[
https://issues.apache.org/jira/browse/IOTDB-5557?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17691188#comment-17691188
]
Yukun Zhou commented on IOTDB-5557:
-----------------------------------
The root cause is the async mechanism of redoing raft log in ratis.
> [ metadata ] The metadata query results are inconsistent
> --------------------------------------------------------
>
> Key: IOTDB-5557
> URL: https://issues.apache.org/jira/browse/IOTDB-5557
> Project: Apache IoTDB
> Issue Type: Bug
> Components: Core/Schema Manager, mpp-cluster
> Affects Versions: 1.1.0-SNAPSHOT
> Reporter: 刘珍
> Assignee: Yukun Zhou
> Priority: Major
> Attachments: image-2023-02-20-14-04-32-611.png
>
>
> master : 0219_0cd4461
> 启动集群，log_datanode_all.log出现enjoy后，查询元数据，出现查询结果不一致（动态增加，直到全部元数据加载到内存）。
> 期望：只要集群已经开始提供查询服务，就要保证查询结果的一致性。
> 测试环境：
> 1. 192.168.10.76 48cpu 384GB 内存
> 元数据信息：1db，1万设备，600序列/dev。
> ConfigNode：
> MAX_HEAP_SIZE="8G"
> DataNode：
> MAX_HEAP_SIZE="256G"
> MAX_DIRECT_MEMORY_SIZE="32G"
> COMMON配置
> time_partition_interval=60480000000
> query_timeout_threshold=36000000
> enable_seq_space_compaction=false
> enable_unseq_space_compaction=false
> enable_cross_space_compaction=false
> 2. 清操作系统缓存，启动数据库，出现enjoy后，执行count devices查看结果
> cat check_device_count.sh
> while true
> do
> v_start=`grep enjoy logs/log_datanode_all.log|wc -l`
> if [[ ${v_start} = "1" ]];then
> for i in {1..100}
> do
> ./sbin/start-cli.sh -h 192.168.10.76 -e "count devices;"
> >> dev_count_during_start.out
> done
> break
> fi
> done
> 下图结果，可以看出，count devices的结果在动态增加，直至10000，完全加载到内存中：
> !image-2023-02-20-14-04-32-611.png!
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
