番外系列说明:该系列所有文章都将作为独立篇章进行知识点讲解,是对其他系列博文进行的补充说明,来自于博客园AskScuti。
主题:关于数据库启动流程的三个阶段
内容预览:本篇涉及数据库启动的三个阶段分别做了什么事及对应操作方式。
目录
1. 启动概述
2. 第一阶段
3. 第二阶段
4. 第三阶段
5. 汇总说明
1. 启动概述
数据库启动从关闭到开启经历三个阶段:NOMOUNT、MOUNT和OPEN。每个阶段所能查看的动态性能视图也不一样,比如:在NOMOUNT阶段,可查询动态性能视图 v$instance,无法查询 v$database,因为后者里面的内容是在控制文件中查询的,而控制文件是在第二阶段才被加载。具体图例请参考官方文档11g Release 2 (11.2) Database Concepts:Overview of Instance and Database Startup
2. 第一阶段
数据库启动的第一阶段将加载参数文件,按照参数文件具体值进行创建SGA,并开启一系列后台进程。
将数据库启动到NOMOUNT阶段
SQL> startup nomount ORACLE instance started. Total System Global Area 830930944 bytes Fixed Size 2232920 bytes Variable Size 591400360 bytes Database Buffers 234881024 bytes Redo Buffers 2416640 bytes
我们来查下当前实例状态
SQL> select status from v$instance; STATUS ------------------------ STARTED
在 NOMOUNT 状态显示为:STARTED状态,也就是加载参数文件的第一阶段状态。
查看日志确定动作,节选日志片段
![](http://img.e-com-net.com/image/info8/b8d97b5613f94ed2ba791cad57d0b2ed.gif)
![](http://img.e-com-net.com/image/info8/2f88dd3f1cd145f59c0e47b51acdbd4b.gif)
1 Starting up: 2 Oracle Database 11g Enterprise Edition Release 11.2.0.3.0 - 64bit Production 3 With the Partitioning, OLAP, Data Mining and Real Application Testing options. 4 ORACLE_HOME = /u01/app/oracle/product/11.2.0/dbhome_1 5 System name: Linux 6 Node name: henry 7 Release: 3.10.0-862.el7.x86_64 8 Version: #1 SMP Wed Mar 21 18:14:51 EDT 2018 9 Machine: x86_64 10 VM name: VMWare Version: 6 11 Using parameter settings in server-side spfile /u01/app/oracle/product/11.2.0/dbhome_1/dbs/spfilePROD1.ora 12 System parameters with non-default values: 13 processes = 150 14 sessions = 247 15 memory_target = 796M 16 control_files = "/u01/app/oracle/oradata/PROD1/control01.ctl" 17 control_files = "/u01/app/oracle/oradata/PROD1/control02.ctl" 18 control_files = "/u01/app/oracle/oradata/PROD1/control03.ctl" 19 control_files = "/u01/app/oracle/oradata/PROD1/control04.ctl" 20 control_files = "/u01/app/oracle/oradata/PROD1/control05.ctl" 21 control_files = "/u01/app/oracle/oradata/PROD1/control06.ctl" 22 control_files = "/u01/app/oracle/oradata/PROD1/control07.ctl" 23 control_files = "/u01/app/oracle/oradata/PROD1/control08.ctl" 24 db_block_size = 8192 25 compatible = "11.2.0.0.0" 26 log_archive_dest_1 = "location=/u01/app/oracle/archive1" 27 db_recovery_file_dest = "/u01/app/oracle/fast_recovery_area" 28 db_recovery_file_dest_size= 4122M 29 undo_tablespace = "UNDOTBS1" 30 remote_login_passwordfile= "EXCLUSIVE" 31 db_domain = "" 32 dispatchers = "(PROTOCOL=TCP) (SERVICE=PROD1XDB)" 