Nginx进程模型
nginx 采用了多进程的进程模型，其中有且只有一个 master 进程，可以配置多个 worker 进程，还有 cache 进程这里暂不讨论。master 进程主要负责监控 worker 进程，当 worker 进程异常退出时，master 进程能够通过信号感知到并重启拉起 worker 进程，保证了服务的高可用性，同时在每个 worker 进程都运行着事件循环，可以很好的利用多核性能。
当然 nginx 也支持单进程模式，不过一般只是在调试的时候才使用，生产上使用的都是多进程模式。
Nginx 进程结构体
- pid 用于存放 worker 进程的进程 id
- status 存放 waitpid 返回的状态
- channel 用于父子进程间通信，使用的 unix 域套接字
- proc 是进程的处理函数，data 作为函数的参数传入
- name 为进程名称
- 最后几个用位域表示进程当前的状态，其中：
- respawn 表示重新生成的进程
- just_spawn 表示进程刚刚产生
- detached 表示进程与父进程分离
- exiting 表示进程正在退出
- exited 表示进程已经退出
typedef struct {
ngx_pid_t pid;
int status;
ngx_socket_t channel[2];
ngx_spawn_proc_pt proc;
void *data;
char *name;
unsigned respawn:1;
unsigned just_spawn:1;
unsigned detached:1;
unsigned exiting:1;
unsigned exited:1;
} ngx_process_t;master 进程主循环
首先设置关注的信号及其信号处理函数 nginx_signal_handler ，然后根据配置中的 worker 进程数量派生出子进程，最后在一个无限 for 循环中调用 sigsuspend阻塞以等待信号到达。
当信号到达后，在信号处理函数中会设置几个全局标志变量的值，然后 sigsuspend解除阻塞后判断全局标志变量，例如 ngx_reap 如果设置为 1 表示有 worker 进程退出了，需要重新拉起来，ngx_terminate 表示整个 nginx 收到了停止命令，这时需要将 worker 进程都退出掉再退出 master 进程。
void
ngx_master_process_cycle(ngx_cycle_t *cycle)
{
char *title;
u_char *p;
size_t size;
ngx_int_t i;
ngx_uint_t sigio;
sigset_t set;
struct itimerval itv;
ngx_uint_t live;
ngx_msec_t delay;
ngx_core_conf_t *ccf;
sigemptyset(&set);
sigaddset(&set, SIGCHLD);
sigaddset(&set, SIGALRM);
sigaddset(&set, SIGIO);
sigaddset(&set, SIGINT);
sigaddset(&set, ngx_signal_value(NGX_RECONFIGURE_SIGNAL));
sigaddset(&set, ngx_signal_value(NGX_REOPEN_SIGNAL));
sigaddset(&set, ngx_signal_value(NGX_NOACCEPT_SIGNAL));
sigaddset(&set, ngx_signal_value(NGX_TERMINATE_SIGNAL));
sigaddset(&set, ngx_signal_value(NGX_SHUTDOWN_SIGNAL));
sigaddset(&set, ngx_signal_value(NGX_CHANGEBIN_SIGNAL));
if (sigprocmask(SIG_BLOCK, &set, NULL) == -1) {
ngx_log_error(NGX_LOG_ALERT, cycle->log, ngx_errno,
"sigprocmask() failed");
}
sigemptyset(&set);
size = sizeof(master_process);
for (i = 0; i < ngx_argc; i++) {
size += ngx_strlen(ngx_argv[i]) + 1;
}
title = ngx_pnalloc(cycle->pool, size);
if (title == NULL) {
/* fatal */
exit(2);
}
p = ngx_cpymem(title, master_process, sizeof(master_process) - 1);
for (i = 0; i < ngx_argc; i++) {
*p++ = ' ';
p = ngx_cpystrn(p, (u_char *) ngx_argv[i], size);
}
ngx_setproctitle(title);
ccf = (ngx_core_conf_t *) ngx_get_conf(cycle->conf_ctx, ngx_core_module);
ngx_start_worker_processes(cycle, ccf->worker_processes,
NGX_PROCESS_RESPAWN);
ngx_start_cache_manager_processes(cycle, 0);
ngx_new_binary = 0;
delay = 0;
sigio = 0;
live = 1;
for ( ;; ) {
if (delay) {
