【监控】Linux部署postgres_exporter及PG配置(非Docker)
目录
- 一、下载及部署
- 二、postgres_exporter配置
- 1. 停止脚本stop.sh
- 2. 启动脚本start.sh
- 3. queries.yaml
- 三、PostgreSQL数据库配置
- 1. 修改postgresql.conf配置文件
- 2. 创建用户、表、扩展等
- 四、参考
一、下载及部署
下载地址
选一个amd64下载
上传至服务器,解压
tax -xvf postgres_exporter-0.11.1.linux-amd64.tar.gz
进入解压后的目录
二、postgres_exporter配置
1. 停止脚本stop.sh
建立停止脚本 stop.sh 。注意unix编码
#!/bin/sh
echo "stop"
#!/bin/bashPID=$(ps -ef | grep postgres_exporter | grep -v grep | awk '{ print $2 }')
if [ "${PID}" ]
thenecho 'Application is stpping...'echo kill $PID DONEkill $PID
elseecho 'Application is already stopped...'
fi
2. 启动脚本start.sh
启动脚本start.sh
-
后面会建立postgres_exporter用户,密码为password
-
–web.listen-address为监听的端口
-
–extend.query-path为自定义查询的文件
sh stop.sh
export DATA_SOURCE_NAME=postgresql://postgres_exporter:password@数据库IP:数据库端口/postgres?sslmode=disablenohup ./postgres_exporter --web.listen-address=0.0.0.0:8001 --extend.query-path=queries.yaml >nohup.out 2>&1 &
3. queries.yaml
pg_replication:query: "SELECT CASE WHEN NOT pg_is_in_recovery() THEN 0 ELSE GREATEST (0, EXTRACT(EPOCH FROM (now() - pg_last_xact_replay_timestamp()))) END AS lag"master: truemetrics:- lag:usage: "GAUGE"description: "Replication lag behind master in seconds"pg_postmaster:query: "SELECT pg_postmaster_start_time as start_time_seconds from pg_postmaster_start_time()"master: truemetrics:- start_time_seconds:usage: "GAUGE"description: "Time at which postmaster started"pg_stat_user_tables:query: |SELECTcurrent_database() datname,schemaname,relname,seq_scan,seq_tup_read,idx_scan,idx_tup_fetch,n_tup_ins,n_tup_upd,n_tup_del,n_tup_hot_upd,n_live_tup,n_dead_tup,n_mod_since_analyze,COALESCE(last_vacuum, '1970-01-01Z') as last_vacuum,COALESCE(last_autovacuum, '1970-01-01Z') as last_autovacuum,COALESCE(last_analyze, '1970-01-01Z') as last_analyze,COALESCE(last_autoanalyze, '1970-01-01Z') as last_autoanalyze,vacuum_count,autovacuum_count,analyze_count,autoanalyze_countFROMpg_stat_user_tablesmetrics:- datname:usage: "LABEL"description: "Name of current database"- schemaname:usage: "LABEL"description: "Name of the schema that this table is in"- relname:usage: "LABEL"description: "Name of this table"- seq_scan:usage: "COUNTER"description: "Number of sequential scans initiated on this table"- seq_tup_read:usage: "COUNTER"description: "Number of live rows fetched by sequential scans"- idx_scan:usage: "COUNTER"description: "Number of index scans initiated on this table"- idx_tup_fetch:usage: "COUNTER"description: "Number of live rows fetched by index scans"- n_tup_ins:usage: "COUNTER"description: "Number of rows inserted"- n_tup_upd:usage: "COUNTER"description: "Number of rows updated"- n_tup_del:usage: "COUNTER"description: "Number of rows deleted"- n_tup_hot_upd:usage: "COUNTER"description: "Number of rows HOT updated (i.e., with no separate index update required)"- n_live_tup:usage: "GAUGE"description: "Estimated number of live rows"- n_dead_tup:usage: "GAUGE"description: "Estimated number of dead rows"- n_mod_since_analyze:usage: "GAUGE"description: "Estimated number of rows changed since last analyze"- last_vacuum:usage: "GAUGE"description: "Last time at which this table was manually vacuumed (not counting VACUUM FULL)"- last_autovacuum:usage: "GAUGE"description: "Last time at which this table was vacuumed by the autovacuum daemon"- last_analyze:usage: "GAUGE"description: "Last time at which this table was manually analyzed"- last_autoanalyze:usage: "GAUGE"description: "Last time at which this table was analyzed by the autovacuum