postmaster.c

PostgreSQL 8.1.4的源码 适用于Linux下的开源数据库系统

			/* Tell pgarch to shut down too; nothing left for it to do */
			if (PgArchPID != 0)
				kill(PgArchPID, SIGQUIT);
			/* Tell pgstat to shut down too; nothing left for it to do */
			if (PgStatPID != 0)
				kill(PgStatPID, SIGQUIT);
			break;

		case SIGQUIT:

			/*
			 * Immediate Shutdown:
			 *
			 * abort all children with SIGQUIT and exit without attempt to
			 * properly shut down data base system.
			 */
			ereport(LOG,
					(errmsg("received immediate shutdown request")));
			if (StartupPID != 0)
				kill(StartupPID, SIGQUIT);
			if (BgWriterPID != 0)
				kill(BgWriterPID, SIGQUIT);
			if (AutoVacPID != 0)
				kill(AutoVacPID, SIGQUIT);
			if (PgArchPID != 0)
				kill(PgArchPID, SIGQUIT);
			if (PgStatPID != 0)
				kill(PgStatPID, SIGQUIT);
			if (DLGetHead(BackendList))
				SignalChildren(SIGQUIT);
			ExitPostmaster(0);
			break;
	}

	PG_SETMASK(&UnBlockSig);

	errno = save_errno;
}

/*
 * Reaper -- signal handler to cleanup after a backend (child) dies.
 */
static void
reaper(SIGNAL_ARGS)
{
	int			save_errno = errno;

#ifdef HAVE_WAITPID
	int			status;			/* backend exit status */
#else
#ifndef WIN32
	union wait	status;			/* backend exit status */
#endif
#endif
	int			exitstatus;
	int			pid;			/* process id of dead backend */

	PG_SETMASK(&BlockSig);

	ereport(DEBUG4,
			(errmsg_internal("reaping dead processes")));
#ifdef HAVE_WAITPID
	while ((pid = waitpid(-1, &status, WNOHANG)) > 0)
	{
		exitstatus = status;
#else
#ifndef WIN32
	while ((pid = wait3(&status, WNOHANG, NULL)) > 0)
	{
		exitstatus = status.w_status;
#else
	while ((pid = win32_waitpid(&exitstatus)) > 0)
	{
		/*
		 * We need to do this here, and not in CleanupBackend, since this is
		 * to be called on all children when we are done with them. Could move
		 * to LogChildExit, but that seems like asking for future trouble...
		 */
		win32_RemoveChild(pid);
#endif   /* WIN32 */
#endif   /* HAVE_WAITPID */

		/*
		 * Check if this child was a startup process.
		 */
		if (StartupPID != 0 && pid == StartupPID)
		{
			StartupPID = 0;
			if (exitstatus != 0)
			{
				LogChildExit(LOG, _("startup process"),
							 pid, exitstatus);
				ereport(LOG,
				(errmsg("aborting startup due to startup process failure")));
				ExitPostmaster(1);
			}

			/*
			 * Startup succeeded - we are done with system startup or
			 * recovery.
			 */
			FatalError = false;

			/*
			 * Load the flat authorization file into postmaster's cache. The
			 * startup process has recomputed this from the database contents,
			 * so we wait till it finishes before loading it.
			 */
			load_role();

			/*
			 * Crank up the background writer.	It doesn't matter if this
			 * fails, we'll just try again later.
			 */
			Assert(BgWriterPID == 0);
			BgWriterPID = StartBackgroundWriter();

			/*
			 * Go to shutdown mode if a shutdown request was pending.
			 * Otherwise, try to start the archiver and stats collector too.
			 * (We could, but don't, try to start autovacuum here.)
			 */
			if (Shutdown > NoShutdown && BgWriterPID != 0)
				kill(BgWriterPID, SIGUSR2);
			else if (Shutdown == NoShutdown)
			{
				if (XLogArchivingActive() && PgArchPID == 0)
					PgArchPID = pgarch_start();
				if (PgStatPID == 0)
					PgStatPID = pgstat_start();
			}

			continue;
		}

		/*
		 * Was it the bgwriter?
		 */
		if (BgWriterPID != 0 && pid == BgWriterPID)
		{
			BgWriterPID = 0;
			if (exitstatus == 0 && Shutdown > NoShutdown && !FatalError &&
				!DLGetHead(BackendList) && AutoVacPID == 0)
			{
				/*
				 * Normal postmaster exit is here: we've seen normal exit of
				 * the bgwriter after it's been told to shut down. We expect
				 * that it wrote a shutdown checkpoint.  (If for some reason
				 * it didn't, recovery will occur on next postmaster start.)
				 *
				 * Note: we do not wait around for exit of the archiver or
				 * stats processes.  They've been sent SIGQUIT by this point,
				 * and in any case contain logic to commit hara-kiri if they
				 * notice the postmaster is gone.
				 */
				ExitPostmaster(0);
			}

