patroni+etcd部署使用callbacks回调函数实现failover时的vip漂移

如果是重启类参数，此时查看集群列表会有显示需要重启生效，需要按先从后主的顺序重启。登录数据库，查看初始化用户已成功创建，但当前的pg_hba规则是全部免密。再次查看集群列表，成功切换，并且vip漂移到了新主节点。pg_cluster1_node1节点。删除集群前，停止所有patroni节点。在所有etcd节点执行。在所有etcd节点执行。在所有etcd节点执行。在所有etcd节点执行。在所有的数据

自撰良方服之卒

623人浏览 · 2026-02-28 17:15:15

自撰良方服之卒 · 2026-02-28 17:15:15 发布

patroni+etcd部署使用callbacks回调函数实现failover时的vip漂移

节点规划

IP	hostname	部署服务
10.0.0.101	PostgreSQL-01	etcd-v3.5.21+PostgreSQL 15.5+ patroni 4.0.5
10.0.0.102	PostgreSQL-02	etcd-v3.5.21+PostgreSQL 15.5+ patroni 4.0.5
10.0.0.103	PostgreSQL-03	etcd-v3.5.21+PostgreSQL 15.5+ patroni 4.0.5
10.0.0.104	PostgreSQL-04	PostgreSQL 15.5 + patroni 4.0.5
10.0.0.105	PostgreSQL-05	PostgreSQL 15.5 + patroni 4.0.5
10.0.0.106	PostgreSQL-06	PostgreSQL 15.5 + patroni 4.0.5

1.安装etcd

在所有etcd节点执行

[root@PostgreSQL-01 ~]# cd /opt
[root@PostgreSQL-01 opt]# wget https://github.com/etcd-io/etcd/releases/download/v3.5.21/etcd-v3.5.21-linux-amd64.tar.gz
[root@PostgreSQL-01 opt]# tar xf etcd-v3.5.21-linux-amd64.tar.gz
[root@PostgreSQL-01 opt]# cd etcd-v3.5.21-linux-amd64
[root@PostgreSQL-01 etcd-v3.5.21-linux-amd64]# cp etcd /usr/bin/
[root@PostgreSQL-01 etcd-v3.5.21-linux-amd64]# cp etcdctl /usr/bin/

[root@PostgreSQL-01 etcd-v3.5.21-linux-amd64]# etcd --version
etcd Version: 3.5.21
Git SHA: a17edfd
Go Version: go1.23.7
Go OS/Arch: linux/amd64

2.配置启动etcd

2.1 编辑etcd配置文件

在所有etcd节点执行

# 所有节点都要配置，按本地ip修改配置
[root@PostgreSQL-01 ~]# mkdir /data/etcd
[root@PostgreSQL-01 ~]# chown -R postgres.postgres /data/etcd
[root@PostgreSQL-01 ~]# cd /data/etcd
[root@PostgreSQL-01 ~]# vim etcd.yaml
# [Member]
name: etcd1
data-dir: "/data/etcd"
listen-client-urls: "http://127.0.0.1:2379,http://10.0.0.101:2379"
listen-peer-urls: "http://10.0.0.101:2380"
# [Cluster]
initial-advertise-peer-urls: "http://10.0.0.101:2380"
initial-cluster: "etcd1=http://10.0.0.101:2380,etcd2=http://10.0.0.102:2380,etcd3=http://10.0.0.103:2380"
initial-cluster-token: "etcd-cluster"
initial-cluster-state: "new"
advertise-client-urls: "http://127.0.0.1:2379,http://10.0.0.101:2379"
# [Logging]
log-outputs:
  - "/data/etcd/etcd.log"  # 必须是数组形式
  - "stderr"               # 可以同时输出到多个目标
log-level: warn
enable-log-rotation: true
log-rotation-config-json: '{"maxsize": 100, "maxage": 7, "maxbackups": 0, "localtime": true, "compress": true}'