daemon"- vacuum_count:usage: "COUNTER"description: "Number of times this table has been manually vacuumed (not counting VACUUM FULL)"- autovacuum_count:usage: "COUNTER"description: "Number of times this table has been vacuumed by the autovacuum daemon"- analyze_count:usage: "COUNTER"description: "Number of times this table has been manually analyzed"- autoanalyze_count:usage: "COUNTER"description: "Number of times this table has been analyzed by the autovacuum daemon"pg_statio_user_tables:query: "SELECT current_database() datname, schemaname, relname, heap_blks_read, heap_blks_hit, idx_blks_read, idx_blks_hit, toast_blks_read, toast_blks_hit, tidx_blks_read, tidx_blks_hit FROM pg_statio_user_tables"metrics:- datname:usage: "LABEL"description: "Name of current database"- schemaname:usage: "LABEL"description: "Name of the schema that this table is in"- relname:usage: "LABEL"description: "Name of this table"- heap_blks_read:usage: "COUNTER"description: "Number of disk blocks read from this table"- heap_blks_hit:usage: "COUNTER"description: "Number of buffer hits in this table"- idx_blks_read:usage: "COUNTER"description: "Number of disk blocks read from all indexes on this table"- idx_blks_hit:usage: "COUNTER"description: "Number of buffer hits in all indexes on this table"- toast_blks_read:usage: "COUNTER"description: "Number of disk blocks read from this table's TOAST table (if any)"- toast_blks_hit:usage: "COUNTER"description: "Number of buffer hits in this table's TOAST table (if any)"- tidx_blks_read:usage: "COUNTER"description: "Number of disk blocks read from this table's TOAST table indexes (if any)"- tidx_blks_hit:usage: "COUNTER"description: "Number of buffer hits in this table's TOAST table indexes (if any)"# WARNING: This set of metrics can be very expensive on a busy server as every unique query executed will create an additional time series
pg_stat_statements:query: "SELECT t2.rolname, t3.datname, queryid, calls, total_time / 1000 as total_time_seconds, min_time / 1000 as min_time_seconds, max_time / 1000 as max_time_seconds, mean_time / 1000 as mean_time_seconds, stddev_time / 1000 as stddev_time_seconds, rows, shared_blks_hit, shared_blks_read, shared_blks_dirtied, shared_blks_written, local_blks_hit, local_blks_read, local_blks_dirtied, local_blks_written, temp_blks_read, temp_blks_written, blk_read_time / 1000 as blk_read_time_seconds, blk_write_time / 1000 as blk_write_time_seconds FROM pg_stat_statements t1 JOIN pg_roles t2 ON (t1.userid=t2.oid) JOIN pg_database t3 ON (t1.dbid=t3.oid) WHERE t2.rolname != 'rdsadmin'"master: truemetrics:- rolname:usage: "LABEL"description: "Name of user"- datname:usage: "LABEL"description: "Name of database"- queryid:usage: "LABEL"description: "Query ID"- calls:usage: "COUNTER"description: "Number of times executed"- total_time_seconds:usage: "COUNTER"description: "Total time spent in the statement, in milliseconds"- min_time_seconds:usage: "GAUGE"description: "Minimum time spent in the statement, in milliseconds"- max_time_seconds:usage: "GAUGE"description: "Maximum time spent in the statement, in milliseconds"- mean_time_seconds:usage: "GAUGE"description: "Mean time spent in the statement, in milliseconds"- stddev_time_seconds:usage: "GAUGE"description: "Population standard deviation of time spent in the statement, in milliseconds"- rows:usage: "COUNTER"description: "Total number of rows retrieved or affected by the statement"- shared_blks_hit:usage: "COUNTER"description: "Total number of shared block cache hits by the