使用默认配置启动CnosDB单节点后客户端无法连接求助

问题详情 #

• 环境：CnosDB 2.4.0，修订版721266b
• 操作流程：
  1. 执行命令启动单节点：sudo ./target/release/cnosdb run -M singleton
  2. 用cnosdb-cli连接后输入\d，触发报错：

     error sending request for url (http://localhost:8902/api/v1/sql?tenant=cnosdb&db=public&chunked=false): error trying to connect: tcp connect error: Connection refused (os error 61)
     

  3. 查看启动日志，未发现服务端口相关输出，启动日志内容如下：

     ➜  cnosdb git:(main) ✗ sudo ./target/release/cnosdb run  -M singleton
     [global]
     node_id = 1001
     host = 'localhost'
     cluster_name = 'cluster_xxx'
     store_metrics = true
     
     [deployment]
     mode = 'query_tskv'
     cpu = 8
     memory = 16
     
     [meta]
     service_addr = ['127.0.0.1:8901']
     report_time_interval_secs = 30
     
     [query]
     max_server_connections = 10240
     query_sql_limit = 16777216
     write_sql_limit = 167772160
     auth_enabled = false
     read_timeout_ms = 3000
     write_timeout_ms = 3000
     stream_trigger_cpu = 1
     stream_executor_cpu = 2
     
     [storage]
     path = '/var/lib/cnosdb/data'
     max_summary_size = '128M'
     base_file_size = '16M'
     flush_req_channel_cap = 16
     max_cached_readers = 32
     max_level = 4
     compact_trigger_file_num = 4
     compact_trigger_cold_duration = '1h'
     max_compact_size = '2G'
     max_concurrent_compaction = 4
     strict_write = false
     reserve_space = '0'
     
     [wal]
     enabled = true
     path = '/var/lib/cnosdb/wal'
     wal_req_channel_cap = 64
     max_file_size = '1G'
     flush_trigger_total_file_size = '2G'
     sync = false
     sync_interval = '0'
     
     [cache]
     max_buffer_size = '128M'
     partition = 8
     
     [log]
     level = 'info'
     path = '/var/log/cnosdb'
     
     [security]
     
     [service]
     grpc_enable_gzip = false
     enable_report = true
     ...
     

排查步骤 #

1. 补全服务端口配置
   客户端请求的8902端口未在日志中出现，说明服务可能未绑定该端口。在配置的[service]段添加明确的端口绑定：

   [service]
   http_addr = "0.0.0.0:8902"
   rpc_addr = "0.0.0.0:8901"
   grpc_enable_gzip = false
   enable_report = true
   

   重启服务后再尝试连接。

2. 验证端口监听状态
   执行命令检查8902端口是否被CnosDB进程监听：

   netstat -tulpn | grep cnosdb
   # 或
   lsof -i :8902
   

   若端口未被监听，去/var/log/cnosdb目录查看详细日志，排查启动失败原因（比如目录权限不足、依赖缺失等）。

3. 修复目录权限
   启动用了sudo，但要确保数据和日志目录权限正确：

   ls -ld /var/lib/cnosdb /var/log/cnosdb
   

   若权限异常，执行命令修正：

   sudo chown -R cnosdb:cnosdb /var/lib/cnosdb /var/log/cnosdb
   

4. 切换启动模式测试
   尝试用standalone模式启动服务，替代当前的singleton：

   sudo ./target/release/cnosdb run -M standalone
   

   启动后再测试客户端连接。

5. 调整客户端连接参数
   若服务端绑定的是0.0.0.0，尝试用IP指定连接：

   cnosdb-cli --host 127.0.0.1
   

内容的提问来源于stack exchange，提问作者flashbo