if (ngx_sigalrm) {
sigio = 0;
delay *= 2;
ngx_sigalrm = 0;
}
ngx_log_debug1(NGX_LOG_DEBUG_EVENT, cycle->log, 0,
"termination cycle: %M", delay);
itv.it_interval.tv_sec = 0;
itv.it_interval.tv_usec = 0;
itv.it_value.tv_sec = delay / 1000;
itv.it_value.tv_usec = (delay % 1000 ) * 1000;
if (setitimer(ITIMER_REAL, &itv, NULL) == -1) {
ngx_log_error(NGX_LOG_ALERT, cycle->log, ngx_errno,
"setitimer() failed");
}
}
ngx_log_debug0(NGX_LOG_DEBUG_EVENT, cycle->log, 0, "sigsuspend");
sigsuspend(&set);
ngx_time_update();
ngx_log_debug1(NGX_LOG_DEBUG_EVENT, cycle->log, 0,
"wake up, sigio %i", sigio);
if (ngx_reap) {
ngx_reap = 0;
ngx_log_debug0(NGX_LOG_DEBUG_EVENT, cycle->log, 0, "reap children");
live = ngx_reap_children(cycle);
}
if (!live && (ngx_terminate || ngx_quit)) {
ngx_master_process_exit(cycle);
}
if (ngx_terminate) {
if (delay == 0) {
delay = 50;
}
if (sigio) {
sigio--;
continue;
}
sigio = ccf->worker_processes + 2 /* cache processes */;
if (delay > 1000) {
ngx_signal_worker_processes(cycle, SIGKILL);
} else {
ngx_signal_worker_processes(cycle,
ngx_signal_value(NGX_TERMINATE_SIGNAL));
}
continue;
}
if (ngx_quit) {
ngx_signal_worker_processes(cycle,
ngx_signal_value(NGX_SHUTDOWN_SIGNAL));
ngx_close_listening_sockets(cycle);
continue;
}
if (ngx_reconfigure) {
ngx_reconfigure = 0;
if (ngx_new_binary) {
ngx_start_worker_processes(cycle, ccf->worker_processes,
NGX_PROCESS_RESPAWN);
ngx_start_cache_manager_processes(cycle, 0);
ngx_noaccepting = 0;
continue;
}
ngx_log_error(NGX_LOG_NOTICE, cycle->log, 0, "reconfiguring");
cycle = ngx_init_cycle(cycle);
if (cycle == NULL) {
cycle = (ngx_cycle_t *) ngx_cycle;
continue;
}
ngx_cycle = cycle;
ccf = (ngx_core_conf_t *) ngx_get_conf(cycle->conf_ctx,
ngx_core_module);
ngx_start_worker_processes(cycle, ccf->worker_processes,
NGX_PROCESS_JUST_RESPAWN);
ngx_start_cache_manager_processes(cycle, 1);
/* allow new processes to start */
ngx_msleep(100);
live = 1;
ngx_signal_worker_processes(cycle,
ngx_signal_value(NGX_SHUTDOWN_SIGNAL));
}
if (ngx_restart) {
ngx_restart = 0;
ngx_start_worker_processes(cycle, ccf->worker_processes,
NGX_PROCESS_RESPAWN);
ngx_start_cache_manager_processes(cycle, 0);
live = 1;
}
if (ngx_reopen) {
ngx_reopen = 0;
ngx_log_error(NGX_LOG_NOTICE, cycle->log, 0, "reopening logs");
ngx_reopen_files(cycle, ccf->user);
ngx_signal_worker_processes(cycle,
ngx_signal_value(NGX_REOPEN_SIGNAL));
}
if (ngx_change_binary) {
ngx_change_binary = 0;
ngx_log_error(NGX_LOG_NOTICE, cycle->log, 0, "changing binary");
ngx_new_binary = ngx_exec_new_binary(cycle, ngx_argv);
