diff --git "a/content/zh/post/gaoyunlong/PostgreSQL\344\270\216openGauss\344\271\213\345\210\206\345\214\272\346\200\247\350\203\275.md" "b/content/zh/post/gaoyunlong/PostgreSQL\344\270\216openGauss\344\271\213\345\210\206\345\214\272\346\200\247\350\203\275.md"
new file mode 100644
index 0000000000000000000000000000000000000000..fc62d44f51625df9f1c7ce88c5c8d433630e7ad1
--- /dev/null
+++ "b/content/zh/post/gaoyunlong/PostgreSQL\344\270\216openGauss\344\271\213\345\210\206\345\214\272\346\200\247\350\203\275.md"
@@ -0,0 +1,232 @@
++++
+
+title = "PostgreSQL与openGauss之分区性能"
+
+date = "2021-03-08"
+
+tags = ["openGauss与PostgreSQL对比"]
+
+archives = "2021-03"
+
+author = "高云龙"
+
+summary = "PostgreSQL与openGauss之分区性能"
+
+img = "/zh/post/gaoyunlong/title/img25.png"
+
+times = "10:40"
+
++++
+
+# PostgreSQL与openGauss之分区性能
+
+## 概述
+
+PostgreSQL与openGauss分区表定义差异，请参考https://www.modb.pro/db/41393。
+
+openGauss1.1.0开始支持hash/list分区，hash分区表最多支持64个分区，否则会报：
+
+```
+ERROR: Un-support feature
+DETAIL: The partition’s length should be less than 65.
+```
+
+本次对PostgreSQL 和 openGauss 64个子分区表的常规操作对比。
+
+服务器配置：虚拟机8G4C50G
+
+数据库版本：PostgreSQL13.1、openGauss1.1.0
+
+## 添加分区表
+
+PostgreSQL数据库：
+
+```
+--创建父表
+CREATE TABLE partition_table(
+ id int,
+ col1 character varying(16),
+ create_time timestamptz
+) PARTITION BY HASH(id);
+
+--添加分区
+SELECT 'CREATE TABLE partition_table_' || n || ' PARTITION of partition_table FOR VALUES WITH (MODULUS 64, REMAINDER ' || n || ');' FROM generate_series(0,63) as n ;\gexec
+
+--初始化数据
+INSERT INTO partition_table(id,col1,create_time) SELECT round(100000000*random()), n || '_col1',now() FROM generate_series(1,10000000) n;
+
+--添加索引
+CREATE INDEX ON partition_table USING BTREE(id);
+CREATE INDEX ON partition_table USING BTREE(col1);
+```
+
+openGauss数据库：
+
+```
+--创建分区表
+create table partition_table(
+ id int,
+ col1 varchar(16),
+ create_time timestamptz default now())
+partition by hash(id)
+(partition part_hash_1,
+partition part_hash_2,
+partition part_hash_3,
+partition part_hash_4,
+partition part_hash_5,
+partition part_hash_6,
+partition part_hash_7,
+partition part_hash_8,
+partition part_hash_9,
+partition part_hash_10,
+partition part_hash_11,
+partition part_hash_12,
+partition part_hash_13,
+partition part_hash_14,
+partition part_hash_15,
+partition part_hash_16,
+partition part_hash_17,
+partition part_hash_18,
+partition part_hash_19,
+partition part_hash_20,
+partition part_hash_21,
+partition part_hash_22,
+partition part_hash_23,
+partition part_hash_24,
+partition part_hash_25,
+partition part_hash_26,
+partition part_hash_27,
+partition part_hash_28,
+partition part_hash_29,
+partition part_hash_30,
+partition part_hash_31,
+partition part_hash_32,
+partition part_hash_33,
+partition part_hash_34,
+partition part_hash_35,
+partition part_hash_36,
+partition part_hash_37,
+partition part_hash_38,
+partition part_hash_39,
+partition part_hash_40,
+partition part_hash_41,
+partition part_hash_42,