statement"- shared_blks_read:usage: "COUNTER"description: "Total number of shared blocks read by the statement"- shared_blks_dirtied:usage: "COUNTER"description: "Total number of shared blocks dirtied by the statement"- shared_blks_written:usage: "COUNTER"description: "Total number of shared blocks written by the statement"- local_blks_hit:usage: "COUNTER"description: "Total number of local block cache hits by the statement"- local_blks_read:usage: "COUNTER"description: "Total number of local blocks read by the statement"- local_blks_dirtied:usage: "COUNTER"description: "Total number of local blocks dirtied by the statement"- local_blks_written:usage: "COUNTER"description: "Total number of local blocks written by the statement"- temp_blks_read:usage: "COUNTER"description: "Total number of temp blocks read by the statement"- temp_blks_written:usage: "COUNTER"description: "Total number of temp blocks written by the statement"- blk_read_time_seconds:usage: "COUNTER"description: "Total time the statement spent reading blocks, in milliseconds (if track_io_timing is enabled, otherwise zero)"- blk_write_time_seconds:usage: "COUNTER"description: "Total time the statement spent writing blocks, in milliseconds (if track_io_timing is enabled, otherwise zero)"pg_process_idle:query: |WITHmetrics AS (SELECTapplication_name,SUM(EXTRACT(EPOCH FROM (CURRENT_TIMESTAMP - state_change))::bigint)::float AS process_idle_seconds_sum,COUNT(*) AS process_idle_seconds_countFROM pg_stat_activityWHERE state = 'idle'GROUP BY application_name),buckets AS (SELECTapplication_name,le,SUM(CASE WHEN EXTRACT(EPOCH FROM (CURRENT_TIMESTAMP - state_change)) <= leTHEN 1ELSE 0END)::bigint AS bucketFROMpg_stat_activity,UNNEST(ARRAY[1, 2, 5, 15, 30, 60, 90, 120, 300]) AS leGROUP BY application_name, leORDER BY application_name, le)SELECTapplication_name,process_idle_seconds_sum as seconds_sum,process_idle_seconds_count as seconds_count,ARRAY_AGG(le) AS seconds,ARRAY_AGG(bucket) AS seconds_bucketFROM metrics JOIN buckets USING (application_name)GROUP BY 1, 2, 3metrics:- application_name:usage: "LABEL"description: "Application Name"- seconds:usage: "HISTOGRAM"description: "Idle time of server processes"
三、PostgreSQL数据库配置
1. 修改postgresql.conf配置文件
- 先根据命令在服务器上找到配置文件在哪
find / -name postgresql.conf
- 修改配置文件postgresql.conf,添加下面三行
shared_preload_libraries = 'pg_stat_statements'
pg_stat_statements.max = 1000
pg_stat_statements.track = all
- 重启pg服务(pg不同安装方式启动方式可能不同)
pg_ctl restart
2. 创建用户、表、扩展等
官网的文档里提示比pg10高或低版本的数据库执行的SQL不同,但是我pg11只执行高版本的SQL失败了。最终高低版本都执行成功。
最好在postgres库下的public模式执行
版本>=10的pg,以下三段SQL都要执行
-- To use IF statements, hence to be able to check if the user exists before
-- attempting creation, we need to switch to procedural SQL (PL/pgSQL)
-- instead of standard SQL.
-- More: https://www.postgresql.org/docs/9.3/plpgsql-overview.html
-- To preserve compatibility with <9.0, DO blocks are not used; instead,
-- a function is created and dropped.
CREATE OR REPLACE FUNCTION __tmp_create_user() returns void as $$
BEGINIF NOT EXISTS (SELECT -- SELECT list can stay empty for thisFROM pg_catalog.pg_userWHERE usename = 'postgres_exporter') THENCREATE USER postgres_exporter;END IF;
END;
$$ language plpgsql;SELECT __tmp_create_user();
DROP FUNCTION __tmp_create_user();ALTER USER postgres_exporter WITH PASSWORD 'password';
ALTER USER postgres_exporter SET SEARCH_PATH TO postgres_exporter,pg_catalog;-- If deploying as non-superuser (for example in AWS RDS), uncomment the GRANT
-- line below and replace <MASTER_USER> with your root user.