}
if (ngx_noaccept) {
ngx_noaccept = 0;
ngx_noaccepting = 1;
ngx_signal_worker_processes(cycle,
ngx_signal_value(NGX_SHUTDOWN_SIGNAL));
}
}
}worker 进程派生
进程派生的主要逻辑在 ngx_spawn_process中，初始时从进程结构体数组中找到一个 pid 为 -1 的位置，然后在这个位置上派生一个 worker 进程，master 进程将 worker 进程的 pid 记录到数组对应位置，worker 进程则直接开始执行处理函数，也就是阻塞在事件循环等待连接到来。
如果是重新拉起的进程，则参数中的 respwn 表示需要拉起的进程在进程结构体数组中的下标，重新拉起后在此位置重新赋值即可。
ngx_pid_t
ngx_spawn_process(ngx_cycle_t *cycle, ngx_spawn_proc_pt proc, void *data,
char *name, ngx_int_t respawn)
{
u_long on;
ngx_pid_t pid;
ngx_int_t s;
if (respawn >= 0) {
s = respawn;
} else {
for (s = 0; s < ngx_last_process; s++) {
if (ngx_processes[s].pid == -1) {
break;
}
}
if (s == NGX_MAX_PROCESSES) {
ngx_log_error(NGX_LOG_ALERT, cycle->log, 0,
"no more than %d processes can be spawned",
NGX_MAX_PROCESSES);
return NGX_INVALID_PID;
}
}
if (respawn != NGX_PROCESS_DETACHED) {
/* Solaris 9 still has no AF_LOCAL */
if (socketpair(AF_UNIX, SOCK_STREAM, 0, ngx_processes[s].channel) == -1)
{
ngx_log_error(NGX_LOG_ALERT, cycle->log, ngx_errno,
"socketpair() failed while spawning \"%s\"", name);
return NGX_INVALID_PID;
}
ngx_log_debug2(NGX_LOG_DEBUG_CORE, cycle->log, 0,
"channel %d:%d",
ngx_processes[s].channel[0],
ngx_processes[s].channel[1]);
if (ngx_nonblocking(ngx_processes[s].channel[0]) == -1) {
ngx_log_error(NGX_LOG_ALERT, cycle->log, ngx_errno,
ngx_nonblocking_n " failed while spawning \"%s\"",
name);
ngx_close_channel(ngx_processes[s].channel, cycle->log);
return NGX_INVALID_PID;
}
if (ngx_nonblocking(ngx_processes[s].channel[1]) == -1) {
ngx_log_error(NGX_LOG_ALERT, cycle->log, ngx_errno,
ngx_nonblocking_n " failed while spawning \"%s\"",
name);
ngx_close_channel(ngx_processes[s].channel, cycle->log);
return NGX_INVALID_PID;
}
on = 1;
if (ioctl(ngx_processes[s].channel[0], FIOASYNC, &on) == -1) {
ngx_log_error(NGX_LOG_ALERT, cycle->log, ngx_errno,
"ioctl(FIOASYNC) failed while spawning \"%s\"", name);
ngx_close_channel(ngx_processes[s].channel, cycle->log);
return NGX_INVALID_PID;
}
if (fcntl(ngx_processes[s].channel[0], F_SETOWN, ngx_pid) == -1) {
ngx_log_error(NGX_LOG_ALERT, cycle->log, ngx_errno,
"fcntl(F_SETOWN) failed while spawning \"%s\"", name);
ngx_close_channel(ngx_processes[s].channel, cycle->log);
return NGX_INVALID_PID;
}
if (fcntl(ngx_processes[s].channel[0], F_SETFD, FD_CLOEXEC) == -1) {
ngx_log_error(NGX_LOG_ALERT, cycle->log, ngx_errno,
"fcntl(FD_CLOEXEC) failed while spawning \"%s\"",
name);
ngx_close_channel(ngx_processes[s].channel, cycle->log);
return NGX_INVALID_PID;