# 以下为不同节点需要修改的内容
name: etcd1
listen-client-urls: "http://127.0.0.1:2379,http://本机ip:2379"
listen-peer-urls: "http://本机ip:2380"
initial-advertise-peer-urls: "http://本机ip:2380"
advertise-client-urls: "http://127.0.0.1:2379,http://本机ip:2379"

2.2 配置system启动文件

在所有etcd节点执行

[root@PostgreSQL-01 ~]# vi /usr/lib/systemd/system/etcd.service
[Unit]
Description=Etcd Server
After=network.target network-online.target
Wants=network-online.target

[Service]
Type=notify
User=postgres
ExecStart=/usr/bin/etcd --config-file=/data/etcd/etcd.yaml
Restart=on-failure
RestartSec=10s
LimitNOFILE=65535

[Install]
WantedBy=multi-user.target

2.3 启动etcd

在所有etcd节点执行

[root@PostgreSQL-01 ~]# systemctl daemon-reload
[root@PostgreSQL-01 ~]# systemctl enable etcd
# 单个节点启动后会hang在前台，因为找不到集群其他节点，其他节点启动后才能进入到后台
[root@PostgreSQL-01 ~]# systemctl start etcd

查看成员列表

[root@PostgreSQL-01 etcd]# etcdctl -w table  member list

在这里插入图片描述

查看主节点

[root@PostgreSQL-01 etcd]# etcdctl --endpoints=10.0.0.101:2379,10.0.0.102:2379,10.0.0.103:2379 -w table endpoint status

在这里插入图片描述

3.安装patroni

3.1 安装patroni

在所有的数据库节点执行

[root@PostgreSQL-01 ~]# mkdir /data/patroni -p
[root@PostgreSQL-01 ~]# pip3 install 'psycopg2-binary==2.9.8'
[root@PostgreSQL-01 ~]# pip3 install 'patroni[etcd]==4.0.5'
# gcc版本较低可能会报错，可以尝试使用以下方式安装
# CFLAGS='-std=c99' pip3 install 'patroni[etcd]==4.0.5'
[root@PostgreSQL-01 ~]# patroni --version
patroni 4.0.5

3.2 配置postgres用户的sudo权限

在所有的数据库节点执行

# failover时需要postgres执行ip与arping等操作，需要sudo权限
[root@PostgreSQL-01 ~]# vim /etc/sudoers
postgres ALL=(ALL) NOPASSWD: ALL

4. 配置启动patroni

4.1 编辑patroni配置文件

在所有数据节点节点配置

# 同一集群中的数据库节点都要配置，注意每个节点不同的参数是：
# name: pg_cluster1_node1
# restapi 里的 connect_address: 10.0.0.101:8008
# postgresql 里的 connect_address: 10.0.0.101:5432

[root@PostgreSQL-01 ~]# cd /data/patroni
[root@PostgreSQL-01 patroni]# vim pg_cluster1.yaml
log:
  dir: /data/patroni
  type: plain
  file_size: 104857600
  file_num: 5
  level: INFO
  date_format: '+%Y-%m-%d %H:%M:%S %z'
# 集群名
scope: pg_cluster1
# 定义 etcd 中存储集群相关信息的目录，在etcd的保存目录为/namespace/scope
namespace: /service
# 节点名称
name: pg_cluster1_node1
# 表示 Patroni 的 REST API 监听的地址和端口
restapi:
  listen: 0.0.0.0:8008
  connect_address: 10.0.0.101:8008

# etcd 集群的地址列表，如果使用etcd2版本，这里就用"etcd:",如果使用etcd3版本,这里就用"etcd3:"
etcd3:
  hosts: 10.0.0.101:2379,10.0.0.102:2379,10.0.0.103:2379

