腾讯云数据库TDSQL-单机安装体验&安装脚本分析
目录
一、机器配置
二、部署规划
三、修改tdsql_hosts文件
四、修改ansible变量group_vars/all文件
五、安装ansible
六、安装part1_site.yml
七、安装part1_site.yml
八、安装hdfs
九、安装kafka
十、安装consumer
十一、安装后的资源使用截图
十二、创建分布式实例
总结
前言
根据腾讯云官方要求，TDSQL安装至少需要三台机器。本人想要学习TDSQL但是没有足够的机器资源，经过仔细分析研究TDSQL安装过程，尝试发现单台机器也可以成功安装TDSQL，现将本人单台机器安装TDSQL的过程以及安装过程中遇到的问题在此分享。具体安装过程可参考腾讯官方文档，本文仅分享与官方文档不一致的步骤。
一、机器配置
本次尝试在win10笔记本电脑安装vmware虚拟机，虚拟机安装centos操作系统，centos安装tdsql
win10笔记本电脑配置：
处理器：Intel(R) Core(TM) i5-8265U CPU @ 1.60GHz 1.80 GHz
内存：8.00 GB (7.78 GB 可用)
虚拟机配置：处理器：2颗CPU 内存：6G 硬盘：80G
虚拟机操作系统：centos 7.6
二、部署规划
| 192.168.92.132 |
|
| zk |
Y |
| scheduler |
Y |
| oss |
Y |
| chitu |
Y |
| monitor（采集监控） |
Y |
| db |
Y |
| proxy |
Y |
| hdfs |
Y |
| kafka | Y |
| consumer | Y |
三、修改tdsql_hosts文件
# 填入所有机器的地址（包括主控机）
# 用于环境检测和初始化配置
# 同一个标签组内的ip地址不要重复，一个ip一行
# 序号从1递增，比如只有2台机器，那么序号保留tdsql_mac1 tdsql_mac2
# 后续如果有扩容，新扩容的机器也需要在这里补上
[tdsql_allmacforcheck]
tdsql_mac1 ansible_ssh_host=192.168.92.132
# zookeeper机器ip地址
# 数量只能是1、3、5， 正式环境建议3台或者5台
# 如果是使用自己已有的zk集群，这里同样要填写正确的zk地址
[tdsql_zk]
tdsql_zk1 ansible_ssh_host=192.168.92.132
# scheduler机器的ip地址，正式环境建议2个
[tdsql_scheduler]
tdsql_scheduler1 ansible_ssh_host=192.168.92.132
# oss机器的ip地址， 正式环境建议2个
[tdsql_oss]
tdsql_oss1 ansible_ssh_host=192.168.92.132
# 赤兔机器的ip地址， 正式环境建议2个
[tdsql_chitu]
tdsql_chitu1 ansible_ssh_host=192.168.92.132
# 监控采集模块的IP地址，正式环境建议2个
[tdsql_monitor]
tdsql_monitor1 ansible_ssh_host=192.168.92.132
# db机器地址
# 有多少台就填多少个ip地址
# 注意tdsql_db序号逐个递增，不要跳跃
[tdsql_db]
tdsql_db1 ansible_ssh_host=192.168.92.132
# proxy机器地址
[tdsql_proxy]
tdsql_proxy1 ansible_ssh_host=192.168.92.132
# hdfs机器地址
# 机器数量只能是1台或者3台
# 如果单节点部署，则只保留tdsql_hdfs1，其他删除
# 正式环境建议高可用部署，初始安装数量只能3台
[tdsql_hdfs]
tdsql_hdfs1 ansible_ssh_host=192.168.92.132
# lvs机器地址，数量固定2台
[tdsql_lvs]
tdsql_lvs1 ansible_ssh_host=192.168.92.132
# kafka机器地址，数量固定3台
[tdsql_kafka]
tdsql_kafka1 ansible_ssh_host=192.168.92.132
# 多源同步消费者服务ip地址，数量固定1台
[tdsql_consumer]
tdsql_consumer1 ansible_ssh_host=192.168.92.132
# es机器地址
[tdsql_es]
tdsql_es1 ansible_ssh_host=192.168.92.132
[tdsql_mc]
tdsql_mc1 ansible_ssh_host=192.168.92.132
[tdsql_newdb]
tdsql_newdb1 ansible_ssh_host=192.168.92.132
[tdsql_ansible_test]
tdsql_ansible_test1 ansible_ssh_host=192.168.92.132四、修改ansible变量group_vars/all文件
---
# scheduler,oss机器网卡
tdsql_sche_netif: ens33
# 操作系统账号tdsql的明文密码
# 如果有规划要部署两个集群做DCN同步, 则这两个集群的tdsql密码要一致
tdsql_os_pass: a+complex+password
# tdsql在zk上的根路径, 保持默认不允许修改
tdsql_zk_rootdir: /tdsqlzk
# zk机器的域名配置, 会写入各配置文件, 并将域名配置到/etc/hosts中
# 正式环境必须用机房或者地区的关键字, 有意义的关键字来命名
# 如果部署多套TDSQL集群, 则名字需要唯一
# 例如: 深圳机房zk的域名可以定义为tdsql_sz_zk
tdsql_zk_domain_name: tdsql_test_zk
# zk端口配置, 保持默认不要改,如果是自建的zk, 则和已有zk端口保持一致
tdsql_zk_clientport: 2118
tdsql_zk_serverport1: 2338
tdsql_zk_serverport2: 2558
# 赤兔监控库配置, 赤兔初始化完成后需要将监控库信息在这里更新
tdsql_metadb_ip: 192.168.92.132