}
if (fcntl(ngx_processes[s].channel[1], F_SETFD, FD_CLOEXEC) == -1) {
ngx_log_error(NGX_LOG_ALERT, cycle->log, ngx_errno,
"fcntl(FD_CLOEXEC) failed while spawning \"%s\"",
name);
ngx_close_channel(ngx_processes[s].channel, cycle->log);
return NGX_INVALID_PID;
}
ngx_channel = ngx_processes[s].channel[1];
} else {
ngx_processes[s].channel[0] = -1;
ngx_processes[s].channel[1] = -1;
}
ngx_process_slot = s;
pid = fork();
switch (pid) {
case -1:
ngx_log_error(NGX_LOG_ALERT, cycle->log, ngx_errno,
"fork() failed while spawning \"%s\"", name);
ngx_close_channel(ngx_processes[s].channel, cycle->log);
return NGX_INVALID_PID;
case 0:
ngx_parent = ngx_pid;
ngx_pid = ngx_getpid();
proc(cycle, data);
break;
default:
break;
}
ngx_log_error(NGX_LOG_NOTICE, cycle->log, 0, "start %s %P", name, pid);
ngx_processes[s].pid = pid;
ngx_processes[s].exited = 0;
if (respawn >= 0) {
return pid;
}
ngx_processes[s].proc = proc;
ngx_processes[s].data = data;
ngx_processes[s].name = name;
ngx_processes[s].exiting = 0;
switch (respawn) {
case NGX_PROCESS_NORESPAWN:
ngx_processes[s].respawn = 0;
ngx_processes[s].just_spawn = 0;
ngx_processes[s].detached = 0;
break;
case NGX_PROCESS_JUST_SPAWN:
ngx_processes[s].respawn = 0;
ngx_processes[s].just_spawn = 1;
ngx_processes[s].detached = 0;
break;
case NGX_PROCESS_RESPAWN:
ngx_processes[s].respawn = 1;
ngx_processes[s].just_spawn = 0;
ngx_processes[s].detached = 0;
break;
case NGX_PROCESS_JUST_RESPAWN:
ngx_processes[s].respawn = 1;
ngx_processes[s].just_spawn = 1;
ngx_processes[s].detached = 0;
break;
case NGX_PROCESS_DETACHED:
ngx_processes[s].respawn = 0;
ngx_processes[s].just_spawn = 0;
ngx_processes[s].detached = 1;
break;
}
if (s == ngx_last_process) {
ngx_last_process++;
}
return pid;
}worker 进程退出后重启
子进程退出后，父进程将会收到 SIGCHLD ，此时需要调用 waitpid 以等待子进程退出，避免僵尸进程产生，在 ngx_siganl_handler 信号处理函数中，当收到此信号时，会调用 ngx_process_get_status 以更新子进程的状态。
static void
ngx_process_get_status(void)
{
int status;
char *process;
ngx_pid_t pid;
ngx_err_t err;
ngx_int_t i;
ngx_uint_t one;
one = 0;
for ( ;; ) {
pid = waitpid(-1, &status, WNOHANG);
if (pid == 0) {
return;
}
if (pid == -1) {
err = ngx_errno;
if (err == NGX_EINTR) {
continue;
}
if (err == NGX_ECHILD && one) {
return;
}
/*
* Solaris always calls the signal handler for each exited process
* despite waitpid() may be already called for this process.
*
* When several processes exit at the same time FreeBSD may
* erroneously call the signal handler for exited process
* despite waitpid() may be already called for this process.