bootstrap:
# 分布式一致性系统（DCS）相关的配置
  dcs:
# Leader 锁的租约时间，如果超过这个时间 Leader 没有续约，将触发故障转移
    ttl: 30
# 主循环的休眠时间，即 Patroni 每次检查集群状态的间隔
    loop_wait: 10
# DCS 和 PostgreSQL 操作重试的超时时间
    retry_timeout: 5
# 以上三个值需要满足 "loop_wait + 2 * retry_timeout <= ttl"
    maximum_lag_on_failover: 1048576
    master_start_timeout: 300
    synchronous_mode: true
    postgresql:
      use_pg_rewind: true
      use_slots: true
      # patroni启动后，会将数据库的原配置文件重命名为postgresql.base.conf，按照以下配置来新生成一份配置文件postgresql.conf，新的配置文件头部会include 'postgresql.base.conf'。
      parameters:
        hot_standby: 'on'
        listen_addresses: '0.0.0.0'
        port: '5432'
        max_connections: '1000'
        password_encryption: 'md5'
        max_locks_per_transaction: '64'
        max_prepared_transactions: '0'
        max_replication_slots: '10'
        max_wal_senders: '10'
        max_worker_processes: '8'
        max_stack_depth: '2MB'
        track_commit_timestamp: 'off'
        log_timezone: 'Asia/Shanghai'
        logging_collector: 'on'
        log_destination: 'stderr'
        log_min_messages: 'warning'   
        log_directory: 'log'
        log_filename: 'postgresql-%Y-%m-%d.log' 
        log_truncate_on_rotation: off 
        log_rotation_age: '1d'
        log_rotation_size: '0'
        log_min_duration_statement: '1000'
        log_statement: 'none'
        lc_messages: 'en_US.UTF-8'
        log_line_prefix: '%t [%p]: user:%u,db:%d,app:%a,client:%h '
        shared_buffers: '1GB'
        temp_buffers: '8MB'
        work_mem: '8MB'
        maintenance_work_mem: '128MB' 
        huge_pages: 'on'
        temp_file_limit: '20GB'
        fsync: 'on'
        commit_delay: '0'
        full_page_writes: 'on'
        synchronous_commit: 'on'
        wal_level: 'replica'
        wal_compression: 'on'
        wal_keep_size: '10GB'
        wal_level: 'replica'
        wal_log_hints: 'on'
        wal_sync_method: 'fdatasync'
        wal_buffers: '16MB'
        max_wal_size: '40GB'
        min_wal_size: '10GB'
        wal_keep_size: '10GB'
        autovacuum: 'on'
        autovacuum_work_mem: '256MB'
postgresql:   
  listen: 0.0.0.0:5432
  connect_address: 10.0.0.101:5432
  data_dir: /data/pgsql/data
  bin_dir: /usr/local/pgsql/bin
  config_dir: /data/pgsql/data
  authentication:
    replication:
      username: repuser
      password: '123'
    superuser:
      username: postgres
      password: '123'
#回调函数，在数据库发生了启动，重启和主库切换这些动作时，会调用配置的脚本进行其他动作，这里配置的是vip生成脚本
  callbacks:
    on_stop: /bin/bash /data/patroni/pg_cluster1_vip.sh
    on_start: /bin/bash /data/patroni/pg_cluster1_vip.sh
    on_restart: /bin/bash /data/patroni/pg_cluster1_vip.sh
    on_role_change: /bin/bash /data/patroni/pg_cluster1_vip.sh

watchdog:
    mode: off

tags:
# 是否禁止该节点成为主库
    nofailover: false
# 是否从负载均衡池中排除
    noloadbalance: false
# 是否允许其他节点从此节点克隆数据
    clonefrom: false
# 是否禁止该节点成为同步备库
    nosync: false

4.2 编辑vip脚本

在所有数据节点节点配置

[root@PostgreSQL-01 patroni]# cd /data/patroni
[root@PostgreSQL-01 patroni]# vim pg_cluster1_vip.sh
#!/bin/bash
# 加强版的 VIP 管理脚本
exec >> /data/patroni/patroni.log 2>&1

readonly OPERATION=$1
readonly ROLE=$2
readonly SCOPE=$3

VIP='10.0.0.201'
PREFIX='32'
DEV='ens33'
LABEL="${DEV}:vip"

function usage() {
    echo "Usage: $0 <on_start|on_stop|on_role_change> <ROLE> <SCOPE>"
    exit 1
}

function log() {
    echo "$(date "+%Y-%m-%d %H:%M:%S %z") $1"
}

function remove_vip() {
    # 检查 VIP 是否存在
    if ip addr show ${DEV} | grep -q "${VIP}/${PREFIX}"; then
        sudo ip addr del ${VIP}/${PREFIX} dev ${DEV} label ${LABEL}
        log "VIP ${VIP} removed successfully"
        