-- GRANT postgres_exporter TO <MASTER_USER>;GRANT CONNECT ON DATABASE postgres TO postgres_exporter;
GRANT pg_monitor to postgres_exporter;
版本<10的pg,只执行下面的SQL
CREATE SCHEMA IF NOT EXISTS postgres_exporter;
GRANT USAGE ON SCHEMA postgres_exporter TO postgres_exporter;CREATE OR REPLACE FUNCTION get_pg_stat_activity() RETURNS SETOF pg_stat_activity AS
$$ SELECT * FROM pg_catalog.pg_stat_activity; $$
LANGUAGE sql
VOLATILE
SECURITY DEFINER;CREATE OR REPLACE VIEW postgres_exporter.pg_stat_activity
ASSELECT * from get_pg_stat_activity();GRANT SELECT ON postgres_exporter.pg_stat_activity TO postgres_exporter;CREATE OR REPLACE FUNCTION get_pg_stat_replication() RETURNS SETOF pg_stat_replication AS
$$ SELECT * FROM pg_catalog.pg_stat_replication; $$
LANGUAGE sql
VOLATILE
SECURITY DEFINER;CREATE OR REPLACE VIEW postgres_exporter.pg_stat_replication
ASSELECT * FROM get_pg_stat_replication();GRANT SELECT ON postgres_exporter.pg_stat_replication TO postgres_exporter;CREATE EXTENSION IF NOT EXISTS pg_stat_statements;
CREATE OR REPLACE FUNCTION get_pg_stat_statements() RETURNS SETOF pg_stat_statements AS
$$ SELECT * FROM public.pg_stat_statements; $$
LANGUAGE sql
VOLATILE
SECURITY DEFINER;CREATE OR REPLACE VIEW postgres_exporter.pg_stat_statements
ASSELECT * FROM get_pg_stat_statements();GRANT SELECT ON postgres_exporter.pg_stat_statements TO postgres_exporter;
来到postgres_exporter安装目录,启动postgres_exporter
sh start.sh
观察nohup.out文件,看是否有报错信息。
如果集成了Grafana,可以发现页面已经能采集到数据了
Grafana+prometheus+postgres_exporter参考
四、参考
- Github地址
- postgres_exporter使用过程中的注意事项
相关文章:
【监控】Linux部署postgres_exporter及PG配置(非Docker)
目录一、下载及部署二、postgres_exporter配置1. 停止脚本stop.sh2. 启动脚本start.sh3. queries.yaml三、PostgreSQL数据库配置1. 修改postgresql.conf配置文件2. 创建用户、表、扩展等四、参考一、下载及部署 下载地址 选一个amd64下载 上传至服务器,解压 tax…...
基于Java+SpringBoot+Vue+Uniapp(有教程)前后端分离健身预约系统设计与实现
博主介绍:✌全网粉丝3W,全栈开发工程师,从事多年软件开发,在大厂呆过。持有软件中级、六级等证书。可提供微服务项目搭建与毕业项目实战✌ 博主作品:《微服务实战》专栏是本人的实战经验总结,《Spring家族及…...
【2023】DevOps、SRE、运维开发面试宝典之Redis相关面试题
文章目录 1、redis主从复制原理2、redis哨兵模式的原理3、reids集群原理4、Redis 哈希表进行的触发时机是什么?5、Redis 的 RDB 和 AOF 机制各自的优缺点是什么?这两种机制是否可以混合使用?6、Redis 经常被称为单线程的系统,你如何理解 Redis 的单线程模型7、redis 的事务…...
十五、MyBatis使用PageHelper
1.limit分页 limit分页原理 mysql的limit后面两个数字: 第一个数字:startIndex(起始下标。下标从0开始。) 第二个数字:pageSize(每页显示的记录条数) 假设已知页码pageNum,还有每页…...