*/
if (err == NGX_ECHILD) {
ngx_log_error(NGX_LOG_INFO, ngx_cycle->log, err,
"waitpid() failed");
return;
}
ngx_log_error(NGX_LOG_ALERT, ngx_cycle->log, err,
"waitpid() failed");
return;
}
one = 1;
process = "unknown process";
for (i = 0; i < ngx_last_process; i++) {
if (ngx_processes[i].pid == pid) {
ngx_processes[i].status = status;
ngx_processes[i].exited = 1;
process = ngx_processes[i].name;
break;
}
}
if (WTERMSIG(status)) {
#ifdef WCOREDUMP
ngx_log_error(NGX_LOG_ALERT, ngx_cycle->log, 0,
"%s %P exited on signal %d%s",
process, pid, WTERMSIG(status),
WCOREDUMP(status) ? " (core dumped)" : "");
#else
ngx_log_error(NGX_LOG_ALERT, ngx_cycle->log, 0,
"%s %P exited on signal %d",
process, pid, WTERMSIG(status));
#endif
} else {
ngx_log_error(NGX_LOG_NOTICE, ngx_cycle->log, 0,
"%s %P exited with code %d",
process, pid, WEXITSTATUS(status));
}
if (WEXITSTATUS(status) == 2 && ngx_processes[i].respawn) {
ngx_log_error(NGX_LOG_ALERT, ngx_cycle->log, 0,
"%s %P exited with fatal code %d "
"and cannot be respawned",
process, pid, WEXITSTATUS(status));
ngx_processes[i].respawn = 0;
}
ngx_unlock_mutexes(pid);
}
}
然后信号处理函数结束后，master 进程判断 ngx_reap 标志，发现需要重新拉起进程，调用 ngx_reap_children 函数，在这个函数中会将需要重新拉起的进程通过 ngx_spawn_process 函数拉起来。
static ngx_uint_t
ngx_reap_children(ngx_cycle_t *cycle)
{
ngx_int_t i, n;
ngx_uint_t live;
ngx_channel_t ch;
ngx_core_conf_t *ccf;
ngx_memzero(&ch, sizeof(ngx_channel_t));
ch.command = NGX_CMD_CLOSE_CHANNEL;
ch.fd = -1;
live = 0;
for (i = 0; i < ngx_last_process; i++) {
ngx_log_debug7(NGX_LOG_DEBUG_EVENT, cycle->log, 0,
"child: %i %P e:%d t:%d d:%d r:%d j:%d",
i,
ngx_processes[i].pid,
ngx_processes[i].exiting,
ngx_processes[i].exited,
ngx_processes[i].detached,
ngx_processes[i].respawn,
ngx_processes[i].just_spawn);
if (ngx_processes[i].pid == -1) {
continue;
}
if (ngx_processes[i].exited) {
if (!ngx_processes[i].detached) {
ngx_close_channel(ngx_processes[i].channel, cycle->log);
ngx_processes[i].channel[0] = -1;
ngx_processes[i].channel[1] = -1;
ch.pid = ngx_processes[i].pid;
ch.slot = i;
for (n = 0; n < ngx_last_process; n++) {
if (ngx_processes[n].exited
|| ngx_processes[n].pid == -1
|| ngx_processes[n].channel[0] == -1)
{
continue;
}
ngx_log_debug3(NGX_LOG_DEBUG_CORE, cycle->log, 0,
"pass close channel s:%i pid:%P to:%P",
ch.slot, ch.pid, ngx_processes[n].pid);
/* TODO: NGX_AGAIN */
ngx_write_channel(ngx_processes[n].channel[0],
&ch, sizeof(ngx_channel_t), cycle->log);
}
}
if (ngx_processes[i].respawn
&& !ngx_processes[i].exiting
&& !ngx_terminate
&& !ngx_quit)
{
if (ngx_spawn_process(cycle, ngx_processes[i].proc,
ngx_processes[i].data,
ngx_processes[i].name, i)
== NGX_INVALID_PID)
{
ngx_log_error(NGX_LOG_ALERT, cycle->log, 0,
"could not respawn %s",
ngx_processes[i].name);
continue;
}
ngx_pass_open_channel(cycle);
live = 1;
continue;
}
if (ngx_processes[i].pid == ngx_new_binary) {
