Details
-
Bug
-
Status: Closed
-
Major
-
Resolution: Fixed
-
0.14.0-SNAPSHOT
-
None
-
2022-11-Cluster
Description
[write]data region leader write many wal files after restarting datanode on it
environment:
3C3D cluster, Nov. 21
reproduction:
1. Using iotdb-benchmarks write data to iotdb cluster for more than 6 hours, only 1 device 1 sensor with double values. 2 replicas.
2. The 46 node failed to writing data, so I restart data node of it, and it's the data region leader
3. Continue writing data to the same timeseries for about 8 hours. I find that most of data lay on 44 node
问题：
1. 为什么重启46前，44，46节点上的数据分布还是很均衡的，重启46后，wal文件几乎就只写在44上了呢
2. 为什么写了那么多的wal文件，远远大于数据数量和size
show regions(before and after restart datanode are the same):
|RegionId|Type|Status|Database|SeriesSlotId|TimeSlotId|DataNodeId|Host|RpcPort|Role|
| 10 | SchemaRegion | Running | root.aggr.g_0 | 1 | 0 | 1 | 172.20.70.44 | 6667 | Follower |
| 10 | SchemaRegion | Running | root.aggr.g_0 | 1 | 0 | 5 | 172.20.70.46 | 6667 | Leader |
| 11 | DataRegion | Running | root.aggr.g_0 | 1 | 10 | 1 | 172.20.70.44 | 6667 | Follower |
| 11 | DataRegion | Running | root.aggr.g_0 | 1 | 10 | 5 | 172.20.70.46 | 6667 | Leader |
iotdb-1: 44
iotdb-2: 45
iotdb-3: 46
files:
atmos@i-rh6m726k root.aggr.g_0]$ ansible allnodes -m shell -a "find $IOTDB_HOME/data/datanode/data/sequence/root.aggr.g_0 -type f |wc -l"
iotdb-1 | CHANGED | rc=0 >>
1694
iotdb-2 | CHANGED | rc=0 >>
966
iotdb-3 | CHANGED | rc=0 >>
183
monitor:
Attachments
Attachments
Issue Links
- links to