: 400000
      schedule.merge-schedule-limit: 16
  - host: 192.168.0.5
    config:
      schedule.max-merge-region-size: 50
      schedule.max-merge-region-keys: 400000
      schedule.merge-schedule-limit: 16

tikv_servers:
  - host: 192.168.0.3
    config:
      raftstore.sync-log: false
      raftstore.region-max-size: "384MB"
      raftstore.region-split-size: "256MB"
      raftstore.region-split-check-diff: "32MB"
      raftstore.split-region-check-tick-interval: 50s
      rocksdb.compaction-readahead-size: "2MB"
      rocksdb.max-manifest-file-size: "20MB"
      rocksdb.defaultcf.block-size: "64KB"
      rocksdb.defaultcf.compression-per-level: ["no", "no", "lz4", "lz4", "lz4", "zstd", "zstd"]
      rocksdb.defaultcf.write-buffer-size: "128MB"
      rocksdb.defaultcf.max-write-buffer-number: 10
      rocksdb.defaultcf.level0-slowdown-writes-trigger: 20
      rocksdb.defaultcf.level0-stop-writes-trigger: 36
      rocksdb.defaultcf.max-bytes-for-level-base: "512MB"
      rocksdb.defaultcf.target-file-size-base: "32MB"
      rocksdb.writecf.compression-per-level: ["no", "no", "lz4", "lz4", "lz4", "zstd", "zstd"]
      rocksdb.writecf.write-buffer-size: "128MB"
      rocksdb.writecf.max-write-buffer-number: 10
      rocksdb.writecf.min-write-buffer-number-to-merge: 1
      rocksdb.writecf.max-bytes-for-level-base: "512MB"
      rocksdb.writecf.target-file-size-base: "32MB"
      raftdb.defaultcf.compression-per-level: ["no", "no", "lz4", "lz4", "lz4", "zstd", "zstd"]
      raftdb.defaultcf.write-buffer-size: "128MB"
      raftdb.defaultcf.max-write-buffer-number: 10
      raftdb.defaultcf.min-write-buffer-number-to-merge: 1
      raftdb.defaultcf.max-bytes-for-level-base: "512MB"
      raftdb.defaultcf.target-file-size-base: "32MB"
  - host: 192.168.0.4
    config:
      raftstore.sync-log: false
      raftstore.region-max-size: "384MB"
      raftstore.region-split-size: "256MB"
      raftstore.region-split-check-diff: "32MB"
      raftstore.split-region-check-tick-interval: 50s
      rocksdb.compaction-readahead-size: "2MB"
      rocksdb.max-manifest-file-size: "20MB"
      rocksdb.defaultcf.block-size: "64KB"
      rocksdb.defaultcf.compression-per-level: ["no", "no", "lz4", "lz4", "lz4", "zstd", "zstd"]
      rocksdb.defaultcf.write-buffer-size: "128MB"
      rocksdb.defaultcf.max-write-buffer-number: 10
      rocksdb.defaultcf.level0-slowdown-writes-trigger: 20
      rocksdb.defaultcf.level0-stop-writes-trigger: 36
      rocksdb.defaultcf.max-bytes-for-level-base: "512MB"
      rocksdb.defaultcf.target-file-size-base: "32MB"
      rocksdb.writecf.compression-per-level: ["no", "no", "lz4", "lz4", "lz4", "zstd", "zstd"]
      rocksdb.writecf.write-buffer-size: "128MB"
      rocksdb.writecf.max-write-buffer-number: 10
      rocksdb.writecf.min-write-buffer-number-to-merge: 1
      rocksdb.writecf.max-bytes-for-level-base: "512MB"
      rocksdb.writecf.target-file-size-base: "32MB"
      raftdb.defaultcf.compression-per-level: ["no", "no", "lz4", "lz4", "lz4", "zstd", "zstd"]
      raftdb.defaultcf.write-buffer-size: "128MB"
      raftdb.defaultcf.max-write-buffer-number: 10
      raftdb.defaultcf.min-write-buffer-number-to-merge: 1
      raftdb.defaultcf.max-bytes-for-level-base: "512MB"
      raftdb.defaultcf.target-file-size-base: "32MB"
  - host: 192.168.0.5
    config:
      raftstore.sync-log: false
      raftstore.region-max-size: "384MB"
      raftstore.region-split-size: "256MB"
      raftstore.region-split-check-diff: "32MB"
      raftstore.split-region-check-tick-interval: 50s
      rocksdb.compaction-readahead-size: "2MB"
      rocksdb.max-manifest-file-size: "20MB"
      rocksdb.defaultcf.block-size: "64KB"
      rocksdb.defaultcf.compression-per-level: ["no", "no", "lz4", "lz4", "lz4", "zstd", "zstd"]
      rocksdb.defaultcf.write-buffer-size: "128MB"
      rocksdb.defaultcf.max-write-buffer-number: 10
      rocksdb.defaultcf.level0-slowdown-writes-trigger: 20
      rocksdb.defaultcf.level0-stop-writes-trigger: 36
      rocksdb.defaultcf.max-bytes-for-level-base: "512MB"
      rocksdb.defaultcf.target-file-size-base: "32MB"
      rocksdb.writecf.compression-per-level: ["no", "no", "lz4", "lz4", "lz4", "zstd", "zstd"]
      rocksdb.writecf.write-buffer-size: "128MB"
      rocksdb.writecf.max-write-buffer-number: 10
      rocksdb.writecf.min-write-buffer-number-to-merge: 1
      rocksdb.writecf.max-bytes-for-level-base: "512MB"
      rocksdb.writecf.target-file-size-base: "32MB"
      raftdb.defaultcf.compression-per-level: ["no", "no", "lz4", "lz4", "lz4", "zstd", "zstd"]
      raftdb.defaultcf.write-buffer-size: "128MB"
      raftdb.defaultcf.max-write-buffer-number: 10
      raftdb.defaultcf.min-write-buffer-number-to-merge: 1
      raftdb.defaultcf.max-bytes-for-level-base: "512MB"
      raftdb.defaultcf.target-file-size-base: "32MB"

