PostgreSQL 源码分析之 CreateLockFile

前端之家收集整理的这篇文章主要介绍了PostgreSQL 源码分析之 CreateLockFile前端之家小编觉得挺不错的,现在分享给大家,也给大家做个参考。

在创建连接套接字之前,数据库主进程Postmaster需要保证当前只有1个Postmaster在运行,如何保证呢?这个要通过函数CreateLockFile来完成,通过函数CreateLockFile在PGDATA数据库目录中创建锁文件postmaster.pid,每次postmaster运行时,都会在PGDATA目录中创建这个文件(创建文件时设置了标志位O_EXCL)。

数据库停止工作时,会删除这个锁文件,在数据库成功启动时,会创建这个锁文件。可以看下启动时该锁文件内容

(1)文件postmaster.pid

[wln@localhost linux]$ pg_ctl -V
pg_ctl (Postgresql) 9.3beta2
[wln@localhost linux]$ cat /home/wln/postgres9.3/data/postmaster.pid
28179
/home/wln/postgres9.3/data
1411423233
5432
/tmp
localhost
5432001 4947977

[wln@localhost linux]$ pg_ctl stop -m f
waiting for server to shut down....... done
server stopped
[wln@localhost linux]$ pg_ctl start
server starting
[wln@localhost linux]$ cat /home/wln/postgres9.3/data/postmaster.pid
16304 --对应进程 PID
/home/wln/postgres9.3/data --对应PGDATA
1411461916
5432 --对应PGPORT
/tmp
localhost --对应PGHOST,本地执行默认localhost
5432001 4980745
[wln@localhost linux]$ ps ux | grep 16304 | grep -v grep
wln 16304 1.2 0.9 152576 10360 pts/3 S 16:45 0:00 /home/wln/postgres9.3/install/bin/postgres

postmaster.pid文件内容具体如下:

* As of 9.1,the contents of the data-directory lock file are:
*
* line #
* 1 postmaster PID (or negative of a standalone backend's PID)
* 2 data directory path
* 3 postmaster start timestamp (time_t representation)
* 4 port number
* 5 first Unix socket directory path (empty if none)
* 6 first listen_address (IP address or "*"; empty if no TCP port)
* 7 shared memory key (not present on Windows)
*
* Lines 6 and up are added via AddToDataDirLockFile() after initial file
* creation.
*
* The socket lock file,if used,has the same contents as lines 1-5.
*/
#define LOCK_FILE_LINE_PID 1
#define LOCK_FILE_LINE_DATA_DIR 2
#define LOCK_FILE_LINE_START_TIME 3
#define LOCK_FILE_LINE_PORT 4
#define LOCK_FILE_LINE_SOCKET_DIR 5
#define LOCK_FILE_LINE_LISTEN_ADDR 6
#define LOCK_FILE_LINE_SHMEM_KEY 7

(2)/tmp下内容

[wln@localhost linux]$ cat /tmp/.s.PGsql.5432.lock
16304
/home/wln/postgres9.3/data
1411461916
5432
/tmp

[wln@localhost tmp]$ ll .s.PGsql.5432
srwxrwxrwx 1 wln wln 0 09-23 16:45 .s.PGsql.5432

[wln@localhost tmp]$ cat .s.PGsql.5432
cat: .s.PGsql.5432: 没有那个设备或地址

--不知怎么报这个错误,该文件内容长度为0,那为什么要存在这个文件呢?不是已经有个.s.PGsql.5432.lock 文件了?