tdsql_metadb_port: 15001
tdsql_metadb_ip_bak: 192.168.92.132
tdsql_metadb_port_bak: 15001
tdsql_metadb_user: hanlon
tdsql_metadb_password: 123456
# hdfs机器的ssh端口
tdsql_hdfs_ssh: 22
# hdfs数据目录, 正式环境要求mount挂载比较大的数据盘
tdsql_hdfs_datadir: /data2/hdfs,/data3/hdfs,/data4/hdfs
# kafka日志目录，正式环境要求mount挂载比较大的数据盘
tdsql_kafka_logdir: /data2/kafka,/data3/kafka,/data4/kafka
# 多源同步消费服务的机器网卡
tdsql_consumer_netif: ens33
# es7配置
tdsql_es7_mem: 4
tdsql_es7_base_path: /data1/es
tdsql_helper_cluster_name: tdsql
# 一致性读MC机器的网卡, 需要安装MC时配置
tdsql_mc_netif: ens33
update_tdsqlinstall_packet: mysqlagent五、安装ansible
在tdsql_install 目录下执行
source scripts/environment_set
sh scripts/install_ansible.sh
验证ansible安装是否成功
ansible --version六、安装part1_site.yml
ansible-playbook -i tdsql_hosts playbooks/tdsql_part1_site.yml该过程安装不是很顺利，首次安装遭遇如下报错提示：
根据报错信息检查发现：/usr/local/php/etc/php-fpm.d目录不存在，
报错环节：TASK[tdsql_chitu:start_php]
进一步分析tdsql_part1_site.yml安装脚本，理清安装过程如下：
1.tdsql_part1_site.yml调用roles/tdsql_chitu/tasks/main.yml
- name: install chitu server
hosts: tdsql_chitu
remote_user: root
gather_facts: false
roles:
- tdsql_chitu
tags: chitu2.roles/tdsql_chitu/tasks/main.yml包含了报错信息中执行的脚本
######## install php ########
- name: copy the tdsql_php file to the dst host
synchronize:
src: "{{ playbook_dir }}/../../group_files/tdsql_php/"
dest: /data/tools/
- name: copy install_php.sh
synchronize:
src: "../files/shell_scripts/install_php.sh"
dest: /data/tools/
- name: copy the php conf
synchronize:
src: "../templates/{{ item }}"
dest: /data/tools/
with_items: ['www.conf', 'php.ini', 'php-fpm.conf']
- name: compile the php
shell: "/bin/bash /data/tools/install_php.sh"
- name: add the x privi
shell: "chmod +x -R /usr/local/php && chown -R nginx:nginx /usr/local/php"
- name: add the nginx privi
shell: "chown -R nginx:nginx /data/website"
- name: start php
shell: "/usr/local/php/sbin/php-fpm -y /usr/local/php/etc/php-fpm.conf -c /usr/local/php/lib/php.ini"
become: true
become_user: nginx
######## install php ########该脚本主要完成了以下几个工作：
（1）拷贝php安装包到/data/tools/目录
（2）拷贝install_php.sh到/data/tools/目录
（3）拷贝php conf文件到/data/tools/目录
（4）执行install_php.sh脚本
（5）添加执行权限，修改归属用户和组
（6）start php
其中start php即为报错信息中执行的脚本，根据报错信息启动php需要在/usr/local/php/etc/php-fpm.d目录找配置文件，但是以上脚本将配置文件拷贝到了/data/tools目录，然后执行了install_php.sh脚本，所以有可能改变配置文件的脚本只可能是install_php.sh脚本
3.查看install_php.sh脚本
#!/usr/bin/env bash
pkill php-fpm
# 如果php-fpm存在则不编译
#test -e /usr/local/php/sbin/php-fpm && exit 0
# 安装PHP
cd /data/tools
test -e php || tar xf php.tar.gz
cd php
export PATH=/usr/bin:${PATH}
# x86
uname -i | grep x86_64 && ./configure --prefix=/usr/local/php \
--enable-fpm \
--with-fpm-user=nginx \
--with-fpm-group=nginx \
--with-libxml-dir \
--with-gd \
--with-jpeg-dir \
--with-png-dir \
--with-zlib-dir \
--enable-gd-jis-conv \