        # 发送 gratuitous ARP 更新邻居
        sudo arping -q -A -c 3 -I ${DEV} ${VIP} 2>/dev/null
    else
        log "VIP ${VIP} not present, no action needed"
    fi
}

function add_vip() {
    # 检查 VIP 是否已存在
    if ! ip addr show ${DEV} | grep -q "${VIP}/${PREFIX}"; then
        sudo ip addr add ${VIP}/${PREFIX} dev ${DEV} label ${LABEL}
        log "VIP ${VIP} added successfully"
        
        # 发送 gratuitous ARP 通知网络
        sudo arping -q -A -c 3 -I ${DEV} ${VIP}
    else
        log "VIP ${VIP} already present"
    fi
}

# 主逻辑
log "Callback triggered: OPERATION=${OPERATION}, ROLE=${ROLE}, SCOPE=${SCOPE}"

case $OPERATION in
    on_stop)
        log "Service stopping, removing VIP"
        remove_vip
        ;;
        
    on_start | on_restart | on_role_change)
        # 兼容多种角色名称
        if [[ $ROLE == 'master' || $ROLE == 'primary' || $ROLE == 'leader' ]]; then
            add_vip
        else
            log "Role is ${ROLE}, removing VIP"
            remove_vip
        fi
        ;;
        
    *)
        log "Unknown operation: ${OPERATION}"
        usage
        ;;
esac

log "Callback completed"

4.3 配置system启动

在所有数据节点配置

[root@PostgreSQL-01 patroni]# vim /usr/lib/systemd/system/patroni.service
[Unit]
Description=patroni - a high-availability PostgreSQL
Documentation=https://patroni.readthedocs.io/en/latest/index.html
After=syslog.target network.target watchdog.service
Wants=network-online.target

[Service]
Type=simple
User=postgres
Group=postgres
PermissionsStartOnly=true
Environment="PGHOME=/usr/local/pgsql" 
Environment="PGDATA=/data/pgsql/data" 
Environment="PGPORT=5432"
Environment="LD_LIBRARY_PATH=/usr/local/pgsql/lib" 
Environment="PATH=/usr/local/pgsql/bin:/usr/local/bin:/usr/bin:/usr/sbin" 
ExecStart= /usr/local/bin/patroni /data/patroni/pg_cluster1.yaml
ExecReload=/bin/kill -HUP $MAINPID
LimitNOFILE=65536
KillMode=process
KillSignal=SIGINT
Restart=on-abnormal
RestartSec=30s
TimeoutSec=0
[Install]
WantedBy=multi-user.target


# 注意
# Type=simple，不能设置为forking，否则会一直hang住
# 这里设置了这么多变量是因为如果不设置，通过system启动后无法找到相关变量信息

5.启动patroni

5.1 启动第一个patroni节点

[root@PostgreSQL-01 patroni]# systemctl daemon-reload
[root@PostgreSQL-01 patroni]# systemctl start patroni

查看日志

# 查看日志可以发现，数据库服务被拉起，当前节点的角色由replica变为primary，VIP添加成功
[root@PostgreSQL-01 patroni]# tail -f patroni.log 
2026-02-28 14:40:28,059 WARNING: Postgresql is not running.
2026-02-28 14:40:28 +0800 Callback triggered: OPERATION=on_start, ROLE=replica, SCOPE=pg_cluster1
2026-02-28 14:40:28 +0800 Role is replica, removing VIP
2026-02-28 14:40:28 +0800 VIP 10.0.0.201 not present, no action needed
2026-02-28 14:40:28 +0800 Callback completed
2026-02-28 14:40:28 +0800 Callback triggered: OPERATION=on_role_change, ROLE=primary, SCOPE=pg_cluster1
2026-02-28 14:40:28 +0800 VIP 10.0.0.201 added successfully
2026-02-28 14:40:30 +0800 Callback completed