			/*
			 * Any unexpected exit of the bgwriter is treated as a crash.
			 */
			HandleChildCrash(pid, exitstatus,
							 _("background writer process"));
			continue;
		}

		/*
		 * Was it the autovacuum process?  Normal exit can be ignored; we'll
		 * start a new one at the next iteration of the postmaster's main
		 * loop, if necessary.
		 *
		 * An unexpected exit must crash the system.
		 */
		if (AutoVacPID != 0 && pid == AutoVacPID)
		{
			AutoVacPID = 0;
			autovac_stopped();
			/* Tell the collector about process termination */
			pgstat_beterm(pid);

			if (exitstatus != 0)
				HandleChildCrash(pid, exitstatus,
								 _("autovacuum process"));
			continue;
		}

		/*
		 * Was it the archiver?  If so, just try to start a new one; no need
		 * to force reset of the rest of the system.  (If fail, we'll try
		 * again in future cycles of the main loop.)
		 */
		if (PgArchPID != 0 && pid == PgArchPID)
		{
			PgArchPID = 0;
			if (exitstatus != 0)
				LogChildExit(LOG, _("archiver process"),
							 pid, exitstatus);
			if (XLogArchivingActive() &&
				StartupPID == 0 && !FatalError && Shutdown == NoShutdown)
				PgArchPID = pgarch_start();
			continue;
		}

		/*
		 * Was it the statistics collector?  If so, just try to start a new
		 * one; no need to force reset of the rest of the system.  (If fail,
		 * we'll try again in future cycles of the main loop.)
		 */
		if (PgStatPID != 0 && pid == PgStatPID)
		{
			PgStatPID = 0;
			if (exitstatus != 0)
				LogChildExit(LOG, _("statistics collector process"),
							 pid, exitstatus);
			if (StartupPID == 0 && !FatalError && Shutdown == NoShutdown)
				PgStatPID = pgstat_start();
			continue;
		}

		/* Was it the system logger? try to start a new one */
		if (SysLoggerPID != 0 && pid == SysLoggerPID)
		{
			SysLoggerPID = 0;
			/* for safety's sake, launch new logger *first* */
			SysLoggerPID = SysLogger_Start();
			if (exitstatus != 0)
				LogChildExit(LOG, _("system logger process"),
							 pid, exitstatus);
			continue;
		}

		/*
		 * Else do standard backend child cleanup.
		 */
		CleanupBackend(pid, exitstatus);
	}							/* loop over pending child-death reports */

	if (FatalError)
	{
		/*
		 * Wait for all important children to exit, then reset shmem and
		 * StartupDataBase.  (We can ignore the archiver and stats processes
		 * here since they are not connected to shmem.)
		 */
		if (DLGetHead(BackendList) || StartupPID != 0 || BgWriterPID != 0 ||
			AutoVacPID != 0)
			goto reaper_done;
		ereport(LOG,
				(errmsg("all server processes terminated; reinitializing")));

		shmem_exit(0);
		reset_shared(PostPortNumber);

		StartupPID = StartupDataBase();

		goto reaper_done;
	}

	if (Shutdown > NoShutdown)
	{
		if (DLGetHead(BackendList) || StartupPID != 0 || AutoVacPID != 0)
			goto reaper_done;
		/* Start the bgwriter if not running */
		if (BgWriterPID == 0)
			BgWriterPID = StartBackgroundWriter();
		/* And tell it to shut down */
		if (BgWriterPID != 0)
			kill(BgWriterPID, SIGUSR2);
		/* Tell pgarch to shut down too; nothing left for it to do */
		if (PgArchPID != 0)
			kill(PgArchPID, SIGQUIT);
		/* Tell pgstat to shut down too; nothing left for it to do */
		if (PgStatPID != 0)
			kill(PgStatPID, SIGQUIT);
	}

reaper_done:
	PG_SETMASK(&UnBlockSig);

	errno = save_errno;
}


/*
 * CleanupBackend -- cleanup after terminated backend.
 *
 * Remove all local state associated with backend.
 */
static void
CleanupBackend(int pid,
			   int exitstatus)	/* child's exit status. */
{
	Dlelem	   *curr;

	LogChildExit(DEBUG2, _("server process"), pid, exitstatus);

	/*
	 * If a backend dies in an ugly way (i.e. exit status not 0) then we must
	 * signal all other backends to quickdie.  If exit status is zero we
	 * assume everything is hunky dory and simply remove the backend from the
	 * active backend list.
	 */
	if (exitstatus != 0)
	{
		HandleChildCrash(pid, exitstatus, _("server process"));
		return;
	}

	for (curr = DLGetHead(BackendList); curr; curr = DLGetSucc(curr))
	{
		Backend    *bp = (Backend *) DLE_VAL(curr);

		if (bp->pid == pid)
		{
			DLRemove(curr);
			free(bp);
			DLFreeElem(curr);
#ifdef EXEC_BACKEND
			ShmemBackendArrayRemove(pid);
#endif
			/* Tell the collector about backend termination */
			pgstat_beterm(pid);
			break;
		}
	}
}