+partition part_hash_43,
+partition part_hash_44,
+partition part_hash_45,
+partition part_hash_46,
+partition part_hash_47,
+partition part_hash_48,
+partition part_hash_49,
+partition part_hash_50,
+partition part_hash_51,
+partition part_hash_52,
+partition part_hash_53,
+partition part_hash_54,
+partition part_hash_55,
+partition part_hash_56,
+partition part_hash_57,
+partition part_hash_58,
+partition part_hash_59,
+partition part_hash_60,
+partition part_hash_61,
+partition part_hash_62,
+partition part_hash_63,
+partition part_hash_64);
+
+--初始化数据
+INSERT INTO partition_table(id,col1,create_time) SELECT round(100000000*random()), n || '_col1',now() FROM generate_series(1,10000000) n;
+
+--添加全局索引
+CREATE INDEX ON partition_table USING BTREE(id);
+CREATE INDEX ON partition_table USING BTREE(col1);
+
+--添加本地索引
+CREATE INDEX ON partition_table USING BTREE(id) local;
+CREATE INDEX ON partition_table USING BTREE(col1) local;
+```
+
+## 测试方法
+
+采用pgbench压测工具，自定义压测脚本的方式来对比。
+
+```
+cat bench.sql
+\set idpp random(1,100000)
+--insert into partition_table values(:idpp,:idpp||'_col1',now());
+--update partition_table set create_time=now() where id=:idpp;
+--update partition_table set create_time=now() where col1=:idpp||'_col1';
+--select * from partition_table where id=:idpp;
+--select * from partition_table where col1=:idpp||'_col1';
+
+pgbench -p 5432 -j 30 -c 30 -M prepared -T 30 -n yunlong -f bench.sql
+```
+
+## 结果对比
+
+
+| |
+分区键查询
+ |
+非分区键查询
+ |
+分区键更新
+ |
+非分区键更新
+ |
+插入
+ |
+
|---|
+
+PostgreSQL
+ |
+0.594 ms
+ |
+7.978 ms
+ |
+1.612 ms
+ |
+17.413 ms
+ |
+17.2ms
+ |
+
+openGauss（全局索引）
+ |
+0.612 ms
+ |
+0.758 ms
+ |
+10.450 ms
+ |
+88.151 ms
+ |
+78.082 ms
+ |
+
+openGauss（本地索引）
+ |
+5.635 ms
+ |
+6.765 ms
+ |
+15.187 ms
+ |
+94.614 ms
+ |
+84.927 ms
+ |
+
+
+
+
+结果对比发现，
+
+1、Postgresql13.1 版本在分区方面总来看优越于openGauss1.1.0。
+
+2、opengauss 全局索引会比本地索引性能更好，但全局索引维护成本高。
+
+3、非分区键查询，带全局索引的opengauss查询性能最快。
+
+此测试受限于服务器环境，数据仅做参考比对。
+
diff --git a/content/zh/post/jiajunfeng/openGauss+KeepAlived.md b/content/zh/post/jiajunfeng/openGauss+KeepAlived.md
new file mode 100644
index 0000000000000000000000000000000000000000..54dee04e9e50acd10230c1b83fdeaeb203b8ac07
--- /dev/null
+++ b/content/zh/post/jiajunfeng/openGauss+KeepAlived.md
@@ -0,0 +1,304 @@
++++
+
+title = "openGauss+KeepAlived"
+
+date = "2021-03-08"
+
+tags = ["openGauss+KeepAlived"]
+
+archives = "2021-03"
+
+author = "贾军锋"
+
+summary = "openGauss+KeepAlived"
+
+img = "/zh/post/jiajunfeng/title/img33.jpg"
+
+times = "12:30"
+
++++
+
+# openGauss+KeepAlived
+
+## 实验环境
+
+操作系统： CentOS 7.6
+
+数据库版本： openGauss 1.1.0Primary
+
+主机/IP： opengaussdb1/192.168.1.11 \(openGauss主备已部署完毕\)
+
+Standby 主机/IP： opengaussdb2/192.168.1.12 \(openGauss主备已部署完毕\)
+
+> **说明：**
+>不建议在云环境\(如：华为云\)下搭建Keepalived进行测试，本人在云环境下测试发现，Keepalived的VIP无法在云环境下与其他主机通信，云环境下如何使用该VIP建议咨询云服务厂商。在踩坑之后，选择使用本地的VMWare workstation进行简单测试。
+
+## 安装KeepAlived软件
+
+```
+## 在所有节点执行安装
+yum install keepalived -y