vip成功挂载

在这里插入图片描述

查看数据目录

[root@PostgreSQL-01 ~]# ll /data/pgsql/data
total 120
drwx------ 5 postgres postgres    33 Feb 28 14:21 base
drwx------ 2 postgres postgres  4096 Feb 28 14:40 global
-rw------- 1 postgres postgres   567 Feb 28 14:21 patroni.dynamic.json
drwx------ 2 postgres postgres     6 Feb 28 14:21 pg_commit_ts
drwx------ 2 postgres postgres     6 Feb 28 14:21 pg_dynshmem
-rw------- 1 postgres postgres  4791 Feb 28 14:21 pg_hba.conf
-rw------- 1 postgres postgres  4791 Feb 28 14:40 pg_hba.conf.backup
-rw------- 1 postgres postgres  1636 Feb 28 14:21 pg_ident.conf
-rw------- 1 postgres postgres  1636 Feb 28 14:40 pg_ident.conf.backup
drwx------ 4 postgres postgres    68 Feb 28 14:40 pg_logical
drwx------ 4 postgres postgres    36 Feb 28 14:21 pg_multixact
drwx------ 2 postgres postgres     6 Feb 28 14:21 pg_notify
drwx------ 2 postgres postgres     6 Feb 28 14:21 pg_replslot
drwx------ 2 postgres postgres     6 Feb 28 14:21 pg_serial
drwx------ 2 postgres postgres     6 Feb 28 14:21 pg_snapshots
drwx------ 2 postgres postgres     6 Feb 28 14:40 pg_stat
drwx------ 2 postgres postgres     6 Feb 28 14:21 pg_stat_tmp
drwx------ 2 postgres postgres    18 Feb 28 14:21 pg_subtrans
drwx------ 2 postgres postgres     6 Feb 28 14:21 pg_tblspc
drwx------ 2 postgres postgres     6 Feb 28 14:21 pg_twophase
-rw------- 1 postgres postgres     3 Feb 28 14:21 PG_VERSION
drwx------ 3 postgres postgres   116 Feb 28 14:40 pg_wal
drwx------ 2 postgres postgres    18 Feb 28 14:21 pg_xact
-rw------- 1 postgres postgres    88 Feb 28 14:21 postgresql.auto.conf
-rw------- 1 postgres postgres 29442 Feb 28 14:21 postgresql.base.conf
-rw------- 1 postgres postgres 29442 Feb 28 14:40 postgresql.base.conf.backup
-rw------- 1 postgres postgres   724 Feb 28 14:40 postgresql.conf
-rw------- 1 postgres postgres   724 Feb 28 14:40 postgresql.conf.backup
-rw------- 1 postgres postgres   447 Feb 28 14:40 postmaster.opts
-rw------- 1 postgres postgres    80 Feb 28 14:40 postmaster.pid

登录数据库，查看初始化用户已成功创建，但当前的pg_hba规则是全部免密

在这里插入图片描述

修改pg_hba文件，并重载配置

[postgres@PostgreSQL-01 data]$ vi pg_hba.conf
# TYPE  DATABASE        USER            ADDRESS                 METHOD
host    replication     repuser         10.0.0.0/24             md5
host    replication     repuser         127.0.0.1/32            md5
host    replication     repuser         ::1/128                 md5
host    all             postgres        10.0.0.0/24             md5
host    all             postgres        127.0.0.1/32            md5
local   all             postgres                                md5
host    all             postgres        ::1/128                 md5

[postgres@PostgreSQL-01 data]$ pg_ctl reload -D $PGDATA

数据库成功启用密码验证

在这里插入图片描述

5.2 启动后续节点

[root@PostgreSQL-02 patroni]# systemctl daemon-reload
[root@PostgreSQL-02 patroni]# systemctl start patroni
[root@PostgreSQL-02 patroni]# vi patroni.log