33 local_listener = "(ADDRESS=(PROTOCOL=tcp)(HOST=henry)(PORT=1566))" 34 local_listener = "(ADDRESS=(PROTOCOL=tcp)(HOST=henry)(PORT=1521))" 35 local_listener = "(ADDRESS=(PROTOCOL=tcp)(HOST=henry)(PORT=1234))" 36 job_queue_processes = 1000 37 audit_file_dest = "/u01/app/oracle/admin/PROD1/adump" 38 audit_trail = "DB" 39 db_name = "PROD1" 40 open_cursors = 300 41 diagnostic_dest = "/u01/app/oracle" 42 Wed May 22 10:21:56 2019 43 PMON started with pid=2, OS id=14957 44 Wed May 22 10:21:56 2019 45 PSP0 started with pid=3, OS id=14960 46 Wed May 22 10:21:57 2019 47 VKTM started with pid=4, OS id=14962 at elevated priority 48 VKTM running at (1)millisec precision with DBRM quantum (100)ms 49 Wed May 22 10:21:58 2019 50 GEN0 started with pid=5, OS id=14967 51 Wed May 22 10:21:58 2019 52 DIAG started with pid=6, OS id=14969 53 Wed May 22 10:21:58 2019 54 DBRM started with pid=7, OS id=14971 55 Wed May 22 10:21:58 2019 56 DIA0 started with pid=8, OS id=14973 57 Wed May 22 10:21:58 2019 58 MMAN started with pid=9, OS id=14975 59 Wed May 22 10:21:58 2019 60 DBW0 started with pid=10, OS id=14977 61 Wed May 22 10:21:58 2019 62 LGWR started with pid=11, OS id=14979 63 Wed May 22 10:21:58 2019 64 CKPT started with pid=12, OS id=14981 65 Wed May 22 10:21:58 2019 66 SMON started with pid=13, OS id=14983 67 Wed May 22 10:21:58 2019 68 RECO started with pid=14, OS id=14986 69 Wed May 22 10:21:58 2019 70 MMON started with pid=15, OS id=14988 71 Wed May 22 10:21:58 2019 72 MMNL started with pid=16, OS id=14990 73 starting up 1 dispatcher(s) for network address '(ADDRESS=(PARTIAL=YES)(PROTOCOL=TCP))'... 74 starting up 1 shared server(s) ... 75 ORACLE_BASE from environment = /u01/app/oracle
可以看到,在第一阶段启动过程中有这么一句话:Using parameter settings in server-side spfile /u01/app/oracle/product/11.2.0/dbhome_1/dbs/spfilePROD1.ora 。默认使用的是动态参数文件,就是按照这个动态参数文件里面的值开始开辟(创建)SGA内存区,包括但不限于整体SGA大小,数据库缓冲区大小,日志缓冲区大小等等。后面可以看到开启了一系列后台进程(这里有6个进程很重要:CKPT、SMON、PMON、DBWn、LGWR、ARCn,其中前5个是必须开启的,第6个是否启动是和当前数据库是否为归档模式有关,如果是归档模式,则启动,如果是非归档模式,则没有此进程)
系统命令查看后台进程
SQL> !ps -ef | grep ora_ oracle 16030 1 0 10:35 ? 00:00:00 ora_pmon_PROD1 oracle 16032 1 0 10:35 ? 00:00:00 ora_psp0_PROD1 oracle 16034 1 14 10:35 ? 00:00:04 ora_vktm_PROD1 oracle 16038 1 0 10:35 ? 00:00:00 ora_gen0_PROD1 oracle 16040 1 0 10:35 ? 00:00:00 ora_diag_PROD1 oracle 16042 1 0 10:35 ? 00:00:00 ora_dbrm_PROD1 oracle 16044 1 0 10:35 ? 00:00:00 ora_dia0_PROD1 oracle 16046 1 2 10:35 ? 00:00:00 ora_mman_PROD1 oracle 16048 1 0 10:35 ? 