[
https://issues.apache.org/jira/browse/IOTDB-4350?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17645746#comment-17645746
]
Xinyu Tan commented on IOTDB-4350:
----------------------------------
自从考虑 SyncStatus 的 IoTConsensus 内存控制完善以后,这个现象已经不复存在。建议复测,如果没什么问题就 close 了吧
> [ MultiLeader Throttle Down] Performance does not return to normal after
> “Throttle Down“
> -----------------------------------------------------------------------------------------
>
> Key: IOTDB-4350
> URL: https://issues.apache.org/jira/browse/IOTDB-4350
> Project: Apache IoTDB
> Issue Type: Bug
> Components: mpp-cluster
> Affects Versions: 0.14.0-SNAPSHOT
> Reporter: 刘珍
> Assignee: 张洪胤
> Priority: Major
> Fix For: 1.0.0
>
> Attachments: image-2022-09-07-14-52-58-266.png, net_restart.conf,
> screenshot-1.png, screenshot-2.png, screenshot-3.png, screenshot-4.png,
> screenshot-5.png, screenshot-6.png
>
>
> m_0905_0095eb3,3副本3C3D
> 3个dataregion , 每个node上有1个leader。
> ip72 断网3分钟(16:52 ~ 16:55),查看集群状态,切主成功后,
> ip73断网2分钟,之后不执行故障操作。
> 同步慢,multiLeader一直在写入限流,但是限流性能也回不去,如下,统计1分钟的写入数据量(bm中的batch)
> !screenshot-6.png!
> IoTDB> select count(latency) from
> root.result.moresession_2022_09_06_04_47_03.INGESTION where okPoint>0 group
> by ([1662454041076000186,1662459764764000179),1m);
> +-----------------------------------+--------------------------------------------------------------------+
> |
> Time|count(root.result.moresession_2022_09_06_04_47_03.INGESTION.latency)|
> +-----------------------------------+--------------------------------------------------------------------+
> |2022-09-06T16:47:21.076000186+08:00|
> 5544|
> |2022-09-06T16:48:21.076000186+08:00|
> 6282|
> |2022-09-06T16:49:21.076000186+08:00|
> 5671|
> |2022-09-06T16:50:21.076000186+08:00|
> 4589|
> |2022-09-06T16:51:21.076000186+08:00|
> 5350|
> |2022-09-06T16:52:21.076000186+08:00|
> 1121|
> |2022-09-06T16:53:21.076000186+08:00|
> 901|
> |2022-09-06T16:54:21.076000186+08:00|
> 201|
> |2022-09-06T16:55:21.076000186+08:00|
> 334|
> |2022-09-06T16:56:21.076000186+08:00|
> 3501|
> |2022-09-06T16:57:21.076000186+08:00|
> 3677|
> |2022-09-06T16:58:21.076000186+08:00|
> 3111|
> |2022-09-06T16:59:21.076000186+08:00|
> 1948|
> |2022-09-06T17:00:21.076000186+08:00|
> 3889|
> |2022-09-06T17:01:21.076000186+08:00|
> 2982|
> |2022-09-06T17:02:21.076000186+08:00|
> 4465|
> |2022-09-06T17:03:21.076000186+08:00|
> 4871|
> |2022-09-06T17:04:21.076000186+08:00|
> 4478|
> |2022-09-06T17:05:21.076000186+08:00|
> 3242|
> |2022-09-06T17:06:21.076000186+08:00|
> 2545|
> |2022-09-06T17:07:21.076000186+08:00|
> 2579|
> |2022-09-06T17:08:21.076000186+08:00|
> 133|
> |2022-09-06T17:09:21.076000186+08:00|
> 488|
> |2022-09-06T17:10:21.076000186+08:00|
> 253|
> |2022-09-06T17:11:21.076000186+08:00|
> 445|
> |2022-09-06T17:12:21.076000186+08:00|
> 2122|
> |2022-09-06T17:13:21.076000186+08:00|
> 1799|
> |2022-09-06T17:14:21.076000186+08:00|
> 1568|
> |2022-09-06T17:15:21.076000186+08:00|
> 355|
> |2022-09-06T17:16:21.076000186+08:00|
> 1127|
> |2022-09-06T17:17:21.076000186+08:00|
> 803|
> |2022-09-06T17:18:21.076000186+08:00|
> 674|
> |2022-09-06T17:19:21.076000186+08:00|
> 621|
> |2022-09-06T17:20:21.076000186+08:00|
> 361|
> |2022-09-06T17:21:21.076000186+08:00|
> 367|
> |2022-09-06T17:22:21.076000186+08:00|
> 999|
> |2022-09-06T17:23:21.076000186+08:00|
> 1119|
> |2022-09-06T17:24:21.076000186+08:00|
> 1113|
> |2022-09-06T17:25:21.076000186+08:00|
> 1737|
> |2022-09-06T17:26:21.076000186+08:00|
> 1282|
> |2022-09-06T17:27:21.076000186+08:00|
> 4454|
> |2022-09-06T17:28:21.076000186+08:00|
> 2013|
> |2022-09-06T17:29:21.076000186+08:00|