--enable-mysqlnd \
--enable-mbstring \
--with-mysqli=mysqlnd \
--with-pdo-mysql=mysqlnd \
--enable-sockets \
--enable-soap \
--with-curl \
--with-openssl
# arm
uname -i | grep aarch64 && ./configure --build=arm --prefix=/usr/local/php \
--enable-fpm \
--with-fpm-user=nginx \
--with-fpm-group=nginx \
--with-libxml-dir \
--with-gd \
--with-jpeg-dir \
--with-png-dir \
--with-zlib-dir \
--enable-gd-native-ttf \
--enable-gd-jis-conv \
--enable-mysqlnd \
--enable-mbstring \
--with-mysql=mysqlnd \
--with-mysqli=mysqlnd \
--with-pdo-mysql=mysqlnd \
--enable-sockets \
--enable-soap \
--with-curl=/usr/local/curl \
--with-mcrypt \
--with-openssl
make -j 4 && make install
cd /data/tools
/bin/cp -a php.ini /usr/local/php/lib/php.ini
/bin/cp -a php-fpm.conf /usr/local/php/etc/php-fpm.conf
/bin/cp -a www.conf /usr/local/php/etc/php-fpm.d/www.conf
mkdir -p /data/website/logs
#启动php-fpm
#/usr/local/php/sbin/php-fpm -y /usr/local/php/etc/php-fpm.conf -c /usr/local/php/lib/php.ini
You have new mail in /var/spool/mail/root该脚本第7行即完成了/usr/local/php/etc/php-fpm.d目录中配置文件的拷贝
经尝试重新执行install_php.sh脚本后/usr/local/php/etc/php-fpm.d目录中产生了www.conf配置文件
再次尝试执行安装part1_site.yml发现之前的报错已经消失，但是产生了新的报错信息：
fatal: [tdsql_chitu1]: FAILED! => {"changed": true, "cmd": "/usr/local/php/sbin/php-fpm -y /usr/local/php/etc/php-fpm.conf -c /usr/local/php/lib/php.ini", "delta": "0:00:00.130908", "end": "2022-05-07 23:42:32.457796", "msg": "non-zero return code", "rc": 70, "start": "2022-05-07 23:42:32.326888", "stderr": "[07-May-2022 23:42:32] NOTICE: [pool www] 'user' directive is ignored when FPM is not running as root\n[07-May-2022 23:42:32] NOTICE: [pool www] 'group' directive is ignored when FPM is not running as root\n[07-May-2022 23:42:32] ERROR: unable to bind listening socket for address '127.0.0.1:9000': Address already in use (98)\n[07-May-2022 23:42:32] ERROR: FPM initialization failed", "stderr_lines": ["[07-May-2022 23:42:32] NOTICE: [pool www] 'user' directive is ignored when FPM is not running as root", "[07-May-2022 23:42:32] NOTICE: [pool www] 'group' directive is ignored when FPM is not running as root", "[07-May-2022 23:42:32] ERROR: unable to bind listening socket for address '127.0.0.1:9000': Address already in use (98)", "[07-May-2022 23:42:32] ERROR: FPM initialization failed"], "stdout": "", "stdout_lines": []}分析报错信息发现报错原因是127.0.0.1:9000这个端口服务已经存在
使用netstat命令查看该端口服务启动的进程
[root@tdsql01 php-fpm.d]# netstat -anp | grep 9000
tcp 0 0 127.0.0.1:9000 0.0.0.0:* LISTEN 70994/php-fpm: mast
You have new mail in /var/spool/mail/root
[root@tdsql01 php-fpm.d]# ps -ef|grep 70994
root 70994 70820 0 23:48 ? 00:00:00 php-fpm: master process (/usr/local/php/etc/php-fpm.conf)
nginx 71006 70994 0 23:48 ? 00:00:00 php-fpm: pool www
nginx 71007 70994 0 23:48 ? 00:00:00 php-fpm: pool www