在这里插入图片描述

[root@PostgreSQL-03 patroni]# systemctl daemon-reload
[root@PostgreSQL-03 patroni]# systemctl start patroni
[root@PostgreSQL-03 patroni]# vi patroni.log

在这里插入图片描述

5.3 查看集群列表

patronictl -c /data/patroni/pg_cluster1.yaml list

在这里插入图片描述

5.4 使用vip登录数据库

在这里插入图片描述

6.运维操作

查看当前集群节点列表

patronictl -c /data/patroni/pg_cluster1.yaml list

查看当前集群参数

patronictl -c /data/patroni/pg_cluster1.yaml show-config

修改集群参数

patronictl -c /data/patroni/pg_cluster1.yaml edit-config
# 修改前的值会有红色标记
# 修改后的值会有绿色标记

在这里插入图片描述

patroni日志中会有参数修改的记录，并且postgresql.conf中的参数值会被修改

[root@PostgreSQL-02 patroni]# grep "max_connections" patroni.log 

[root@PostgreSQL-02 patroni]# grep "max_connections" /data/pgsql/data/postgresql.conf

在这里插入图片描述

如果是重启类参数，此时查看集群列表会有显示需要重启生效，需要按先从后主的顺序重启

在这里插入图片描述

重启节点

[root@PostgreSQL-01 patroni]# patronictl -c /data/patroni/pg_cluster1.yaml restart pg_cluster1 pg_cluster1_node2

switchover操作

[root@PostgreSQL-01 patroni]# patronictl -c /data/patroni/pg_cluster1.yaml switchover