(3)函数CreateLockFile

代码的从上到下顺序对应《Posgresql数据库内核分析》P27内容介绍顺序

<span style="font-size:14px;">/*
 * Create a lockfile.
 *
 * filename is the path name of the lockfile to create.
 * amPostmaster is used to determine how to encode the output PID.
 * socketDir is the Unix socket directory path to include (possibly empty).
 * isDDLock and refName are used to determine what error message to produce.
 */
static void
CreateLockFile(const char *filename,bool amPostmaster,const char *socketDir,bool isDDLock,const char *refName)
{
	int			fd;
	char		buffer[MAXPGPATH * 2 + 256];
	int			ntries;
	int			len;
	int			encoded_pid;
	pid_t		other_pid;
	pid_t		my_pid,my_p_pid,my_gp_pid;
	const char *envvar;

	/*
	 * If the PID in the lockfile is our own PID or our parent's or
	 * grandparent's PID,then the file must be stale (probably left over from
	 * a prevIoUs system boot cycle).  We need to check this because of the
	 * likelihood that a reboot will assign exactly the same PID as we had in
	 * the prevIoUs reboot,or one that's only one or two counts larger and
	 * hence the lockfile's PID now refers to an ancestor shell process.  We
	 * allow pg_ctl to pass down its parent shell PID (our grandparent PID)
	 * via the environment variable PG_GRANDPARENT_PID; this is so that
	 * launching the postmaster via pg_ctl can be just as reliable as
	 * launching it directly.  There is no provision for detecting
	 * further-removed ancestor processes,but if the init script is written
	 * carefully then all but the immediate parent shell will be root-owned
	 * processes and so the kill test will fail with EPERM.  Note that we
	 * cannot get a false negative this way,because an existing postmaster
	 * would surely never launch a competing postmaster or pg_ctl process
	 * directly.
	 */
	my_pid = getpid();

#ifndef WIN32
	my_p_pid = getppid();
#else

	/*
	 * Windows hasn't got getppid(),but doesn't need it since it's not using
	 * real kill() either...
	 */
	my_p_pid = 0;
#endif

	envvar = getenv("PG_GRANDPARENT_PID");
	if (envvar)
		my_gp_pid = atoi(envvar);
	else
		my_gp_pid = 0;

	/*
	 * We need a loop here because of race conditions.	But don't loop forever
	 * (for example,a non-writable $PGDATA directory might cause a failure
	 * that won't go away).  100 tries seems like plenty.
	 */
	for (ntries = 0;; ntries++)
	{
		/*
		 * Try to create the lock file --- O_EXCL makes this atomic.
		 *
		 * Think not to make the file protection weaker than 0600.	See
		 * comments below.
		 */
		fd = open(filename,O_RDWR | O_CREAT | O_EXCL,0600);
		if (fd >= 0)
			break;				/* Success; exit the retry loop */

		/*
		 * Couldn't create the pid file. Probably it already exists.
		 */
		if ((errno != EEXIST && errno != EACCES) || ntries > 100)
			ereport(FATAL,(errcode_for_file_access(),errmsg("could not create lock file \"%s\": %m",filename)));

		/*
		 * Read the file to get the old owner's PID.  Note race condition
		 * here: file might have been deleted since we tried to create it.
		 */
		fd = open(filename,O_RDONLY,0600);
		if (fd < 0)
		{
			if (errno == ENOENT)
				continue;		/* race condition; try again */
			ereport(FATAL,errmsg("could not open lock file \"%s\": %m",filename)));
		}
		if ((len = read(fd,buffer,sizeof(buffer) - 1)) < 0)
			ereport(FATAL,errmsg("could not read lock file \"%s\": %m",filename)));
		close(fd);

		if (len == 0)
		{
			ereport(FATAL,(errcode(ERRCODE_LOCK_FILE_EXISTS),errmsg("lock file \"%s\" is empty",filename),errhint("Either another server is starting,or the lock file is the remnant of a prevIoUs server startup crash.")));
		}

		buffer[len] = '\0';
		encoded_pid = atoi(buffer);

		/* if pid < 0,the pid is for postgres,not postmaster */
		other_pid = (pid_t) (encoded_pid < 0 ? -encoded_pid : encoded_pid);

		if (other_pid <= 0)
			elog(FATAL,"bogus data in lock file \"%s\": \"%s\"",filename,buffer);