ccf = (ngx_core_conf_t *) ngx_get_conf(cycle->conf_ctx,
ngx_core_module);
if (ngx_rename_file((char *) ccf->oldpid.data,
(char *) ccf->pid.data)
== NGX_FILE_ERROR)
{
ngx_log_error(NGX_LOG_ALERT, cycle->log, ngx_errno,
ngx_rename_file_n " %s back to %s failed "
"after the new binary process \"%s\" exited",
ccf->oldpid.data, ccf->pid.data, ngx_argv[0]);
}
ngx_new_binary = 0;
if (ngx_noaccepting) {
ngx_restart = 1;
ngx_noaccepting = 0;
}
}
if (i == ngx_last_process - 1) {
ngx_last_process--;
} else {
ngx_processes[i].pid = -1;
}
} else if (ngx_processes[i].exiting || !ngx_processes[i].detached) {
live = 1;
}
}
return live;
}修改进程名
master-worker 进程模型的主要处理如上所示，然后一般为了方便用户识别 worker 进程和 master 进程，nginx 会对进程名称进行修改，其主要的方案就是修改 argv[0] 的值，不过有一些细节问题需要注意，那就是一般 argv 后面跟着 environ，其中存放着系统的环境变量，如果修改的进程名超过原来的进程，则会修改到 environ 环境变量数组中的内容，故修改方案是重新分配一块内存用于存放 environ 中的内容，这样其它使用到环境变量的地方就不会出现问题了。
首先在 ngx_init_setproctitle 进行相关的初始化，即分配一块新的内存用于存放 environ，注意 environ 是 Linux 系统默认导出的，只需要 extern char **environ 即可引用，初始化完成后，ngx_os_argv_last 指针将会指向可用缓存空间的最后一个字符。
ngx_int_t
ngx_init_setproctitle(ngx_log_t *log)
{
u_char *p;
size_t size;
ngx_uint_t i;
size = 0;
for (i = 0; environ[i]; i++) {
size += ngx_strlen(environ[i]) + 1;
}
p = ngx_alloc(size, log);
if (p == NULL) {
return NGX_ERROR;
}
ngx_os_argv_last = ngx_os_argv[0];
for (i = 0; ngx_os_argv[i]; i++) {
if (ngx_os_argv_last == ngx_os_argv[i]) {
ngx_os_argv_last = ngx_os_argv[i] + ngx_strlen(ngx_os_argv[i]) + 1;
}
}
for (i = 0; environ[i]; i++) {
if (ngx_os_argv_last == environ[i]) {
size = ngx_strlen(environ[i]) + 1;
ngx_os_argv_last = environ[i] + size;
ngx_cpystrn(p, (u_char *) environ[i], size);
environ[i] = (char *) p;
p += size;
}
}
ngx_os_argv_last--;
return NGX_OK;
}实际设置进程名称的地方在 ngx_setproctitle 函数中，将参数中的字符串拷贝到 argv[0] 即可
void
ngx_setproctitle(char *title)
{
u_char *p;
#if (NGX_SOLARIS)
ngx_int_t i;
size_t size;
#endif
ngx_os_argv[1] = NULL;
p = ngx_cpystrn((u_char *) ngx_os_argv[0], (u_char *) "nginx: ",
ngx_os_argv_last - ngx_os_argv[0]);
p = ngx_cpystrn(p, (u_char *) title, ngx_os_argv_last - (char *) p);
#if (NGX_SOLARIS)
size = 0;
for (i = 0; i < ngx_argc; i++) {
size += ngx_strlen(ngx_argv[i]) + 1;
}
if (size > (size_t) ((char *) p - ngx_os_argv[0])) {
/*
* ngx_setproctitle() is too rare operation so we use
* the non-optimized copies
*/
p = ngx_cpystrn(p, (u_char *) " (", ngx_os_argv_last - (char *) p);
for (i = 0; i < ngx_argc; i++) {
p = ngx_cpystrn(p, (u_char *) ngx_argv[i],
ngx_os_argv_last - (char *) p);
p = ngx_cpystrn(p, (u_char *) " ", ngx_os_argv_last - (char *) p);
}
if (*(p - 1) == ' ') {
*(p - 1) = ')';
}
}
#endif
if (ngx_os_argv_last - (char *) p) {
ngx_memset(p, NGX_SETPROCTITLE_PAD, ngx_os_argv_last - (char *) p);
}
ngx_log_debug1(NGX_LOG_DEBUG_CORE, ngx_cycle->log, 0,
"setproctitle: \"%s\"", ngx_os_argv[0]);
}All articles in this blog are licensed under CC BY-NC-SA 4.0 unless stating additionally.
