1、关于systemd
systemd 是Linux系统启动和服务器守护进程管理器，负责在系统启动或运行时，激活系统资源，服务器进程和其它进程，systemd被设计用来改进原来sysvinit中的多个缺点。
CentOS 7的systemd服务程序脚本存放在 /usr/lib/systemd/目录下，并区分 system 和 user，每一个服务程序脚本以 .service 结尾，例如 /usr/lib/systemd/system/sshd.service。
2、编辑systemd服务程序脚本
设定 GreatSQL 二进制文件放在 /usr/local/GreatSQL-8.0.23-14/ 目录下，即设定 basedir 为此目录，先进入到这个工作目录中。
[root@greatsql~]# cd /usr/local/GreatSQL-8.0.23-14/
复制 support-files/greatsql.server 程序脚本到 /usr/lib/systemd/system/ 目录下：
[root@greatsql~]# cp -f ./support-files/greatsql.server /usr/lib/systemd/system/
该脚本内容如下，基本上不需要再修改什么内容：
[root@greatsql~]# cat /usr/lib/systemd/system/greatsql.service[Unit]Description=GreatSQL ServerDocumentation=man:mysqld(8)Documentation=http://dev.mysql.com/doc/refman/en/using-systemd.htmlAfter=network.targetAfter=syslog.target[Install]WantedBy=multi-user.target[Service]User=mysqlGroup=mysqlType=notifyTimeoutSec=0PermissionsStartOnly=trueExecStartPre=/usr/local/GreatSQL-8.0.23-14/bin/mysqld_pre_systemdExecStart=/usr/local/GreatSQL-8.0.23-14/bin/mysqld $MYSQLD_OPTSEnvironmentFile=-/etc/sysconfig/mysqlLimitNOFILE = 10000Restart=on-failureRestartPreventExitStatus=1Environment=MYSQLD_PARENT_PID=1PrivateTmp=false
3、准备my.cnf及其他配置文件
复制 support-files/my.cnf 到 /etc/ 目录下，替换原来的配置文件（原来的 /etc/my.cnf 建议先备份），并确认 datadir、port、server_id 等参数是否要修改：
[root@greatsql~]# cp -f ./support-files/my.cnf /etc/my.cnf[root@greatsql~]# cat /etc/my.cnf#my.cnf[mysqld]user = mysqlport = 3306server_id = 3306basedir=/usr/local/GreatSQL-8.0.23-14datadir = /data/GreatSQLsocket = /data/GreatSQL/mysql.sockpid-file = mysql.pidcharacter-set-server = UTF8MB4skip_name_resolve = 1#若你的MySQL数据库主要运行在境外，请务必根据实际情况调整本参数default_time_zone = "+8:00"#performance setttingslock_wait_timeout = 3600open_files_limit = 65535back_log = 1024max_connections = 512max_connect_errors = 1000000table_open_cache = 1024table_definition_cache = 1024thread_stack = 512Ksort_buffer_size = 4Mjoin_buffer_size = 4Mread_buffer_size = 8Mread_rnd_buffer_size = 4Mbulk_insert_buffer_size = 64Mthread_cache_size = 768interactive_timeout = 600wait_timeout = 600tmp_table_size = 32Mmax_heap_table_size = 32M#log settingslog_timestamps = SYSTEMlog_error = /data/GreatSQL/error.loglog_error_verbosity = 3slow_query_log = 1log_slow_extra = 1slow_query_log_file = /data/GreatSQL/slow.loglong_query_time = 0.1log_queries_not_using_indexes = 1log_throttle_queries_not_using_indexes = 60min_examined_row_limit = 100log_slow_admin_statements = 1log_slow_slave_statements = 1log_bin = /data/GreatSQL/binlogbinlog_format = ROWsync_binlog = 1binlog_cache_size = 4Mmax_binlog_cache_size = 2Gmax_binlog_size = 1Gbinlog_rows_query_log_events = 1binlog_expire_logs_seconds = 604800#MySQL 8.0.22前，想启用MGR的话，需要设置binlog_checksum=NONE才行binlog_checksum = CRC32gtid_mode = ONenforce_gtid_consistency = TRUE#myisam settingskey_buffer_size = 32Mmyisam_sort_buffer_size = 128M#replication settingsmaster_info_repository = TABLErelay_log_info_repository = TABLErelay_log_recovery = 1slave_parallel_type = LOGICAL_CLOCK#可以设置为逻辑CPU数量的2倍slave_parallel_workers = 64binlog_transaction_dependency_tracking = WRITESETslave_preserve_commit_order = 1slave_checkpoint_period = 2#mgr settingsloose-plugin_load_add = 'mysql_clone.so'loose-plugin_load_add = 'group_replication.so'loose-group_replication_group_name = "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaa1"#MGR本地节点IP:PORT，请自行替换loose-group_replication_local_address = "172.16.16.10:33061"#MGR集群所有节点IP:PORT，请自行替换loose-group_replication_group_seeds = "172.16.16.10:33061,172.16.16.11:33061,172.16.16.12:33061"loose-group_replication_start_on_boot = OFFloose-group_replication_bootstrap_group = OFFloose-group_replication_exit_state_action = READ_ONLYloose-group_replication_flow_control_mode = "DISABLED"loose-group_replication_single_primary_mode = ON#innodb settingstransaction_isolation = REPEATABLE-READinnodb_buffer_pool_size = 2Ginnodb_buffer_pool_instances = 8innodb_data_file_path = ibdata1:12M:autoextendinnodb_flush_log_at_trx_commit = 1innodb_log_buffer_size = 32Minnodb_log_file_size = 1Ginnodb_log_files_in_group = 3innodb_max_undo_log_size = 4G# 根据您的服务器IOPS能力适当调整# 一般配普通SSD盘的话，可以调整到 10000 - 20000# 配置高端PCIe SSD卡的话，则可以调整的更高，比如 50000 - 80000innodb_io_capacity = 4000innodb_io_capacity_max = 8000innodb_open_files = 65535innodb_flush_method = O_DIRECTinnodb_lru_scan_depth = 4000innodb_lock_wait_timeout = 10innodb_rollback_on_timeout = 1innodb_print_all_deadlocks = 1innodb_online_alter_log_max_size = 4Ginnodb_print_ddl_logs = 1innodb_status_file = 1#注意: 开启 innodb_status_output & innodb_status_output_locks 后, 可能会导致log_error文件增长较快innodb_status_output = 0innodb_status_output_locks = 1innodb_sort_buffer_size = 67108864#innodb monitor settingsinnodb_monitor_enable = "module_innodb"innodb_monitor_enable = "module_server"innodb_monitor_enable = "module_dml"innodb_monitor_enable = "module_ddl"innodb_monitor_enable = "module_trx"innodb_monitor_enable = "module_os"innodb_monitor_enable = "module_purge"innodb_monitor_enable = "module_log"innodb_monitor_enable = "module_lock"innodb_monitor_enable = "module_buffer"innodb_monitor_enable = "module_index"innodb_monitor_enable = "module_ibuf_system"innodb_monitor_enable = "module_buffer_page"innodb_monitor_enable = "module_adaptive_hash"#pfs settingsperformance_schema = 1#performance_schema_instrument = '%memory%=on'performance_schema_instrument = '%lock%=on'
再复制 support-files/sysconfig/mysql 文件到 /etc/sysconfig 目录下。
[root@greatsql~]# cp -f ./support-files/sysconfig/mysql /etc/sysconfig/mysql[root@greatsql~]# cat /etc/sysconfig/mysqlLD_PRELOAD=/usr/lib64/libjemalloc.so.1
这是为了能让GreatSQL启动时加载jemalloc，获得更好的内存管理性能。
4、启动GreatSQL服务
上述准备工作完成后，还要将 GreatSQL 服务正式加载到 systemd 列表中：
[root@greatsql~]# systemctl daemon-reload
这就会自动将GreatSQL这个新服务加载进来了。
然后开始启动服务：
[root@greatsql~]# systemctl start greatsql#如果加上 .service 后缀也可以，例如[root@greatsql~]# systemctl start greatsql.service
如果启动失败，可能会输出类似的报错提示：
[root@greatsql~]# systemctl start greatsqlJob for greatsql.service failed because the control process exited with error code. See "systemctl status greatsql.service" and "journalctl -xe" for details.
按照错误提示，执行 systemctl status greatsql.service 或 journalctl -xe 查看失败的原因，亦可查看系统日志 /var/log/messages 或 GreatSQL日志 /data/GreatSQL/error.log 确认失败原因。
第一次会比较慢，因为还要初始化InnoDB系统表空间、redo、undo日志等。初始化时，还会为root生成临时密码，例如这样：
...[MY-010454] [Server] A temporary password is generated for root@localhost: dofI,0HE#E>C...
首次登入需要用这个密码，并立即修改成新密码，新密码还必须符合安全规范才行，不能是弱密码：
# 先登入[root@greatsql~]# mysql -uroot -p -S/data/GreatSQL/mysql.sockEnter password: ************# 而后修改密码，否则不能做其他操作[root@GreatSQL][(none)]> ALTER USER CURRENT_USER() IDENTIFIED BY 'GreatSQL3##)^';
这是因为用默认加载了 validate_password 组件（component），对密码有合规性要求。
[root@GreatSQL][(none)]> SELECT * FROM mysql.component;+--------------+--------------------+------------------------------------+| component_id | component_group_id | component_urn |+--------------+--------------------+------------------------------------+| 1 | 1 | file://component_validate_password |+--------------+--------------------+------------------------------------+
如果嫌麻烦可以卸载掉（但非常不建议这么做）：
[root@GreatSQL][(none)]> UNINSTALL COMPONENT 'file://component_validate_password';
最后，介绍systemctl主要几个的命令：
- 重载系统服务：systemctl daemon-reload
- 启用开机启动：systemctl enable greatsql
- 禁用开机启动：systemctl disable greatsql
- 启动服务：systemctl start greatsql
- 停止服务：systemctl stop greatsql
- 重启服务：systemctl restart greatsql
- 查看服务状态：systemctl status greatsql
全文完。
Enjoy GreatSQL 🙂
