- 本节主要解释hape gs子命令的结果
- 需要注意的是以下状态中只有havenask状态中的sqlClientInfo的clusterStatus为READY表示最终集群可服务。表的READY状态只能表示表注册完成,不代表可服务
- 对应子命令hape gs havenask
- 格式如下,含义:
- serviceZk 表示服务运行的zk地址
- leaderAlive 表示admin是否正常运行
- hippoZk 表示服务worker进程调度的zk地址
- processors 表示admin和worker的进程信息
- role表示该节点在集群中的地位。例如2列的searcher分别对应两个role:database_partition_0和database.database_partition_1。一个role下可能有多个相同的机器服务,称为replica
- havenask的role主要有admin、qrs和多分片的searcher
- containerName表示容器在所在机器上的名称
- containerStatus表示容器是否正常运行
- processorStatus表示进程是否正常运行
- sqlClusterInfo 表示havenask worker节点调度的状态
- replicaNodeId 表示某个role下的所有replica节点Id
- slaveAddress 表示节点地址
- clusterStatus 表示该集群是否达到可服务的状态
{
"serviceZk": "xxx",
"leaderAlive": true,
"hippoZk": "xxx",
"processors": [
{
"ip": "xxx",
"processorName": "suez_admin_worker",
"role": "admin",
"containerName": "havenask-sql-local_appmaster",
"containerStatus": "RUNNING",
"processorStatus": "RUNNING"
},
...
],
sqlClusterInfo": {
"clusterStatus": "READY",
"qrs": [
{
"replicaNodeId": "qrs.qrs_partition_0.0",
"workerStatus": "WS_READY",
"serviceStatus": "SVT_AVAILABLE",
"healthStatus": "HT_ALIVE",
"slotId": {
"id": 1,
"declareTime": 1693527281,
"slaveAddress": "xxx"
}
}
],
"database": {
"database_partition_0": [
{
"replicaNodeId": "database.database_partition_0.0",
"workerStatus": "WS_READY",
"serviceStatus": "SVT_AVAILABLE",
"healthStatus": "HT_ALIVE",
"slotId": {
"id": 0,
"declareTime": 1693527281,
"slaveAddress": "xxx.8"
}
}
]
}
},
}
- 对应子命令hape gs table
- 格式如下,含义:
- status: 表是否准备好
- tableName: 表名
- tableStructureConfig: 表分片和类型
- shardCount 分片数
- shardFields 主键
- buildType DIRECT表示直写表,如果该字段不存在表示全量表
- columns: 表字段
- indexes: 表索引
{
"status": "READY",
"tableName": "in0",
"tableStructureConfig": {
"buildType": "DIRECT",
"shardInfo": {
"shardFunc": "HASH",
"shardFields": [
"id"
],
"shardCount": 1
},
"tableType": "NORMAL"
},
"indexes": [
{
"indexConfig": {
"indexFields": [
{
"fieldName": "id"
}
]
},
"name": "id",
"indexType": "PRIMARY_KEY64"
},
...
],
"operationMeta": {},
"version": "3",
"databaseName": "database",
"columns": [
{
"type": "UINT32",
"name": "id"
},
...
],
"catalogName": "catalog"
}
- 对应子命令hape gs swift
- 格式如下,含义:
- serviceZk 表示服务运行的zk地址
- leaderAlive 表示admin是否正常运行
- hippoZk 表示服务worker进程调度的zk地址
- processors 含义同havenask状态中的processors
- swift的role主要有admin、和多分片的broker
{
"serviceZk": "xxx",
"leaderAlive": true,
"hippoZk": "xxx",
"processors": [
{
"ip": "xxxx",
"processorName": "swift_admin",
"role": "admin",
"containerName": "xxx",
"containerStatus": "RUNNING",
"processorStatus": "RUNNING"
},
...
]
}
- 对应子命令hape gs bs
- 格式如下,含义:
- serviceZk 表示服务运行的zk地址
- leaderAlive 表示admin是否正常运行
- hippoZk 表示服务worker进程调度的zk地址
- processors 含义同havenask状态中的processors
- bs的role主要有以下,会根据情况动态拉起:
- prepare_data_source 全量准备节点
- processor.full全量文档预处理
- builder.full 全量索引构建
- merger.full 全量索引合并
- processor.inc 实时文档预处理
- builder.inc 增量索引构建
- merger.inc 增量索引合并
- 当full进程全部结束,inc进程拉起时代表全量完成
- buildInfo 表示当前正在构建的索引任务
- generationId 表示任务的序号
- tableName 表示正在构建索引的表
{
"serviceZk": "xxx",
"leaderAlive": true,
"hippoZk": "xxx",
"processors": [
{
"ip": "xxx",
"processorName": "build_service_worker",
"role": "xxx",
"containerName": "xxx",
"containerStatus": "RUNNING",
"processorStatus": "RUNNING"
},
...
],
"buildInfo": {
"in1": {
"buildId": {
"partitionName": "in1_part",
"tableName": "in1",
"generationId": "xxx",
"databaseName": "database",
"catalogName": "catalog"
},
"current": {
"buildState": "RUNNING",
"configPath": "xxx"
},
"target": {
"type": "BATCH_BUILD",
"buildState": "RUNNING",
"configPath": "xxx"
}
}
}
}