00:00:00 ora_dbw0_PROD1 oracle 16050 1 0 10:35 ? 00:00:00 ora_lgwr_PROD1 oracle 16052 1 0 10:35 ? 00:00:00 ora_ckpt_PROD1 oracle 16054 1 0 10:35 ? 00:00:00 ora_smon_PROD1 oracle 16056 1 0 10:35 ? 00:00:00 ora_reco_PROD1 oracle 16058 1 3 10:35 ? 00:00:01 ora_mmon_PROD1 oracle 16060 1 0 10:35 ? 00:00:00 ora_mmnl_PROD1 oracle 16062 1 0 10:35 ? 00:00:00 ora_d000_PROD1 oracle 16064 1 0 10:35 ? 00:00:00 ora_s000_PROD1 oracle 16131 1 0 10:35 ? 00:00:00 ora_arc0_PROD1 oracle 16134 1 7 10:35 ? 00:00:01 ora_arc1_PROD1 oracle 16136 1 0 10:35 ? 00:00:00 ora_arc2_PROD1 oracle 16138 1 0 10:35 ? 00:00:00 ora_arc3_PROD1 oracle 16143 1 0 10:35 ? 00:00:00 ora_qmnc_PROD1 oracle 16161 1 2 10:35 ? 00:00:00 ora_cjq0_PROD1 oracle 16166 1 5 10:35 ? 00:00:00 ora_j000_PROD1 oracle 16168 1 2 10:35 ? 00:00:00 ora_j001_PROD1 oracle 16170 1 0 10:35 ? 00:00:00 ora_j002_PROD1 oracle 16173 1 1 10:35 ? 00:00:00 ora_q000_PROD1 oracle 16176 1 0 10:35 ? 00:00:00 ora_q001_PROD1 oracle 16186 14851 0 10:35 pts/1 00:00:00 /bin/bash -c ps -ef | grep ora_ oracle 16188 16186 0 10:35 pts/1 00:00:00 grep ora_
其中,归档模式下,ARCH进程默认开启4个,可通过修改参数 log_archive_max_processes 最多可以开启30个,查看官方文档11g Release 2 (11.2) Database Administrator's Guide:Adjusting the Number of Archiver Processes
3. 第二阶段
数据库启动的第二阶段将加载控制文件,为打开数据文件和日志文件做准备。
将数据库更改到MOUNT状态
SQL> alter database mount; Database altered.
查看日志片段,可看到在控制文件里面记录了当前数据库的DBID(唯一ID,类似于身份证号),有了这个,就确定了接下来要打开哪个数据库了。
alter database mount Wed May 22 11:09:49 2019 Successful mount of redo thread 1, with mount id 2230329145 Database mounted in Exclusive Mode Lost write protection disabled Completed: alter database mount
数据库在MOUNT阶段后,就可以查询 v$database 数据字典了,刚才说过,它里面的东西都是从控制文件中查出来的。
SQL> select open_mode from v$database; OPEN_MODE ------------------------------------- MOUNTED
也可以查询数据文件、日志文件等,因为这些动态性能视图查的都是控制文件里面的内容。
SQL> select name from v$datafile; NAME -------------------------------------------------- /u01/app/oracle/oradata/PROD1/system01.dbf /u01/app/oracle/oradata/PROD1/sysaux01.dbf /u01/app/oracle/oradata/PROD1/undotbs01.dbf /u01/app/oracle/oradata/PROD1/users01.dbf /u01/app/oracle/oradata/PROD1/example01.dbf /u01/app/oracle/oradata/PROD1/abc01.dbf /u01/app/oracle/oradata/PROD1/abcd01.dbf /u01/app/oracle/oradata/PROD1/tbs_c01.dbf /u01/app/oracle/oradata/PROD1/aaa01.dbf /u01/app/oracle/oradata/PROD1/aaa02.dbf 10 rows selected.
但是无法查询数据文件里面的具体内容,因为数据文件还没有被加载,数据库还没有被打开。
4. 第三阶段
将数据库开启
SQL> alter database open; Database altered.