+```
+
+## 配置keepalived
+
+> **说明：**
+>采用nopreempt不抢占VIP，主备节点的state均设置为BACKUP。
+
+- 主节点配置文件。
+
+```
+# vi /etc/keepalived/keepalived.conf
+--------------------------------------------
+! Configuration File for keepalived
+## 全局定义
+global_defs {
+ router_id Keepalived_openGauss #运行 keepalived 服务器的一个标识
+ script_user root #执行脚本的用户
+}
+
+## VRRP实例定义
+## 通常如果master服务Down掉后backup会变成master，但是当master服务又好了的时候 master此时会抢占VIP，这样就会发生两次数据库切换。
+## 建议使用nopreempt参数设置为非抢占模式，此时主库从故障中恢复后，不会从新的主库抢回VIP，但这需要将master和backup的state都设置成backup。
+vrrp_instance VI_1 {
+ state BACKUP #指定Keepalived的角色(BACKUP需大写)
+ interface eth0 #指定 HA 监测的网络接口
+ virtual_router_id 59 #虚拟路由的数字标识，同一个 vrrp_instance 下，MASTER 和 BACKUP 一致
+ nopreempt #非抢占模式，主库从故障中恢复后，不会从新的主库抢回VIP
+ priority 100 #优先级，备节点需要适当降低优先级
+ advert_int 1 #MASTER 和 BACKUP 负载均衡器同步检查的时间间隔（秒）
+ authentication { #设置验证码和验证类型
+ auth_type PASS
+ auth_pass 1111
+ }
+ virtual_ipaddress { #设置虚拟 IP 地址，可以设置多个，每个一行
+ 192.168.1.10
+ }
+}
+
+## 虚拟服务器定义
+virtual_server 192.168.1.10 26000 { #设置虚拟服务器的 IP 和端口，用空格隔开
+ delay_loop 6 #设置运行情况检查时间，单位是秒
+# lb_algo rr #负载调度算法(轮询)
+# lb_kind DR #负载均衡机制(NAT、TUN、DR)
+ persistence_timeout 50 #会话保持时间(秒)
+ protocol TCP #转发协议类型
+ real_server 192.168.1.11 26000 { #配置服务节点
+ weight 100 #配置服务节点的权重
+ notify_down /gauss/failoverdb.sh #故障响应脚本
+ TCP_CHECK { #使用TCP_CHECK方式进行健康检查
+ connect_timeout 10 #10秒无响应即超时
+ delay_before_retry 3 #重试间隔时间
+ }
+ }
+}
+```
+
+- 主节点故障切换脚本（仅适用openGauss进程崩溃故障处理，不适用Primary操作系统宕机故障处理）。
+
+```
+vi /gauss/failoverdb.sh
+--------------------------------------------
+#!/bin/bash
+echo "Start to failover openGauss database."
+pkill keepalived
+ssh 192.168.1.12 "su - omm -c 'gs_ctl failover -D /gauss/data/db1'"
+ssh 192.168.1.12 "su - omm -c 'gs_om -t refreshconf'"
+echo 'Failover operation is completed.'
+--------------------------------------------
+chmod 764 /gauss/failoverdb.sh
+```
+
+- 备节点配置文件。
+
+```
+# vi /etc/keepalived/keepalived.conf
+--------------------------------------------
+! Configuration File for keepalived
+## 全局定义
+global_defs {
+ router_id Keepalived_openGauss #运行 keepalived 服务器的一个标识
+ script_user root #执行脚本的用户
+}
+
+## VRRP实例定义
+## 通常如果master服务Down掉后backup会变成master，但是当master服务又好了的时候 master此时会抢占VIP，这样就会发生两次数据库切换。
+## 建议使用nopreempt参数设置为非抢占模式，此时主库从故障中恢复后，不会从新的主库抢回VIP，但这需要将master和backup的state都设置成backup。
+vrrp_instance VI_1 {
+ state BACKUP #指定Keepalived的角色(BACKUP需大写)
+ interface eth0 #指定 HA 监测的网络接口
+ virtual_router_id 59 #虚拟路由的数字标识，同一个 vrrp_instance 下，MASTER 和 BACKUP 一致
+ nopreempt #非抢占模式，主库从故障中恢复后，不会从新的主库抢回VIP
+ priority 60 #优先级，备节点需要适当降低优先级
+ advert_int 1 #MASTER 和 BACKUP 负载均衡器同步检查的时间间隔（秒）
+ authentication { #设置验证码和验证类型
+ auth_type PASS
+ auth_pass 1111
+ }
+ virtual_ipaddress { #设置虚拟 IP 地址，可以设置多个，每个一行
+ 192.168.1.10
+ }
+}
+
+## 虚拟服务器定义
+virtual_server 192.168.1.10 26000 { #设置虚拟服务器的 IP 和端口，用空格隔开
+ delay_loop 6 #设置运行情况检查时间，单位是秒
+# lb_algo rr #负载调度算法(轮询)
+# lb_kind DR #负载均衡机制(NAT、TUN、DR)
+ persistence_timeout 50 #会话保持时间(秒)
+ protocol TCP #转发协议类型