【MySQL】B+ 树索引
一、索引是什么 ? 为什么需要索引 ? 索引就是目录,目录就是索引。 索引从 InnoDB 存储引擎数据存储结构上来看,就是为各个页建立的目录。保证我们在查询时,可以通过二分法快速定位到页,再在页内通过二分法…...
Android Gradle Plugin Version 和 Gradle Version 的对应关系
官网参考 以下是插件版本和Gradle 版本对应关系: 插件版本所需的最低 Gradle 版本Android Gradle Plugin VersionGradle Version1.0.0 - 1.1.32.2.1 - 2.31.2.0 - 1.3.12.2.1 - 2.91.5.02.2.1 - 2.132.0.0 - 2.1.22.10 - 2.132.1.3 - 2.2.32.14.1 - 3.52.3.03.33.0…...
更多单词/词组/短语补充和总结(二)
auto 美 /[ˈɔːtoʊ] n.汽车adj.与汽车有关的,汽车的。不要记成“自动的” mobile 美 /[ˈmoʊbl] adj.可移动的;流动的;不要记成“手机”,手机是mobile phone automobile 美 /[ˈɔːtəməbiːl] n.汽车adj.自动的 automatic 美 /[ˌɔːtəˈmtɪk]…...
HEC-HMS和HEC-RAS快速入门、防洪评价报告编制及洪水建模、洪水危险性评价等应用
目录 ①HEC-RAS一维、二维建模方法及实践技术应用 ②HEC-HMS水文模型实践技术应用 ③新导则下的防洪评价报告编制方法及洪水建模实践技术应用 ④基于ArcGIS水文分析、HEC-RAS模拟技术在洪水危险性及风险评估 ⑤山洪径流过程模拟及洪水危险性评价 ①HEC-RAS一维、二维建模方…...
全面了解 B 端产品设计 — 基础扫盲篇
在今天,互联网的影响力与作用与日俱增,除了我们日常生活领域的改变以外,对于商业领域的渗透也见效颇丰。 越来越多的企业开始使用数字化的解决方案来助力企业发展,包括日常管理、运营、统计等等。或者通过互联网的方式开发出新的业务形态,进行产业升级,如这几年风头正劲的…...
顺序表(增删查改)
目录一、什么是顺序表二、顺序表的增删查改2.1 结构体的声明2.2 顺序表的初始化2.3 顺序表检查容量2.4 顺序表尾部插入数据2.5 顺序表头部插入数据2.6 顺序表尾部删除数据2.7 顺序表头部删除数据2.8 顺序表查找数据2.9 顺序表任意位置插入数据2.10 顺序表任意位置删除数据2.11 …...
一款优秀的低代码开发平台是什么样的?
目录 一、一款优秀的低代码平台应该是什么样的? 二、低代码核心能力 01、全栈可视化编程: 02、全生命周期管理: 03、低代码扩展能力: 三、小结 一、一款优秀的低代码平台应该是什么样的? 从企业角度来说&#x…...
ElasticSearch 学习笔记总结(四)
文章目录一、ES继承 Spring Data 框架二、SpringData 功能集成三、ES SpringData 文档搜索四、ES 优化 硬件选择五、ES 优化 分片策略六、ES 优化 路由选择七、ES 优化 写入速度优化七、ES 优化 内存设置八、ES 优化 重要配置一、ES继承 Spring Data 框架 Spring Data 是一个用…...
HDFS文件块大小
HDFS中的文件在物理上是分块存储(Block),块的大小可以通过配置参数(dfs.blocksize)来规定,默认大小在Hadooop2X版本中是128M,老版本中是64M。 思考:为什么块的大小不能设置太小&…...
C++——优先级队列(priority_queue)的使用及实现
目录 一.priority_queue的使用 1.1、基本介绍 1.2、优先级队列的定义 1.3、基本操作(常见接口的使用) 1.4、重写仿函数支持自定义数据类型 二.priority_queue的模拟实现 2.1、构造&&重要的调整算法 2.2、常见接口的实现 push() pop() top() empt…...