		/*
		 * Check to see if the other process still exists
		 *
		 * Per discussion above,my_pid,and my_gp_pid can be
		 * ignored as false matches.
		 *
		 * Normally kill() will fail with ESRCH if the given PID doesn't
		 * exist.
		 *
		 * We can treat the EPERM-error case as okay because that error
		 * implies that the existing process has a different userid than we
		 * do,which means it cannot be a competing postmaster.  A postmaster
		 * cannot successfully attach to a data directory owned by a userid
		 * other than its own.	(This is now checked directly in
		 * checkDataDir(),but has been true for a long time because of the
		 * restriction that the data directory isn't group- or
		 * world-accessible.)  Also,since we create the lockfiles mode 600,* we'd have Failed above if the lockfile belonged to another userid
		 * --- which means that whatever process kill() is reporting about
		 * isn't the one that made the lockfile.  (NOTE: this last
		 * consideration is the only one that keeps us from blowing away a
		 * Unix socket file belonging to an instance of Postgres being run by
		 * someone else,at least on machines where /tmp hasn't got a
		 * stickybit.)
		 */
		if (other_pid != my_pid && other_pid != my_p_pid &&
			other_pid != my_gp_pid)
		{
			if (kill(other_pid,0) == 0 ||
				(errno != ESRCH && errno != EPERM))
			{
				/* lockfile belongs to a live process */
				ereport(FATAL,errmsg("lock file \"%s\" already exists",isDDLock ?
						 (encoded_pid < 0 ?
						  errhint("Is another postgres (PID %d) running in data directory \"%s\"?",(int) other_pid,refName) :
						  errhint("Is another postmaster (PID %d) running in data directory \"%s\"?",refName)) :
						 (encoded_pid < 0 ?
						  errhint("Is another postgres (PID %d) using socket file \"%s\"?",refName) :
						  errhint("Is another postmaster (PID %d) using socket file \"%s\"?",refName))));
			}
		}

		/*
		 * No,the creating process did not exist.	However,it could be that
		 * the postmaster crashed (or more likely was kill -9'd by a clueless
		 * admin) but has left orphan backends behind.	Check for this by
		 * looking to see if there is an associated shmem segment that is
		 * still in use.
		 *
		 * Note: because postmaster.pid is written in multiple steps,we might
		 * not find the shmem ID values in it; we can't treat that as an
		 * error.
		 */
		if (isDDLock)
		{
			char	   *ptr = buffer;
			unsigned long id1,id2;
			int			lineno;

			for (lineno = 1; lineno < LOCK_FILE_LINE_SHMEM_KEY; lineno++)
			{
				if ((ptr = strchr(ptr,'\n')) == NULL)
					break;
				ptr++;
			}

			if (ptr != NULL &&
				sscanf(ptr,"%lu %lu",&id1,&id2) == 2)
			{
				if (PGSharedMemoryIsInUse(id1,id2))
					ereport(FATAL,errmsg("pre-existing shared memory block "
									"(key %lu,ID %lu) is still in use",id1,id2),errhint("If you're sure there are no old "
									 "server processes still running,remove "
									 "the shared memory block "
									 "or just delete the file \"%s\".",filename)));
			}
		}

		/*
		 * Looks like nobody's home.  Unlink the file and try again to create
		 * it.	Need a loop because of possible race condition against other
		 * would-be creators.
		 */
		if (unlink(filename) < 0)
			ereport(FATAL,errmsg("could not remove old lock file \"%s\": %m",errhint("The file seems accidentally left over,but "
						   "it could not be removed. Please remove the file "
							 "by hand and try again.")));
	}

	/*
	 * Successfully created the file,now fill it.	See comment in miscadmin.h
	 * about the contents.	Note that we write the same first five lines into
	 * both datadir and socket lockfiles; although more stuff may get added to
	 * the datadir lockfile later.
	 */
	snprintf(buffer,sizeof(buffer),"%d\n%s\n%ld\n%d\n%s\n",amPostmaster ? (int) my_pid : -((int) my_pid),DataDir,(long) MyStartTime,PostPortNumber,socketDir);