+ real_server 192.168.1.12 26000 { #配置服务节点
+ weight 60 #配置服务节点的权重
+ notify_down /gauss/failoverdb.sh #虚拟服务故障响应脚本
+ MISC_CHECK { ## 使用 MISC_CHECK 方式自定义脚本做健康检查
+ misc_path "/gauss/check.sh" ## 检测脚本
+ misc_timeout 10 ## 执行脚本的超时时间
+ misc_dynamic ## 根据退出状态码动态调整服务器的权重
+ }
+ }
+}
+--------------------------------------------
+## 备节点选择MISC_CHECK方式的原因：
+## 测试发现，当主节点直接断电宕机后，Keepalived的VIP会漂移至备节点，此时如果使用TCP_CHECK方式做健康检查，会因为备机可读的原因使得VIP:26000连接正常，造成keepalived健康检查的误判。
+## 最终导致主节点断电宕机后，备节点虽获取了VIP，但并没有执行openGauss的failover操作，备节点依旧只读，无法对外提供业务。
+## 为了纠正这一点，建议使用MISC_CHECK方式自定义脚本，登录主节点做数据库健康检查(简单示例脚本：/gauss/check.sh)
+```
+
+- 备节点健康检查脚本\[ ssh登录主节点进行数据库连接检查 \]。
+
+```
+vi /gauss/check.sh
+-------------------------------------------
+ssh 192.168.1.11 "su - omm -c \"gsql -d postgres -p 26000 -t -A -c 'select 1;'\""
+-------------------------------------------
+```
+
+- 备节点故障切换脚本。
+
+```
+vi /gauss/failoverdb.sh
+--------------------------------------------
+#!/bin/bash
+echo "Start to failover openGauss database."
+pkill keepalived
+su - omm -c "gs_ctl failover -D /gauss/data/db1"
+su - omm -c "gs_om -t refreshconf"
+echo 'Failover operation is completed.'
+--------------------------------------------
+chmod 764 /gauss/failoverdb.sh
+```
+
+## openGauss配置
+
+- 修改openGauss侦听地址。
+
+```
+$ gs_guc set -I all -N all -c "listen_addresses = '0.0.0.0'"
+$ gs_guc set -I all -N all -c "local_bind_address = '0.0.0.0'"
+```
+
+- 修改所有节点replconninfo参数（避免端口冲突）。
+
+```
+$ vi /gauss/data/db1/postgresql.conf
+--------------------------------------------
+修改：localport --> 26011
+修改：remoteport --> 26011
+--------------------------------------------
+```
+
+- 重启openGauss数据库,并检查服务器状态。
+
+```
+## 重启openGauss
+[omm@prod db1]$ gs_om -t stop && gs_om -t start
+
+## 检查openGauss状态
+[root@opengaussdb1 ~]# su - omm -c "gs_om -t status --detail"
+[ Cluster State ]
+cluster_state : Normal
+redistributing : No
+current_az : AZ_ALL
+[ Datanode State ]
+node node_ip instance state |
+-----------------------------------------------------------------------
+1 opengaussdb1 192.168.1.11 6001 /gauss/data/db1 P Primary Normal |
+2 opengaussdb2 192.168.1.12 6002 /gauss/data/db1 S Standby Normal
+
+## 检查KeepAlived进程状态
+[omm@opengaussdb1 ~]$ ps -ef|grep keep|grep -v grep
+root 15664 1 0 16:15 ? 00:00:00 /usr/sbin/keepalived -D
+root 15665 15664 0 16:15 ? 00:00:00 /usr/sbin/keepalived -D
+root 15666 15664 0 16:15 ? 00:00:00 /usr/sbin/keepalived -D
+
+## 检查VIP状态
+[root@opengaussdb1 ~]# ip a
+1: lo: mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
+ link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
+ inet 127.0.0.1/8 scope host lo
+ valid_lft forever preferred_lft forever
+ inet6 ::1/128 scope host
+ valid_lft forever preferred_lft forever
+2: ens33: mtu 1500 qdisc pfifo_fast state UP group default qlen 1000
+ link/ether 00:0c:29:da:60:c0 brd ff:ff:ff:ff:ff:ff
+ inet 192.168.1.11/24 brd 192.168.1.255 scope global noprefixroute ens33
+ valid_lft forever preferred_lft forever