Linux学习记录——십일 环境变量
文章目录1、认识2、通过代码获取环境变量1、手动获取2、函数获取3、重新认识环境变量1、认识 在云服务器上写程序时,最终的执行需要./文件名,点表示当前目录,/是文件分隔符,之后就会打印程序,这是用户的操作ÿ…...
【人工智能 Open AI 】我们程序员真的要下岗了- 全能写Go / C / Java / C++ / Python / JS 人工智能机器人
文章目录[toc]人工智能 AI Code 写代码测试用golang实现冒泡排序用golang实现计算环比函数goroutine and channel用golang实现二叉树遍历代码用golang实现线程安全的HashMap操作代码using C programming language write a tiny Operation Systemuse C language write a tiny co…...
STM32 EXTI外部中断
本文代码使用 HAL 库。 文章目录前言一、什么是外部中断?二、外部中断中断线三、STM32F103的引脚复用四、相关函数:总结前言 一、什么是外部中断? 外部中断 是单片机实时地处理外部事件的一种内部机制。当某种外部事件发生时,单片…...
Mapper代理开发——书接MaBatis的简单使用
在这个mybatis的普通使用中依旧存在硬编码问题,虽然静态语句比原生jdbc都写更少了但是还是要写,Mapper就是用来解决原生方式中的硬编码还有简化后期执行SQL UserMapper是一个接口,里面有很多方法,都是一一和配置文件里面的sql语句的id名称所对…...
实体对象说明
1.工具类层Utilutil 工具顾明思义,util层就是存放工具类的地方,对于一些独立性很高的小功能,或重复性很高的代码片段,可以提取出来放到Util层中。2.数据层POJO对象(概念比较大) 包含了以下POJO plain ord…...
JAVA中加密与解密
BASE64加密/解密 Base64 编码会将字符串编码得到一个含有 A-Za-z0-9/ 的字符串。标准的 Base64 并不适合直接放在URL里传输,因为URL编码器会把标准 Base64 中的“/”和“”字符变为形如 “%XX” 的形式,而这些 “%” 号在存入数据库时还需要再进行转换&…...
改进YOLO系列 | ICLR2022 | OMNI-DIMENSIONAL DYNAMIC CONVOLUTION: 全维动态卷积
单个静态卷积核是现代卷积神经网络(CNNs)的常见训练范式。然而,最近的动态卷积研究表明,学习加权为其输入依赖注意力的n个卷积核的线性组合可以显著提高轻量级CNNs的准确性,同时保持高效的推理。然而,我们观察到现有的作品通过卷积核空间的一个维度(关于卷积核数量)赋予…...
信息收集之Github搜索语法
信息收集之Github搜索语法1.Github的搜索语法2.使用 Github 进行邮件配置信息收集3.使用Github进行数据库信息收集4.使用Github进行 SVN 信息收集5.使用Github进行综合信息收集在测试的信息收集阶段,可以去Github和码云上搜索与目标有关的信息,或者就有意…...
【案例教程】拉格朗日粒子扩散模式FLEXPART
拉格朗日粒子扩散模式FLEXPART通过计算点、线、面或体积源释放的大量粒子的轨迹,来描述示踪物在大气中长距离、中尺度的传输、扩散、干湿沉降和辐射衰减等过程。该模式既可以通过时间的前向运算来模拟示踪物由源区向周围的扩散,也可以通过后向运算来确定…...
试题 算法训练 自行车停放
问题描述 有n辆自行车依次来到停车棚,除了第一辆自行车外,每辆自行车都会恰好停放在已经在停车棚里的某辆自行车的左边或右边。(e.g.停车棚里已经有3辆自行车,从左到右编号为:3,5,1。现在编号为2的第4辆自行车要停在5号自行车的左…...
泛型与Map接口
Java学习之道 泛型 泛型这种参数类型可以用在类、方法和接口中,分别被称为泛型类,泛型方法,泛型接口 参数化类型:将类型由原来的具体的类型参数化,在使用/调用时传入具体的类型JDK5引入特性提供了安全检测机制…...