nginx 71008 70994 0 23:48 ? 00:00:00 php-fpm: pool www
nginx 71009 70994 0 23:48 ? 00:00:00 php-fpm: pool www
nginx 71010 70994 0 23:48 ? 00:00:00 php-fpm: pool www
nginx 71011 70994 0 23:48 ? 00:00:00 php-fpm: pool www
nginx 71012 70994 0 23:48 ? 00:00:00 php-fpm: pool www
nginx 71013 70994 0 23:48 ? 00:00:00 php-fpm: pool www
nginx 71014 70994 0 23:48 ? 00:00:00 php-fpm: pool www
nginx 71015 70994 0 23:48 ? 00:00:00 php-fpm: pool www
nginx 71016 70994 0 23:48 ? 00:00:00 php-fpm: pool www
nginx 71017 70994 0 23:48 ? 00:00:00 php-fpm: pool www
nginx 71018 70994 0 23:48 ? 00:00:00 php-fpm: pool www
nginx 71019 70994 0 23:48 ? 00:00:00 php-fpm: pool www
nginx 71020 70994 0 23:48 ? 00:00:00 php-fpm: pool www
nginx 71021 70994 0 23:48 ? 00:00:00 php-fpm: pool www
nginx 71022 70994 0 23:48 ? 00:00:00 php-fpm: pool www
nginx 71023 70994 0 23:48 ? 00:00:00 php-fpm: pool www
nginx 71024 70994 0 23:48 ? 00:00:00 php-fpm: pool www
nginx 71025 70994 0 23:48 ? 00:00:00 php-fpm: pool www
root 73778 1170 0 23:58 pts/8 00:00:00 grep --color=auto 70994
[root@tdsql01 php-fpm.d]# ps -ef|grep 70820
root 70820 1 0 23:48 ? 00:00:00 /usr/bin/python /usr/bin/supervisord -c /etc/supervisord.conf
root 70836 70820 0 23:48 ? 00:00:00 nginx: master process /usr/local/nginx/sbin/nginx -g daemon off; -c /usr/local/nginx/conf/nginx.conf
root 70994 70820 0 23:48 ? 00:00:00 php-fpm: master process (/usr/local/php/etc/php-fpm.conf)
root 73796 1170 0 23:58 pts/8 00:00:00 grep --color=auto 70820查看/etc/supervisord.conf配置文件，最后两行包含以下内容
[include]
files = /etc/supervisord.d/*.conf查看查看/etc/supervisord.d目录下的文件，包含supervisord_php.conf脚本，内容如下：
; TDSQL supervisor configure file for php-fpm
[program:php-fpm]
command=/usr/local/php/sbin/php-fpm -y /usr/local/php/etc/php-fpm.conf --nodaemonize -c /usr/local/php/lib/php.ini
user=root
autostart=true
autorestart=true
startsecs=3
startretries=100
stdout_logfile=/var/log/php-fpm_stdout.log
stdout_logfile_maxbytes=1MB
stdout_logfile_backups=2
stderr_logfile=/var/log/php-fpm_stderr.log
stderr_logfile_maxbytes=1MB
stderr_logfile_backups=2该脚本包含了启动php-fpm的脚本
Supervisor 是一个用 Python 写的进程管理工具，可以很方便的用来在 UNIX-like 系统（不支持 Windows）下启动、重启（自动重启程序）、关闭进程。根据以上内容分析，tdsql安装脚本在/etc/supervisord.d目录放置了启动php-fpm的脚本，用于自动启动php-fpm进程，再次安装报错是因为php-fpm进程被Supervisor进程自动拉起。删除掉该目录下的文件后再次尝试执行安装part1_site.yml，未报错，安装成功。安装成功后显示信息如下：
2022-05-08 00:17:42,588 p=34880 u=root n=ansible | tdsql_chitu1 : ok=39 changed=26 unreachable=0 failed=0 skipped=0 rescued=0 ignored=0
2022-05-08 00:17:42,588 p=34880 u=root n=ansible | tdsql_db1 : ok=9 changed=9 unreachable=0 failed=0 skipped=0 rescued=0 ignored=0
2022-05-08 00:17:42,588 p=34880 u=root n=ansible | tdsql_mac1 : ok=61 changed=41 unreachable=0 failed=0 skipped=0 rescued=0 ignored=0
