thinkphp在PostgreSQL下兼容ON CONFLICT DO NOTHING处理方法

由于thinkphp只对Mysql有INSERT IGNORE方法，PostgreSQL是ON CONFLICT  DO NOTHING方法，但是thinkphp orm一直没有支持，只能暂时在model用这个方法.

    public static function insertIgnore(array $data, string $conflictTarget = null): int
    {
        $rawSql = static::fetchSql(true)->insert($data);

        if ($conflictTarget) {
            $rawSql = "{$rawSql} ON CONFLICT {$conflictTarget}  DO NOTHING";
        }
        return Db::execute($rawSql);
    }

PostgreSQL插件pgclone 类似git版本控制方案.

前段时间看到老冯的公众号推送pgclone这玩意想搞来用一下 https://vonng.com/pg/pg-clone/ 

翻了下文档要求磁盘分区必须支持reflink才可以, 这就意味着ext4没法玩耍了, 必须XFS  Btrfs ZFS分区格式才能玩.

php类备忘录

最蠢的营销广告

用女人拍擦边性暗示的视频推广卖男装是我见过最愚蠢的精致营销。

买衣服要穿着合身和舒适， 其次才是好看。

即使我这种好色的老色批看这种营销内容都能感到满满的生理厌恶。

PostgreSQL18 主从同步复制部署备忘录

安装postgresql-18

apt install -y postgresql-common
/usr/share/postgresql-common/pgdg/apt.postgresql.org.sh
install -d /usr/share/postgresql-common/pgdg
curl -o /usr/share/postgresql-common/pgdg/apt.postgresql.org.asc --fail https://www.postgresql.org/media/keys/ACCC4CF8.asc
apt update
apt install postgresql-18

主库配置

USER="replica"
PASSWORD=$(openssl rand -base64 24)
NETWOKR="0.0.0.0/0"   //从库IP段
IP="1.1.1.1" #主库IP 

#主库
cat>>/etc/postgresql/18/main/postgresql.conf<<EOF
listen_addresses = '*'
wal_level = replica
max_wal_senders = 10
max_replication_slots = 10
wal_keep_size = 1GB
archive_mode = off
#archive_command = 'cp %p /var/lib/postgresql/wal_archive/%f'
EOF
#archive_mode归档模式很占空间
echo "host replication $USER $NETWOKR md5" >>/etc/postgresql/18/main/pg_hba.conf
# 创建用户（幂等）
sudo -u postgres psql -tc "SELECT 1 FROM pg_roles WHERE rolname='$USER'" | grep -q 1 || \
sudo -u postgres psql -c "CREATE ROLE $USER WITH REPLICATION LOGIN PASSWORD '$PASSWORD';"
systemctl restart postgresql
echo "REPL $USER PASSWORD: $PASSWORD"

echo -e "\n\n从库快速导入备份bash shell："
echo -e "USER=$USER\nPASSWORD=$PASSWORD"
echo "sudo -u postgres pg_basebackup -h $IP  -U  $USER -D /home/postgresql/18/main -P -R -X stream -v"

IP="1.1.1.1" #主库IP
#rm -rf /var/lib/postgresql/18/main/*   
cat>>/etc/postgresql/18/main/postgresql.conf<<EOF
listen_addresses = '*'
#recovery_min_apply_delay='24h'   #延迟24小时同步
hot_standby = on
wal_keep_size = 2GB
max_wal_senders = 10
EOF
systemctl stop postgresql
mkdir /home/postgresql
chown -R postgres:postgres /home/postgresql
chmod 700 /home/postgresql
sudo -u postgres pg_basebackup -h $IP -U $USER -D /home/postgresql/18/main -P -R -X stream -v
systemctl start postgresql

tail -n 100 /var/log/postgresql/postgresql-18-main.log
#验证主库
sudo -u postgres psql -c "SELECT client_addr,state FROM pg_stat_replication;"

#验证从库
sudo -u postgres psql -c "SELECT pg_is_in_recovery();"

dnswatch

https://www.dnswatch.info/dns/dnslookup?la=en&host=164.210.103.in-addr.arpa&type=NS&submit=Resolve

一个在线验证rpki网站

https://bgprpki.com/

youtube的Shorts就是shift.........