Unity Bug记录本
//个人记录,持续更新 1、将此代码挂载到空脚本上: bool flag (object)GetComponent<Camera>() null; bool flag1 (object)GetComponent<Text>() null; Debug.Log(flag"::"flag1); //输出结果:False::True bool…...
B. The Number of Products)厉害
You are given a sequence a1,a2,…,ana1,a2,…,an consisting of nn non-zero integers (i.e. ai≠0ai≠0). You have to calculate two following values: the number of pairs of indices (l,r)(l,r) (l≤r)(l≤r) such that al⋅al1…ar−1⋅aral⋅al1…ar−1⋅ar is neg…...
一起Talk Android吧(第五百一十二回:自定义Dialog)
文章目录整体思路实现方法第一步第二步第三步第四步各位看官们大家好,上一回中咱们说的例子是"自定义Dialog主题",这一回中咱们说的例子是" 自定义Dialog"。闲话休提,言归正转, 让我们一起Talk Android吧!整体…...
GinVueAdmin源码分析3-整合MySQL
目录文件结构数据库准备配置文件处理config.godb_list.gogorm_mysql.gosystem.go初始化数据库gorm.gogorm_mysql.go开始初始化测试数据库定义实体类 Userserviceapi开始测试!文件结构 本文章将使用到上一节创建的 CommonService 接口,用于测试连接数据库…...
大数据框架之Hadoop:MapReduce(三)MapReduce框架原理——MapReduce开发总结
在编写MapReduce程序时,需要考虑如下几个方面: 1、输入数据接口:InputFormat 默认使用的实现类是:TextInputFormatTextInputFormat的功能逻辑是:一次读一行文本,然后将该行的起始偏移量作为key࿰…...
鸿鹄网站建设/各大网站排名
本文目标是充当使用Selenium Docker镜像以及CodeceptJS和Express服务器的“操作方法”指南。其中,我们将涵盖:什么是E2E验收测试?为什么要使用Docker?松耦合的测试工具测试工具层创建测试项目E2E验收测试验收测试是典型软件开发过…...
河北做网站找谁/网络推广公司联系方式
jvm参数配置解析 参数说明-Xmx2048M JVM最大堆内存 -Xms2048M JVM最初始堆内存 -Xmn256MJVM年轻代大小(整个堆大小年轻代大小 年老代大小 持久代大小 。持久代一般固定大小为64m)-XX:MaxMetaspaceSize256MMaxMetaspaceSize是没有上限的,最…...
浙江网站建设品牌设计/企业品牌推广
腾讯云存储团队正式发布数据加速器 GooseFS 1.4 版本(含 GooseFS 1.4.0 和 GooseFS 1.4.1 版本),该版本针对 AI、大数据场景提供了文件解压缩等便捷易用的工具,同时针对海量文件读写下的集群性能和稳定性问题进行了针对性优化&…...
wordpress the_field/如何让自己的网站排名靠前
前言 很多后台服务很容被系统干掉,尤其红米类型的手机,它们可能是为了省电一些服务它们会自动消除掉。试了很多种方法,不如提高优先级什么的都没有太大的效果,这个ndk双进程守护相对好一点,但是也并不是多么的保险&am…...
学做网站必须php吗/软文推广发布
开发环境:系统环境:龙芯1B开发板(mips32指令)、Linux 3.0.0内核编译环境:Ubuntu 10.04 ,gcc-3.4.6-2f本文要用到的相关文件(cramfs-1.1.tar.gz、yaffs2-d43e901.tar.gz、mtd-utils-1.0.0.tar.gz)下载:用户名与密码都是www.linuxid…...
php网站建设用什么软件/今日头条新闻大事
目标 熟悉安骑士的架构和基本功能使用“基线检查”功能对ECS进行安全检测设置周期任务定期监控ECS的安全风险安骑士基本介绍 安骑士:运行在服务器上的轻量级插件,通过与云端的大数据威胁情报库联动,提供服务器整体的高危风险检查、实时入侵告…...