	/*
	 * In a standalone backend,the next line (LOCK_FILE_LINE_LISTEN_ADDR)
	 * will never receive data,so fill it in as empty now.
	 */
	if (isDDLock && !amPostmaster)
		strlcat(buffer,"\n",sizeof(buffer));

	errno = 0;
	if (write(fd,strlen(buffer)) != strlen(buffer))
	{
		int			save_errno = errno;

		close(fd);
		unlink(filename);
		/* if write didn't set errno,assume problem is no disk space */
		errno = save_errno ? save_errno : ENOSPC;
		ereport(FATAL,errmsg("could not write lock file \"%s\": %m",filename)));
	}
	if (pg_fsync(fd) != 0)
	{
		int			save_errno = errno;

		close(fd);
		unlink(filename);
		errno = save_errno;
		ereport(FATAL,filename)));
	}
	if (close(fd) != 0)
	{
		int			save_errno = errno;

		unlink(filename);
		errno = save_errno;
		ereport(FATAL,filename)));
	}

	/*
	 * Arrange to unlink the lock file(s) at proc_exit.  If this is the first
	 * one,set up the on_proc_exit function to do it; then add this lock file
	 * to the list of files to unlink.
	 */
	if (lock_files == NIL)
		on_proc_exit(UnlinkLockFiles,0);

	lock_files = lappend(lock_files,pstrdup(filename));
}</span>

(4)learn CreateLockFile

.1 进程、环境变量
取得当前进程PID my_pid = getpid();
取得其父进程PID my_p_pid = getppid();
取得环境变量的值 const char *envvar = getenv("PG_GRANDPARENT_PID");

.2. 文件标识符
有关文件不存在则创建,为什么要添加标志位 O_EXCL
fd = open(filename,0600);
::open(pathname,O_RDWR | O_CREAT,0666); 打开成功,返回一个大于0的fd
open(pathname,O_RDWR | O_CREAT | O_EXCL,0666); 打开失败,返回-1
可参考问题讨论:http://bbs.csdn.net/topics/350001997

.3.出错标志位 errno
errno 是记录系统的最后一次错误代码代码是一个int型的值,在errno.h中定义
errno 只是一个INT 数值,那么如何查看该数值对应的是什么错误描述呢?这就需要和strerror进行关联,详细参见文章
#include <stdio.h>
#include <string.h>
#include <errno.h>
#include <fcntl.h>
int main(void)
{
int   fd;
extern int errno;

if((fd = open("/dev/dsp2",O_WRONLY)) < 0)
{
   printf("errno=%d\n",errno);
char * mesg = strerror(errno);
   printf("Mesg:%s\n",mesg);
}
  
exit(0);
}
[wln@localhost linux]$ ./erron2
errno=2
Mesg:No such file or directory

.4. kill函数

kill(传送信号给指定的进程)
相关函数 raise,signal
表头文件 #include<sys/types.h>
#include<signal.h>
定义函数 int kill(pid_t pid,int sig);
函数说明
kill()可以用来送参数sig指定的信号给参数pid指定的进程。参数
pid有几种情况:
pid>0 将信号传给进程识别码为pid 的进程。
pid=0 将信号传给和目前进程相同进程组的所有进程
pid=-1 将信号广播传送给系统内所有的进程
pid<0 将信号传给进程组识别码为pid绝对值的所有进程
参数sig代表的信号编号可参考附录D
返回值 执行成功则返回0,如果有错误则返回-1。
错误代码 EINVAL 参数sig 不合法
ESRCH 参数pid 所指定的进程或进程组不存在
EPERM 权限不够无法传送信号给指定进程



.5 strchr
strchr函数原型:extern char *strchr(const char *s,char c);查找字符串s中首次出现字符c的位置

猜你在找的Postgre SQL相关文章