monitoring_servers:
  - host: 192.168.0.18

grafana_servers:
  - host: 192.168.0.18

alertmanager_servers:
  - host: 192.168.0.18

【遇到的问题：问题现象及影响】Tikv集群有6万多key，每个key每10秒更新一次，但是Tikv集群的数据量一直增加，同时Region一直增加
【资源配置】进入到 TiDB Dashboard -集群信息 (Cluster Info) -主机(Hosts) 截图此页面

【附件：截图/日志/监控】
Tikv监控：

系统信息：

裤衩儿飞上天 · 2023 年5 月 24 日 02:41

Tidb的insert update delete 都会写入新的数据，
最终是gc删除数据的~

兴文-逍遥子 · 2023 年5 月 24 日 02:46

因为我的需求是想直接用tikv这个kv存储功能，所以没有部署tidb服务

pingyu · 2023 年5 月 24 日 03:27

是使用 RawKV 还是 Transaction KV 方式写入的数据？
如果是 Transaction KV，需要部署至少一个 TiDB 节点用于驱动 GC。

兴文-逍遥子 · 2023 年5 月 24 日 05:58

使用的是 Transaction KV写入数据，那我明白了，我去部署一个TIDB节点试试看

兴文-逍遥子 · 2023 年5 月 24 日 06:58

如果我的应用服务直接使用tikv的接口写入的数据，tidb在gc的时候会处理不是从自己这里写入的数据啊？

pingyu · 2023 年5 月 24 日 07:31

会处理的

另外有一个地方需要注意，Transaction KV 写入的 key 不要以 t 和 m 开头
避免跟 TiDB 冲突。即使 TiDB 没有负载，也会写入少量元数据

兴文-逍遥子 · 2023 年5 月 24 日 08:23

好的，目前启动了TIDB看着确实是在GC了

system · 2023 年7 月 23 日 08:24

此话题已在最后回复的 60 天后被自动关闭。不再允许新回复。