/*
 * HandleChildCrash -- cleanup after failed backend, bgwriter, or autovacuum.
 *
 * The objectives here are to clean up our local state about the child
 * process, and to signal all other remaining children to quickdie.
 */
static void
HandleChildCrash(int pid, int exitstatus, const char *procname)
{
	Dlelem	   *curr,
			   *next;
	Backend    *bp;

	/*
	 * Make log entry unless there was a previous crash (if so, nonzero exit
	 * status is to be expected in SIGQUIT response; don't clutter log)
	 */
	if (!FatalError)
	{
		LogChildExit(LOG, procname, pid, exitstatus);
		ereport(LOG,
				(errmsg("terminating any other active server processes")));
	}

	/* Process regular backends */
	for (curr = DLGetHead(BackendList); curr; curr = next)
	{
		next = DLGetSucc(curr);
		bp = (Backend *) DLE_VAL(curr);
		if (bp->pid == pid)
		{
			/*
			 * Found entry for freshly-dead backend, so remove it.
			 */
			DLRemove(curr);
			free(bp);
			DLFreeElem(curr);
#ifdef EXEC_BACKEND
			ShmemBackendArrayRemove(pid);
#endif
			/* Tell the collector about backend termination */
			pgstat_beterm(pid);
			/* Keep looping so we can signal remaining backends */
		}
		else
		{
			/*
			 * This backend is still alive.  Unless we did so already, tell it
			 * to commit hara-kiri.
			 *
			 * SIGQUIT is the special signal that says exit without proc_exit
			 * and let the user know what's going on. But if SendStop is set
			 * (-s on command line), then we send SIGSTOP instead, so that we
			 * can get core dumps from all backends by hand.
			 */
			if (!FatalError)
			{
				ereport(DEBUG2,
						(errmsg_internal("sending %s to process %d",
										 (SendStop ? "SIGSTOP" : "SIGQUIT"),
										 (int) bp->pid)));
				kill(bp->pid, (SendStop ? SIGSTOP : SIGQUIT));
			}
		}
	}

	/* Take care of the bgwriter too */
	if (pid == BgWriterPID)
		BgWriterPID = 0;
	else if (BgWriterPID != 0 && !FatalError)
	{
		ereport(DEBUG2,
				(errmsg_internal("sending %s to process %d",
								 (SendStop ? "SIGSTOP" : "SIGQUIT"),
								 (int) BgWriterPID)));
		kill(BgWriterPID, (SendStop ? SIGSTOP : SIGQUIT));
	}

	/* Take care of the autovacuum daemon too */
	if (pid == AutoVacPID)
		AutoVacPID = 0;
	else if (AutoVacPID != 0 && !FatalError)
	{
		ereport(DEBUG2,
				(errmsg_internal("sending %s to process %d",
								 (SendStop ? "SIGSTOP" : "SIGQUIT"),
								 (int) AutoVacPID)));
		kill(AutoVacPID, (SendStop ? SIGSTOP : SIGQUIT));
	}

	/* Force a power-cycle of the pgarch process too */
	/* (Shouldn't be necessary, but just for luck) */
	if (PgArchPID != 0 && !FatalError)
	{
		ereport(DEBUG2,
				(errmsg_internal("sending %s to process %d",
								 "SIGQUIT",
								 (int) PgArchPID)));
		kill(PgArchPID, SIGQUIT);
	}

	/* Force a power-cycle of the pgstat processes too */
	/* (Shouldn't be necessary, but just for luck) */
	if (PgStatPID != 0 && !FatalError)
	{
		ereport(DEBUG2,
				(errmsg_internal("sending %s to process %d",
								 "SIGQUIT",
								 (int) PgStatPID)));
		kill(PgStatPID, SIGQUIT);
	}

	/* We do NOT restart the syslogger */

	FatalError = true;
}

/*
 * Log the death of a child process.
 */
static void
LogChildExit(int lev, const char *procname, int pid, int exitstatus)
{
	if (WIFEXITED(exitstatus))
		ereport(lev,

		/*
		 * translator: %s is a noun phrase describing a child process, such as
		 * "server process"
		 */
				(errmsg("%s (PID %d) exited with exit code %d",
						procname, pid, WEXITSTATUS(exitstatus))));
	else if (WIFSIGNALED(exitstatus))
		ereport(lev,

		/*
		 * translator: %s is a noun phrase describing a child process, such as
		 * "server process"
		 */
				(errmsg("%s (PID %d) was terminated by signal %d",
						procname, pid, WTERMSIG(exitstatus))));
	else
		ereport(lev,

		/*
		 * translator: %s is a noun phrase describing a child process, such as
		 * "server process"
		 */
				(errmsg("%s (PID %d) exited with unexpected status %d",
						procname, pid, exitstatus)));
}

/*
 * Send a signal to all backend children (but NOT special children)
 */
static void
SignalChildren(int signal)
{
	Dlelem	   *curr;

	for (curr = DLGetHead(BackendList); curr; curr = DLGetSucc(cu