[ 
https://issues.apache.org/jira/browse/IOTDB-5164?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Gaofei Cao reassigned IOTDB-5164:
---------------------------------

      Sprint: 2023-1-Storage  (was: 2023-1-ConfigNode)
    Assignee: Jinrui Zhang  (was: Gaofei Cao)

> [disk]datanode takes too much disk space, should improve
> --------------------------------------------------------
>
>                 Key: IOTDB-5164
>                 URL: https://issues.apache.org/jira/browse/IOTDB-5164
>             Project: Apache IoTDB
>          Issue Type: Improvement
>            Reporter: changxue
>            Assignee: Jinrui Zhang
>            Priority: Major
>         Attachments: iotdb-common.properties, iotdb-confignode.properties, 
> iotdb-datanode.properties
>
>
> [disk]datanode takes too much disk space, should improve
> Here is the disk taking state of one node, it shows 124G data would take 230G 
> in one node, and there are 3 nodes with 3 replicas, so 124G data would take 6 
> times of real one. This is too much.
> {code}
> 124G  ./datanode/data/sequence
> 51M   ./datanode/data/unsequence
> 104G  ./datanode/data/snapshot
> 228G  ./datanode/data
> 414M  ./datanode/wal/root.test-0
> 401M  ./datanode/wal/root.test-3
> 394M  ./datanode/wal/root.test-1
> 410M  ./datanode/wal/root.test-2
> 394M  ./datanode/wal/root.test-4
> 2.0G  ./datanode/wal
> 4.0K  ./datanode/system/compression_ratio
> 16K   ./datanode/system/schema
> 4.0K  ./datanode/system/roles
> 8.0K  ./datanode/system/users
> 48K   ./datanode/system/databases
> 4.0K  ./datanode/system/upgrade
> 8.0K  ./datanode/system/udf
> 100K  ./datanode/system
> 5.2M  ./datanode/consensus/schema_region
> 356K  ./datanode/consensus/data_region
> 5.6M  ./datanode/consensus
> 230G  ./datanode
> 4.0K  ./confignode/system/roles
> 8.0K  ./confignode/system/users
> 4.0K  ./confignode/system/procedure
> 24K   ./confignode/system
> 4.1M  ./confignode/consensus/47474747-4747-4747-4747-000000000000
> 4.1M  ./confignode/consensus
> 4.1M  ./confignode
> 230G  .
> {code}
> 124G的数据,单个节点上要占用230G的空间,这是个3节点集群配置的3副本,所以,它总共要占用6倍的磁盘空间。这实在太多了,我觉得需要优化。咱们snapshot的设计是否有部分重复。这部分空间是否可以复用。
> 说明:可能是因为磁盘空间不足导致readonly, 然后snapshot。



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to