2022-05-08 00:17:42,588 p=34880 u=root n=ansible | tdsql_oss1 : ok=15 changed=12 unreachable=0 failed=0 skipped=1 rescued=0 ignored=0
2022-05-08 00:17:42,590 p=34880 u=root n=ansible | tdsql_proxy1 : ok=9 changed=8 unreachable=0 failed=0 skipped=0 rescued=0 ignored=0
2022-05-08 00:17:42,590 p=34880 u=root n=ansible | tdsql_scheduler1 : ok=30 changed=29 unreachable=0 failed=0 skipped=0 rescued=0 ignored=0
2022-05-08 00:17:42,590 p=34880 u=root n=ansible | tdsql_zk1 : ok=13 changed=8 unreachable=0 failed=0 skipped=0 rescued=0 ignored=0 之后按照腾讯官方提供的安装步骤执行，未产生新的报错。
七、安装part1_site.yml
安装成功显示信息如下：
2022-05-08 00:36:52,749 p=23280 u=root n=ansible | tdsql_chitu1 : ok=18 changed=15 unreachable=0 failed=0 skipped=4 rescued=0 ignored=0
2022-05-08 00:36:52,749 p=23280 u=root n=ansible | tdsql_mac1 : ok=46 changed=32 unreachable=0 failed=0 skipped=0 rescued=0 ignored=0
2022-05-08 00:36:52,749 p=23280 u=root n=ansible | tdsql_monitor1 : ok=13 changed=13 unreachable=0 failed=0 skipped=0 rescued=0 ignored=0
2022-05-08 00:36:52,750 p=23280 u=root n=ansible | tdsql_oss1 : ok=15 changed=10 unreachable=0 failed=0 skipped=1 rescued=0 ignored=0
2022-05-08 00:36:52,750 p=23280 u=root n=ansible | tdsql_scheduler1 : ok=38 changed=35 unreachable=0 failed=0 skipped=0 rescued=0 ignored=0 八、安装hdfs
安装成功显示信息如下：
2022-05-08 21:45:19,518 p=42548 u=root n=ansible | tdsql_hdfs1 : ok=12 changed=12 unreachable=0 failed=0 skipped=1 rescued=0 ignored=0
2022-05-08 21:45:19,519 p=42548 u=root n=ansible | tdsql_mac1 : ok=58 changed=42 unreachable=0 failed=0 skipped=3 rescued=0 ignored=0 九、安装kafka
安装报错信息如下：
2022-05-08 21:52:01,006 p=67751 u=root n=ansible | fatal: [tdsql_kafka1]: UNREACHABLE! => {"changed": true, "msg": "All items completed", "results": [{"ansible_loop_var": "item", "changed":
true, "failed": false, "invocation": {"module_args": {"after": null, "attributes": null, "backup": false, "before": null, "content": null, "delimiter": null, "dest": "/data/application/kaf
ka/config/server.properties", "directory_mode": null, "encoding": "utf-8", "follow": false, "force": null, "group": null, "mode": null, "owner": null, "path": "/data/application/kafka/confi
g/server.properties", "regexp": "wait_for_change_brokerid", "remote_src": null, "replace": "1", "selevel": null, "serole": null, "setype": null, "seuser": null, "src": null, "unsafe_writes"
: null, "validate": null}}, "item": 1, "msg": "1 replacements made"}, {"ansible_loop_var": "item", "item": 2, "msg": "Failed to connect to the host via ssh: ssh: Could not resolve hostname
tdsql_kafka2: Name or service not known", "unreachable": true}, {"ansible_loop_var": "item", "item": 3, "msg": "Failed to connect to the host via ssh: ssh: Could not resolve hostname tdsql_