+ inet 192.168.1.10/32 scope global ens33 ## VIP：192.168.1.10
+ valid_lft forever preferred_lft forever
+ inet6 2408:8270:237:ded0:c89c:adab:e7b:8bd6/64 scope global noprefixroute dynamic
+ valid_lft 258806sec preferred_lft 172406sec
+ inet6 fe80::c4f2:8ad1:200d:ce9b/64 scope link noprefixroute
+ valid_lft forever preferred_lft forever
+```
+
+## 故障模拟测试
+
+- 主节点\[192.168.1.11\]操作。
+
+```
+## kill数据库进程
+[root@opengaussdb1 ~]# ps -ef|grep gauss
+omm 18115 1 4 16:30 ? 00:00:35 /gauss/app/bin/gaussdb -D /gauss/data/db1 -M primary
+root 19254 9299 0 16:42 pts/0 00:00:00 grep --color=auto gauss
+[root@opengaussdb1 ~]# kill -9 18115
+
+## 检查message日志[检测到故障，执行notify_down脚本，并关闭keepalived服务]
+# tail -fn 200 /var/log/messages
+Feb 19 16:42:57 opengaussdb1 Keepalived_healthcheckers[18816]: TCP connection to [192.168.1.11]:26000 failed.
+Feb 19 16:43:00 opengaussdb1 Keepalived_healthcheckers[18816]: TCP connection to [192.168.1.11]:26000 failed.
+Feb 19 16:43:00 opengaussdb1 Keepalived_healthcheckers[18816]: Check on service [192.168.1.11]:26000 failed after 1 retry.
+Feb 19 16:43:00 opengaussdb1 Keepalived_healthcheckers[18816]: Removing service [192.168.1.11]:26000 from VS [192.168.1.10]:26000
+Feb 19 16:43:00 opengaussdb1 Keepalived_healthcheckers[18816]: IPVS (cmd 1160, errno 2): No such destination
+Feb 19 16:43:00 opengaussdb1 Keepalived_healthcheckers[18816]: Executing [/gauss/failoverdb.sh] for service [192.168.1.11]:26000 in VS [192.168.1.10]:26000
+Feb 19 16:43:00 opengaussdb1 Keepalived_healthcheckers[18816]: Lost quorum 1-0=1 > 0 for VS [192.168.1.10]:26000
+Feb 19 16:43:00 opengaussdb1 Keepalived[18815]: Stopping
+Feb 19 16:43:00 opengaussdb1 Keepalived_healthcheckers[18816]: pid 19258 exited due to signal 15
+Feb 19 16:43:00 opengaussdb1 Keepalived_vrrp[18817]: VRRP_Instance(VI_1) sent 0 priority
+Feb 19 16:43:00 opengaussdb1 Keepalived_vrrp[18817]: VRRP_Instance(VI_1) removing protocol VIPs.
+Feb 19 16:43:00 opengaussdb1 Keepalived_healthcheckers[18816]: IPVS (cmd 1156, errno 2): No such file or directory
+Feb 19 16:43:00 opengaussdb1 Keepalived_healthcheckers[18816]: Stopped
+Feb 19 16:43:01 opengaussdb1 Keepalived_vrrp[18817]: Stopped
+Feb 19 16:43:01 opengaussdb1 Keepalived[18815]: Stopped Keepalived v1.3.5 (03/19,2017), git commit v1.3.5-6-g6fa32f2
+```
+
+- 备节点\[192.168.1.12\]检查。
+
+```
+## 检查VIP是否已漂移
+[root@opengaussdb2 ~]# ip a|grep 192.168
+ inet 192.168.1.12/24 brd 192.168.1.255 scope global noprefixroute ens33
+ inet 192.168.1.10/32 scope global ens33
+
+## 检查数据库状态[已failover成为Primary]
+[omm@opengaussdb2 ~]$ gs_om -t status --detail
+[ Cluster State ]
+cluster_state : Degraded
+redistributing : No
+current_az : AZ_ALL
+[ Datanode State ]
+node node_ip instance state |
+---------------------------------------------------------------------------------
+1 opengaussdb1 192.168.1.11 6001 /gauss/data/db1 P Down Manually stopped |
+2 opengaussdb2 192.168.1.12 6002 /gauss/data/db1 S Primary Normal
+```
+