> 623|
> |2022-09-06T17:30:21.076000186+08:00|
> 313|
> |2022-09-06T17:31:21.076000186+08:00|
> 455|
> |2022-09-06T17:32:21.076000186+08:00|
> 353|
> |2022-09-06T17:33:21.076000186+08:00|
> 347|
> |2022-09-06T17:34:21.076000186+08:00|
> 587|
> |2022-09-06T17:35:21.076000186+08:00|
> 1370|
> |2022-09-06T17:36:21.076000186+08:00|
> 341|
> |2022-09-06T17:37:21.076000186+08:00|
> 1555|
> |2022-09-06T17:38:21.076000186+08:00|
> 3266|
> |2022-09-06T17:39:21.076000186+08:00|
> 1344|
> |2022-09-06T17:40:21.076000186+08:00|
> 1057|
> |2022-09-06T17:41:21.076000186+08:00|
> 682|
> |2022-09-06T17:42:21.076000186+08:00|
> 231|
> |2022-09-06T17:43:21.076000186+08:00|
> 170|
> |2022-09-06T17:44:21.076000186+08:00|
> 729|
> |2022-09-06T17:45:21.076000186+08:00|
> 118|
> |2022-09-06T17:46:21.076000186+08:00|
> 135|
> |2022-09-06T17:47:21.076000186+08:00|
> 109|
> |2022-09-06T17:48:21.076000186+08:00|
> 167|
> |2022-09-06T17:49:21.076000186+08:00|
> 139|
> |2022-09-06T17:50:21.076000186+08:00|
> 138|
> |2022-09-06T17:51:21.076000186+08:00|
> 321|
> |2022-09-06T17:52:21.076000186+08:00|
> 138|
> |2022-09-06T17:53:21.076000186+08:00|
> 326|
> |2022-09-06T17:54:21.076000186+08:00|
> 166|
> |2022-09-06T17:55:21.076000186+08:00|
> 70|
> |2022-09-06T17:56:21.076000186+08:00|
> 302|
> |2022-09-06T17:57:21.076000186+08:00|
> 587|
> |2022-09-06T17:58:21.076000186+08:00|
> 25|
> |2022-09-06T17:59:21.076000186+08:00|
> 427|
> |2022-09-06T18:00:21.076000186+08:00|
> 2|
> |2022-09-06T18:01:21.076000186+08:00|
> 96|
> |2022-09-06T18:02:21.076000186+08:00|
> 72|
> |2022-09-06T18:03:21.076000186+08:00|
> 94|
> |2022-09-06T18:04:21.076000186+08:00|
> 99|
> |2022-09-06T18:05:21.076000186+08:00|
> 66|
> |2022-09-06T18:06:21.076000186+08:00|
> 230|
> |2022-09-06T18:07:21.076000186+08:00|
> 10|
> |2022-09-06T18:08:21.076000186+08:00|
> 335|
> |2022-09-06T18:09:21.076000186+08:00|
> 25|
> |2022-09-06T18:10:21.076000186+08:00|
> 10|
> |2022-09-06T18:11:21.076000186+08:00|
> 18|
> |2022-09-06T18:12:21.076000186+08:00|
> 142|
> |2022-09-06T18:13:21.076000186+08:00|
> 281|
> |2022-09-06T18:14:21.076000186+08:00|
> 30|
> |2022-09-06T18:15:21.076000186+08:00|
> 14|
> |2022-09-06T18:16:21.076000186+08:00|
> 8|
> |2022-09-06T18:17:21.076000186+08:00|
> 7|
> |2022-09-06T18:18:21.076000186+08:00|
> 38|
> |2022-09-06T18:19:21.076000186+08:00|
> 13|
> |2022-09-06T18:20:21.076000186+08:00|
> 40|
> |2022-09-06T18:21:21.076000186+08:00|
> 12|
> |2022-09-06T18:22:21.076000186+08:00|
> 10|
> +-----------------------------------+--------------------------------------------------------------------+
> Total line number = 96
> 复现流程:
> 1. 机器配置 192.168.10.72/73/74 48核386GB
> bm在ip71
> 2. 数据库配置
> ConfigNode
> MAX_HEAP_SIZE="8G"
> schema_region_consensus_protocol_class=org.apache.iotdb.consensus.ratis.RatisConsensus
> data_region_consensus_protocol_class=org.apache.iotdb.consensus.multileader.MultiLeaderConsensus
> schema_replication_factor=3
> data_replication_factor=3
> DataNode
> MAX_HEAP_SIZE="256G"
> MAX_DIRECT_MEMORY_SIZE="32G"
> max_connection_for_internal_service=1100
> max_waiting_time_when_insert_blocked=3600000
> query_timeout_threshold=3600000
> 2. benchmark见附件
> 故障前的region信息
> !screenshot-1.png!
> 3. 断网
> ip72断网
> cat restart_network.sh
> #!/bin/bash
> sudo ifconfig enp129s0f1 down
> sleep $1
> sudo ifconfig enp129s0f1 up
> nohup sh -x restart_network.sh "120" > a.log &
> 查看region状态,切主成功后,
> !screenshot-2.png!
> ip73也执行断网,ip73恢复后
> !screenshot-3.png!
> 网络传输
> !screenshot-4.png!
> !screenshot-5.png!
--
This message was sent by Atlassian Jira
(v8.20.10#820010)