kafka3: Name or service not known", "unreachable": true}], "warnings": ["The value 1 (type int) in a string field was converted to u'1' (type string). If this does not look like what you ex
pect, quote the entire value to ensure it does not change."]}
2022-05-08 21:52:01,006 p=67751 u=root n=ansible | NO MORE HOSTS LEFT ***********************************************************************************************************************
***************************************************
2022-05-08 21:52:01,007 p=67751 u=root n=ansible | PLAY RECAP *******************************************************************************************************************************
***************************************************
2022-05-08 21:52:01,007 p=67751 u=root n=ansible | tdsql_kafka1 : ok=51 changed=36 unreachable=1 failed=0 skipped=0 rescued=0 ignored=0 错误信息提取关键内容：Could not resolve hostname tdsql_kafka2
分析原因：安装脚本要求kafka必须三台机器，本人只有一台机器，所以需要修改kafka安装脚本
查看kafka安装脚本playbooks/tdsql_kafka.yml，内容如下：
---
- name: install kafka beginning
hosts: tdsql_kafka
remote_user: root
gather_facts: false
roles:
- tdsql_beginning
- name: install kafka
hosts: tdsql_kafka
remote_user: root
gather_facts: true
roles:
- tdsql_kafka查看roles/tdsql_kafka/tasks/main.yml包含以下关键内容
- name: set brokerid for kafka
replace:
dest: /data/application/kafka/config/server.properties
regexp: 'wait_for_change_brokerid'
replace: "{{ item }}"
run_once: true
delegate_to: tdsql_kafka{{ item }}
with_items: [1, 2, 3]这里指定了kafka安装主机需要三台，修改为1台，修改后的内容如下：
- name: set brokerid for kafka
replace:
dest: /data/application/kafka/config/server.properties
regexp: 'wait_for_change_brokerid'
replace: "{{ item }}"
run_once: true
delegate_to: tdsql_kafka{{ item }}
with_items: [1]再次执行安装脚本，提示安装成功
tdsql_kafka1 : ok=57 changed=40 unreachable=0 failed=0 skipped=0 rescued=0 ignored=0根据安装文档查看kafka启动信息，发现/data/application/kafka/logs目录下没有server.log文件
[root@tdsql01 tdsql_install]# cd /data/application/kafka/logs
[root@tdsql01 logs]# ls
kafkaServer-gc.log kafkaServer.out
[root@tdsql01 logs]# more kafkaServer-gc.log
OpenJDK 64-Bit Server VM (25.242-b1) for linux-amd64 JRE (1.8.0_242-b1), built on Feb 19 2020 08:17:47 by "root" with gcc 4.4.7 20120313 (Red Hat 4.4.7-23)
Memory: 4k page, physical 6925020k(2979164k free), swap 0k(0k free)
CommandLine flags: -XX:+ExplicitGCInvokesConcurrent -XX:InitialHeapSize=6442450944 -XX:InitiatingHeapOccupancyPercent=35 -XX:+ManagementServer -XX:MaxGCPauseMillis=20 -XX:MaxHeapSize=644245
0944 -XX:+PrintGC -XX:+PrintGCDateStamps -XX:+PrintGCDetails -XX:+PrintGCTimeStamps -XX:+UseCompressedClassPointers -XX:+UseCompressedOops -XX:+UseG1GC
You have new mail in /var/spool/mail/root
[root@tdsql01 logs]# more kafkaServer.out
OpenJDK 64-Bit Server VM warning: INFO: os::commit_memory(0x0000000680000000, 6442450944, 0) failed; error='Cannot allocate memory' (errno=12)