两三年前刷过，老是重复出现看过的视频， 现在还是一点都没变。。。。。。

真是shi。。。。。。。。 shift

workerman和swoole单文件静态php

worker和swoole直接常驻内存运行抛弃fpm直接高效运行， 官方提供了现成的静态编译php文件，拿来直接可以运行，不用再自己吭哧吭哧再去编译，直接带上nginx就可以跑，如果将php打包成二进制文件会更简单。

几个相关项目发布地址

https://github.com/swoole/swoole-cli/releases
https://github.com/dixyes/phpmicro
https://github.com/crazywhalecc/static-php-cli

https://www.swoole.com/download
https://www.workerman.net/download

swoole-cli静态php

wget -O - https://github.com/swoole/swoole-cli/releases/download/v6.1.4.0/swoole-cli-v6.1.4-linux-x64.tar.xz | tar -xJ 
install swoole-cli /usr/bin/swoole

workerman官方静态php

wget https://www.workerman.net/download/php/php-8.4-linux-x86_64.tar.gz|  tar -xz

phpmicro可以打包php为二进制文件

swoole启动fpm

swoole-cli -P -h
# 运行 FPM
swoole-cli -P --fpm-config /opt/php-8.1/etc/php-fpm.conf -p /opt/php-8.1/var
# 关闭守护进程
swoole-cli -P --fpm-config /opt/php-8.1/etc/php-fpm.conf -p /opt/php-8.1/var -F
# 使用 root 账户启动
swoole-cli -P --fpm-config /opt/php-8.1/etc/php-fpm.conf -p /opt/php-8.1/var -F -R

发现两个新的邮局方案mox和stalwart

https://github.com/stalwartlabs/stalwart/releases

https://github.com/mjl-/mox/releases

PostgreSQL查找慢查询

统计CPU

ps -u postgres -o %cpu | awk '{s+=$1} END {print "Postgres Total CPU: " s "%"}'

查看活跃查询

SELECT pid, now() - query_start AS duration, query, state FROM pg_stat_activity WHERE state = 'active' ORDER BY duration DESC;
SELECT pid, now() - query_start AS duration, query, state FROM pg_stat_activity WHERE state = 'active' ORDER BY duration DESC LIMIT 5;

查看连接IP和查询语句

SELECT pid, now() - query_start AS 执行时间, usename AS 用户名, 
client_addr AS 连接IP, state AS 状态, query AS 查询语句         
FROM pg_stat_activity WHERE
 pid <> pg_backend_pid() ORDER BY 执行时间 DESC;

查看扫表

SELECT 
    relname AS 表名,
    seq_scan AS 全表扫描次数,
    seq_tup_read AS 全表扫描读取行数,
    idx_scan AS 索引扫描次数,
    idx_tup_fetch AS 索引提取行数
FROM pg_stat_user_tables
WHERE seq_scan > 0
ORDER BY seq_tup_read DESC
LIMIT 10;

启用插件

CREATE EXTENSION IF NOT EXISTS pg_stat_statements; 

 重置自动增续ID

SELECT setval('tag_id_seq', (SELECT MAX(id) FROM tag));

nginx隐藏真实文件路径

给浏览器浏览器返回头

$file=decode($url);
header("X-Accel-Redirect: /privatefiles/$file");

nginx配置

    location /privatefiles {
        internal;  #内部访问
        alias /var/www/html/xaccel/restrictedfiles;
    }   

将文件名url加密， 然后接收解密，通过X-Accel-Redirect再转给nginx返回数据。

为什么要这么做？  一些文件路径在某些原因下不想暴露真实路径，又不通过程序再去读一遍问题直接转交个nginx响应。

转载自：https://dev.to/gbhorwood/nginx-serving-private-files-with-x-accel-redirect-57dl

Thinkphp使用postgresql新库报错解决

DROP FUNCTION IF EXISTS public.table_msg(varchar) CASCADE;
DROP FUNCTION IF EXISTS public.table_msg(varchar, varchar) CASCADE;
DROP FUNCTION IF EXISTS public.pgsql_type(varchar) CASCADE;
DROP TYPE IF EXISTS public.tablestruct CASCADE;

CREATE OR REPLACE FUNCTION public.pgsql_type(a_type character varying) RETURNS character varying AS
$BODY$
DECLARE
     v_type character varying;
BEGIN
     CASE a_type
         WHEN 'int8' THEN
              v_type := 'bigint';
         WHEN 'int4' THEN
              v_type := 'integer';
         WHEN 'int2' THEN
              v_type := 'smallint';
         WHEN 'bpchar' THEN
              v_type := 'char';
         ELSE
              v_type := a_type;
     END CASE;
     RETURN v_type;
END;
$BODY$
LANGUAGE plpgsql;