# 注意，只能切换到同步流节点，否则会报错
Switchover failed, details: 412, candidate name does not match with sync_standby

在这里插入图片描述

再次查看集群列表，成功切换，并且vip漂移到了新主节点

在这里插入图片描述

查看切换日志

pg_cluster1_node1节点

2026-02-28 16:31:43,407 INFO: received switchover request with leader=pg_cluster1_node1 candidate=pg_cluster1_node2 scheduled_at=None
2026-02-28 16:31:43,442 INFO: Got response from pg_cluster1_node2 http://10.0.0.102:8008/patroni: {"state": "running", "postmaster_start_time": "2026-02-28 15:36:36.085886+08:00", "role": "replica", "server_version": 150005, "xlog": {"received_location": 83886408, "replayed_location": 83886408, "replayed_timestamp": "2026-02-28 15:36:53.229199+08:00", "paused": false}, "sync_standby": true, "timeline": 1, "replication_state": "streaming", "dcs_last_seen": 1772267496, "database_system_identifier": "7611814854766790160", "pending_restart": true, "pending_restart_reason": {"max_connections": {"old_value": "1000", "new_value": "500"}}, "patroni": {"version": "4.0.5", "scope": "pg_cluster1", "name": "pg_cluster1_node2"}}
2026-02-28 16:31:43,629 INFO: Got response from pg_cluster1_node2 http://10.0.0.102:8008/patroni: {"state": "running", "postmaster_start_time": "2026-02-28 15:36:36.085886+08:00", "role": "replica", "server_version": 150005, "xlog": {"received_location": 83886408, "replayed_location": 83886408, "replayed_timestamp": "2026-02-28 15:36:53.229199+08:00", "paused": false}, "sync_standby": true, "timeline": 1, "replication_state": "streaming", "dcs_last_seen": 1772267496, "database_system_identifier": "7611814854766790160", "pending_restart": true, "pending_restart_reason": {"max_connections": {"old_value": "1000", "new_value": "500"}}, "patroni": {"version": "4.0.5", "scope": "pg_cluster1", "name": "pg_cluster1_node2"}}
2026-02-28 16:31:43,577 INFO: Lock owner: pg_cluster1_node1; I am pg_cluster1_node1
2026-02-28 16:31:43,772 INFO: switchover: demoting myself
2026-02-28 16:31:43,772 INFO: Demoting self (graceful)
2026-02-28 16:31:44 +0800 Callback triggered: OPERATION=on_stop, ROLE=primary, SCOPE=pg_cluster1
2026-02-28 16:31:44 +0800 Service stopping, removing VIP
2026-02-28 16:31:44 +0800 VIP 10.0.0.201 removed successfully
2026-02-28 16:31:44 +0800 Callback completed
2026-02-28 16:31:44,941 INFO: Leader key released
2026-02-28 16:31:44,943 INFO: Lock owner: None; I am pg_cluster1_node1
2026-02-28 16:31:44,943 INFO: not healthy enough for leader race
2026-02-28 16:31:44,943 INFO: switchover: demote in progress
2026-02-28 16:31:44,947 INFO: Lock owner: pg_cluster1_node2; I am pg_cluster1_node1
2026-02-28 16:31:44,947 INFO: switchover: demote in progress
2026-02-28 16:31:46,211 INFO: Lock owner: pg_cluster1_node2; I am pg_cluster1_node1
2026-02-28 16:31:46,211 INFO: switchover: demote in progress
2026-02-28 16:31:46,947 INFO: Local timeline=1 lsn=0/50001F8
2026-02-28 16:31:46,955 INFO: primary_timeline=2
2026-02-28 16:31:46,959 INFO: primary: history=1        0/5000270       no recovery target specified
2026-02-28 16:31:46,960 INFO: closed patroni connections to postgres
2026-02-28 16:31:46,963 INFO: max_connections value in pg_controldata: 1000, in the global configuration: 500. pg_controldata value will be used. Setting 'Pending restart' flag
2026-02-28 16:31:47,304 INFO: postmaster pid=10608
2026-02-28 16:31:47 +0800 Callback triggered: OPERATION=on_role_change, ROLE=replica, SCOPE=pg_cluster1
2026-02-28 16:31:47 +0800 Role is replica, removing VIP
2026-02-28 16:31:47 +0800 VIP 10.0.0.201 not present, no action needed
2026-02-28 16:31:47 +0800 Callback completed
2026-02-28 16:31:56,208 INFO: Lock owner: pg_cluster1_node2; I am pg_cluster1_node1
2026-02-28 16:31:56,209 INFO: establishing a new patroni heartbeat connection to postgres
2026-02-28 16:31:56,307 INFO: no action. I am (pg_cluster1_node1), a secondary, and following a leader (pg_cluster1_node2)
2026-02-28 16:32:06,256 INFO: no action. I am (pg_cluster1_node1), a secondary, and following a leader (pg_cluster1_node2)

删除集群

删除集群前，停止所有patroni节点

patronictl -c /data/patroni/pg_cluster1.yaml  remove pg_cluster1

2048 AI社区

有“AI”的1024 = 2048，欢迎大家加入2048 AI社区

更多推荐

首都在线海外大带宽筑基，加速大模型出海全球化进程

2048 AI社区

《数据库性能飞跃：SQL优化与Explain实战指南》

2048 AI社区

大模型登记流程全解析

2048 AI社区

所有评论(0)

查看更多评论

自撰良方服之卒

@weixin_41690116

已为社区贡献1条内容

patroni+etcd部署 使用callbacks回调函数实现failover时的vip漂移

自撰良方服之卒

patroni+etcd部署 使用callbacks回调函数实现failover时的vip漂移

节点规划

1.安装etcd

2.配置启动etcd

2.1 编辑etcd配置文件

2.2 配置system启动文件

2.3 启动etcd

3.安装patroni

3.1 安装patroni

3.2 配置postgres用户的sudo权限

4. 配置启动patroni

4.1 编辑patroni配置文件

4.2 编辑vip脚本

4.3 配置system启动

5.启动patroni

5.1 启动第一个patroni节点

5.2 启动后续节点

5.3 查看集群列表

5.4 使用vip登录数据库

6.运维操作

查看当前集群节点列表

查看当前集群参数

修改集群参数

重启节点

switchover操作

删除集群

所有评论(0)

自撰良方服之卒

patroni+etcd部署使用callbacks回调函数实现failover时的vip漂移

patroni+etcd部署使用callbacks回调函数实现failover时的vip漂移