节点启动或停止失败类问题 可能影响 DN主节点启动失败,会导致访问到该DN的节点SQL报错,实例部分不可用; CN主节点启动失败,会导致流入该节点的SQL语句报错,流入其它CN主节点的DDL语句报错; CN/DN备节点失败失败,如果开启同步复制,同步复制节点数量不足且未启用退化策略时,会导致DDL、DML语句卡住; CN/DN备节点失败失败,可能会导致无可用备节点,主节点再次异常会导致实例不可用,有数据丢失风险。 解决步骤 1. 按照错误提示,修正postgresql.conf文件内容; 2. 重新发起节点启动任务,或等待节点自动拉起。 端口被占用导致启动失败问题 问题描述 节点启动失败,启动日志(在Agent目录logs/clslog下对应节点的日志pgctl.startxxx.log)显示报错: 地址已在使用,如: 20240115 09:37:40.881 CST 22855,coord(0,0) 0LOG: could not bind IPv4 address "0.0.0.0": 地址已在使用 20240115 09:37:40.881 CST 22855,coord(0,0) 0HINT: Is another postmaster already running on port 11006? If not, wait a few seconds and retry. 20240115 09:37:40.881 CST 22855,coord(0,0) 0LOG: could not bind IPv6 address "::": 地址已在使用 20240115 09:37:40.881 CST 22855,coord(0,0) 0HINT: Is another postmaster already running on port 11006? If not, wait a few seconds and retry. 20240115 09:37:40.881 CST 22855,coord(0,0) 0WARNING: could not create listen socket for "" 20240115 09:37:40.881 CST 22855,coord(0,0) 0FATAL: could not create any TCP/IP sockets20240115 09:37:40.881 CST 22855,coord(0,0) 0LOG: database system is shut down 这里显示11006端口已被使用。