因为控制文件里面指定了数据文件、在线日志文件等具体路径,因此在OPEN阶段为数据库运行做最后检查确认并打开所有文件。至此,数据库完成启动。
SQL> select ename,sal from scott.emp; ENAME SAL ---------- ----- SMITH 800 ALLEN 1600 WARD 1250 JONES 2975 MARTIN 1250 BLAKE 2850 CLARK 2450 SCOTT 3000 KING 5000 TURNER 1500 ADAMS 1100 JAMES 950 FORD 3000 MILLER 1300 14 rows selected.
5. 汇总说明
对于初学者,数据库分步启动命令常与启动命令选项搞混淆。
你可以把数据库启动到NOMOUNT状态,仅加载参数文件:前提是数据库处于关闭状态。
SQL> startup nomount ORACLE instance started. Total System Global Area 830930944 bytes Fixed Size 2232920 bytes Variable Size 591400360 bytes Database Buffers 234881024 bytes Redo Buffers 2416640 bytes
你也可以把数据库直接启动到MOUNT状态,加载参数文件和控制文件:前提是数据库处于关闭状态。
SQL> startup mount ORACLE instance started. Total System Global Area 830930944 bytes Fixed Size 2232920 bytes Variable Size 591400360 bytes Database Buffers 234881024 bytes Redo Buffers 2416640 bytes Database mounted.
你也可以直接把数据库启动到OPEN状态,加载参数文件和控制文件,并打开其他所有文件:前提是数据库处于关闭状态。
SQL> startup ORACLE instance started. Total System Global Area 830930944 bytes Fixed Size 2232920 bytes Variable Size 591400360 bytes Database Buffers 234881024 bytes Redo Buffers 2416640 bytes Database mounted. Database opened.
请注意:
如果处于NOMOUNT状态,之后要打开数据库,可以先将状态更改为MOUNT,然后再将状态更改为OPEN,就是不能跨阶段。
SQL> startup nomount ORACLE instance started. Total System Global Area 830930944 bytes Fixed Size 2232920 bytes Variable Size 591400360 bytes Database Buffers 234881024 bytes Redo Buffers 2416640 bytes SQL> alter database mount; Database altered. SQL> alter database open; Database altered.
也可以在NOMOUNT状态,将数据库关闭,然后直接STARTUP。
SQL> startup nomount ORACLE instance started. Total System Global Area 830930944 bytes Fixed Size 2232920 bytes Variable Size 591400360 bytes Database Buffers 234881024 bytes Redo Buffers 2416640 bytes SQL> shutdown immediate ORA-01507: database not mounted ORACLE instance shut down. SQL> startup ORACLE instance started. Total System Global Area 830930944 bytes Fixed Size 2232920 bytes Variable Size 591400360 bytes Database Buffers 234881024 bytes Redo Buffers 2416640 bytes Database mounted. Database opened.
如果处于MOUNT状态,打开数据库需要用ALTER语句。
SQL> select status from v$instance; STATUS ------------------------ MOUNTED SQL> alter database open; Database altered.
或者直接关闭数据库,发出STARTUP命令启动。
最后,老手常用的 startup force 命令,初学者的你请慎用!因为你不知道这个命令是在何时且在什么场景下才能使用,实验测试环境可以随便。这个命令包含两个动作:shutdown abort 和 startup 。
SQL> startup force; ORACLE instance started. Total System Global Area 830930944 bytes Fixed Size 2232920 bytes Variable Size 591400360 bytes Database Buffers 234881024 bytes Redo Buffers 2416640 bytes Database mounted. Database opened.
查看日志片段
Wed May 22 11:54:00 2019 Shutting down instance (abort) License high water mark = 3 USER (ospid: 23787): terminating the instance Instance terminated by USER, pid = 23787 Wed May 22 11:54:01 2019 Instance shutdown complete
注意shutdown abort,它比较特殊,属于非一致性关闭,类似于服务器突然断电。生产库中使用这个命令,是很有可能丢失数据的,因为日志来不及写。shutdown abort 属于数据库关闭四种模式中的其中一种,在数据库关闭的四种模式博文中进行介绍。