-- 2. tablestruct 类型
CREATE TYPE public.tablestruct AS (
  fields_key_name varchar(100),
  fields_name varchar(200),
  fields_type varchar(20),
  fields_length bigint,
  fields_not_null varchar(10),
  fields_default varchar(500),
  fields_comment varchar(1000)
);

-- 3. table_msg 函数（带 schema 参数）
CREATE OR REPLACE FUNCTION public.table_msg(a_schema_name varchar, a_table_name varchar)
RETURNS SETOF public.tablestruct AS
$body$
DECLARE
     v_ret tablestruct;
     v_oid oid;
     v_sql varchar;
     v_rec record;
     v_key varchar;
BEGIN
     -- 获取表 oid
     SELECT c.oid INTO v_oid
     FROM pg_class c
     INNER JOIN pg_namespace n ON c.relnamespace = n.oid AND lower(n.nspname) = lower(a_schema_name)
     WHERE c.relname = a_table_name;

     IF NOT FOUND THEN
         RETURN;
     END IF;

     -- 构造动态查询
     v_sql := '
     SELECT
           a.attname AS fields_name,
           a.attnum AS fields_index,
           pgsql_type(t.typname::varchar) AS fields_type,
           a.atttypmod - 4 AS fields_length,
           CASE WHEN a.attnotnull THEN ''not null'' ELSE '''' END AS fields_not_null,
           pg_get_expr(ad.adbin, ad.adrelid) AS fields_default,
           d.description AS fields_comment
     FROM pg_attribute a
     INNER JOIN pg_class c ON a.attrelid = c.oid
     INNER JOIN pg_type t ON a.atttypid = t.oid
     LEFT JOIN pg_attrdef ad ON ad.adrelid = c.oid AND ad.adnum = a.attnum
     LEFT JOIN pg_description d ON d.objoid = c.oid AND d.objsubid = a.attnum
     WHERE a.attnum > 0 AND a.attisdropped = false AND c.oid = ' || v_oid || '
     ORDER BY a.attnum;
     ';

     -- 遍历字段
     FOR v_rec IN EXECUTE v_sql LOOP
         v_ret.fields_name := v_rec.fields_name;
         v_ret.fields_type := v_rec.fields_type;
         IF v_rec.fields_length > 0 THEN
             v_ret.fields_length := v_rec.fields_length;
         ELSE
             v_ret.fields_length := NULL;
         END IF;
         v_ret.fields_not_null := v_rec.fields_not_null;
         v_ret.fields_default := v_rec.fields_default;
         v_ret.fields_comment := v_rec.fields_comment;

         SELECT constraint_name INTO v_key
         FROM information_schema.key_column_usage
         WHERE table_schema = a_schema_name
           AND table_name = a_table_name
           AND column_name = v_rec.fields_name;

         IF FOUND THEN
             v_ret.fields_key_name := v_key;
         ELSE
             v_ret.fields_key_name := '';
         END IF;

         RETURN NEXT v_ret;
     END LOOP;

     RETURN;
END;
$body$
LANGUAGE plpgsql VOLATILE CALLED ON NULL INPUT SECURITY INVOKER;

COMMENT ON FUNCTION public.table_msg(a_schema_name varchar, a_table_name varchar) IS '获得表信息';

-- 4. table_msg 函数（简化版，只传 table_name）
CREATE OR REPLACE FUNCTION public.table_msg(a_table_name varchar)
RETURNS SETOF public.tablestruct AS
$body$
DECLARE
    v_ret tablestruct;
BEGIN
    FOR v_ret IN SELECT * FROM public.table_msg('public', a_table_name) LOOP
        RETURN NEXT v_ret;
    END LOOP;
    RETURN;
END;
$body$
LANGUAGE plpgsql VOLATILE CALLED ON NULL INPUT SECURITY INVOKER;

COMMENT ON FUNCTION public.table_msg(a_table_name varchar) IS '获得表信息';

rhel系更新启动内核

dracut --force #强制更新initramfs
#dracut --force /boot/initramfs-$(uname -r).img $(uname -r)
#grubby --set-default /boot/vmlinuz-5.19.0-ml.x86_64   #设置指定内核
grubby --set-default 0         #设置启动第一个内核
grubby --default-kernel       #查看启动内核

初学golang之用beego快速创建项目

现安装golang https://go.dev/dl/

go install github.com/beego/bee/v2@latest
export PATH=$PATH:$(go env GOPATH)/bin
bee new app
cd app
go mod tidy
bee run