#
# There is insufficient memory for the Java Runtime Environment to continue.
# Native memory allocation (mmap) failed to map 6442450944 bytes for committing reserved memory.
# An error report file with more information is saved as:
# /data/application/kafka/bin/hs_err_pid22721.log
[root@tdsql01 logs]# free -g
total used free shared buff/cache available
Mem: 6 2 2 0 1 3
Swap: 0 0 0查看目录下的kafkaServer.out文件发现报错：内存不足
进一步分析kafka安装脚本，尝试修改内存限制
查看roles/tdsql_kafka/tasks/main.yml包含以下kafka启动脚本
- name: start kafka service
shell: "source /etc/profile ; cd /data/application/kafka/bin ; ./kafka-server-start.sh -daemon ../config/server.properties; sleep 5"
become: true
become_user: tdsql查看/data/application/kafka/bin/kafka-server-start.sh包含以下内存要求
if [ "x$KAFKA_HEAP_OPTS" = "x" ]; then
export KAFKA_HEAP_OPTS="-Xmx6G -Xms6G"
fi因为本机仅剩2G内存，这里修改JVM最大可用内存Xmx 和 初始内存Xms为1G
if [ "x$KAFKA_HEAP_OPTS" = "x" ]; then
export KAFKA_HEAP_OPTS="-Xmx1G -Xms1G"
fi再次执行启动kafka的脚本，并查看日志，发现生成了server.log
[root@tdsql01 bin]# ./kafka-server-start.sh -daemon ../config/server.properties
[root@tdsql01 bin]# cd ../logs
[root@tdsql01 logs]# ls
controller.log kafka-authorizer.log kafka-request.log kafkaServer-gc.log kafkaServer.out log-cleaner.log server.log state-change.log查看server.log包含启动kafka的关键信息
至此kafka安装成功
十、安装consumer
安装成功后显示信息及进程信息如下：
tdsql_consumer1 : ok=56 changed=37 unreachable=0 failed=0 skipped=0 rescued=0 ignored=0
You have new mail in /var/spool/mail/root
[root@tdsql01 tdsql_install]# ps -ef | grep consumer
root 28982 1 0 23:34 ? 00:00:00 ./binlogconsumermgn --zklist tdsql_test_zk1:2118 --zkrootpath /tdsqlzk --kafkazklist tdsql_test_zk1:2118 --kafkazkrootpath /kafka --dev ens33
root 28984 28982 0 23:34 ? 00:00:00 ./binlogconsumermgn --zklist tdsql_test_zk1:2118 --zkrootpath /tdsqlzk --kafkazklist tdsql_test_zk1:2118 --kafkazkrootpath /kafka --dev ens33
root 32007 61018 0 23:35 pts/1 00:00:00 grep --color=auto consumer十一、安装后的资源使用截图
安装后的机型规格截图：
设备资源截图：
十二、创建分布式实例
因电脑资源有限，创建分布式实例CPU，内存，磁盘均需手动修改设置为最小
容灾设置选择一主零备
其他设置选择默认即可，创建成功截图：
总结
1.实验证明tdsql可以单机安装，希望对于没有足够的机器资源但是想学习tdsql的小伙伴可以有所帮助，经本人测试除了tdsql必须的核心组件zk，scheduler，oss，chitu，monitor，db,proxy可以成功安装，非核心组件hdfs，kafka,consumer单机也可以成功安装，可以学习使用tdsql的绝大部分功能。
2.首次安装未成功产生启动php-frm的配置文件，具体原因后续有时间打算再做进一步尝试研究。
3.tdsql安装会在/etc/supervisord.d目录放置nginx，php，zookeeper的启动脚本，再次安装需要把这三个文件清理掉，否则安装会报错，提示端口已被使用。
4.kafka安装需要修改roles/tdsql_kafka/tasks/main.yml中数量，官方脚本限定kafka必须是3台，需要修改为1台。
5.kafka安装需要修改/data/application/kafka/bin/kafka-server-start.sh脚本中kafka启动的JVM内存设置，本次测试修改为1G，可以正常启动kafka
6.LVS负载均衡组件是否安装不影响tdsql的学习使用，故本次测试未安装LVS
7.es数据检索及分析工具是否安装不影响tdsql的学习使用，故本次测试未安装LVS