Example #1
0
void ap_relieve_child_processes(void)
{
    int i;
    extra_process_t *cur_extra;
    int max_daemons;

    ap_mpm_query(AP_MPMQ_MAX_DAEMON_USED, &max_daemons);

    /* now see who is done */
    for (i = 0; i < max_daemons; ++i) {
        pid_t pid = MPM_CHILD_PID(i);

        if (pid == 0) {
            continue; /* not every scoreboard entry is in use */
        }

        if (reclaim_one_pid(pid, DO_NOTHING)) {
            MPM_NOTE_CHILD_KILLED(i);
        }
    }

    cur_extra = extras;
    while (cur_extra) {
        extra_process_t *next = cur_extra->next;

        if (reclaim_one_pid(cur_extra->pid, DO_NOTHING)) {
            AP_DEBUG_ASSERT(1 == ap_unregister_extra_mpm_process(cur_extra->pid));
        }
        cur_extra = next;
    }
}
Example #2
0
int ap_mpm_run(apr_pool_t *_pconf, apr_pool_t *plog, server_rec *s)
{
    int index;
    int remaining_children_to_start;
    apr_status_t rv;

    ap_log_pid(pconf, ap_pid_fname);

    first_server_limit = server_limit;
    if (changed_limit_at_restart) {
        ap_log_error(APLOG_MARK, APLOG_WARNING, 0, s,
                     "WARNING: Attempt to change ServerLimit "
                     "ignored during restart");
        changed_limit_at_restart = 0;
    }

    /* Initialize cross-process accept lock */
    ap_lock_fname = apr_psprintf(_pconf, "%s.%" APR_PID_T_FMT,
                                 ap_server_root_relative(_pconf, ap_lock_fname),
                                 ap_my_pid);

    rv = apr_proc_mutex_create(&accept_mutex, ap_lock_fname,
                               ap_accept_lock_mech, _pconf);
    if (rv != APR_SUCCESS) {
        ap_log_error(APLOG_MARK, APLOG_EMERG, rv, s,
                     "Couldn't create accept lock (%s) (%d)",
                     ap_lock_fname, ap_accept_lock_mech);
        mpm_state = AP_MPMQ_STOPPING;
        return 1;
    }

#if APR_USE_SYSVSEM_SERIALIZE
    if (ap_accept_lock_mech == APR_LOCK_DEFAULT ||
        ap_accept_lock_mech == APR_LOCK_SYSVSEM) {
#else
    if (ap_accept_lock_mech == APR_LOCK_SYSVSEM) {
#endif
        rv = unixd_set_proc_mutex_perms(accept_mutex);
        if (rv != APR_SUCCESS) {
            ap_log_error(APLOG_MARK, APLOG_EMERG, rv, s,
                         "Couldn't set permissions on cross-process lock; "
                         "check User and Group directives");
            mpm_state = AP_MPMQ_STOPPING;
            return 1;
        }
    }

    if (!is_graceful) {
        if (ap_run_pre_mpm(s->process->pool, SB_SHARED) != OK) {
            mpm_state = AP_MPMQ_STOPPING;
            return 1;
        }
        /* fix the generation number in the global score; we just got a new,
         * cleared scoreboard
         */
        ap_scoreboard_image->global->running_generation = ap_my_generation;
    }

    set_signals();

    if (one_process) {
        AP_MONCONTROL(1);
        make_child(ap_server_conf, 0);
    }
    else {
    if (ap_daemons_max_free < ap_daemons_min_free + 1)  /* Don't thrash... */
        ap_daemons_max_free = ap_daemons_min_free + 1;

    /* If we're doing a graceful_restart then we're going to see a lot
     * of children exiting immediately when we get into the main loop
     * below (because we just sent them AP_SIG_GRACEFUL).  This happens pretty
     * rapidly... and for each one that exits we'll start a new one until
     * we reach at least daemons_min_free.  But we may be permitted to
     * start more than that, so we'll just keep track of how many we're
     * supposed to start up without the 1 second penalty between each fork.
     */
    remaining_children_to_start = ap_daemons_to_start;
    if (remaining_children_to_start > ap_daemons_limit) {
        remaining_children_to_start = ap_daemons_limit;
    }
    if (!is_graceful) {
        startup_children(remaining_children_to_start);
        remaining_children_to_start = 0;
    }
    else {
        /* give the system some time to recover before kicking into
         * exponential mode
         */
        hold_off_on_exponential_spawning = 10;
    }

    ap_log_error(APLOG_MARK, APLOG_NOTICE, 0, ap_server_conf,
                "%s configured -- resuming normal operations",
                ap_get_server_description());
    ap_log_error(APLOG_MARK, APLOG_INFO, 0, ap_server_conf,
                "Server built: %s", ap_get_server_built());
#ifdef AP_MPM_WANT_SET_ACCEPT_LOCK_MECH
    ap_log_error(APLOG_MARK, APLOG_DEBUG, 0, ap_server_conf,
                "AcceptMutex: %s (default: %s)",
                apr_proc_mutex_name(accept_mutex),
                apr_proc_mutex_defname());
#endif
    restart_pending = shutdown_pending = 0;

    mpm_state = AP_MPMQ_RUNNING;

    while (!restart_pending && !shutdown_pending) {
        int child_slot;
        apr_exit_why_e exitwhy;
        int status, processed_status;
        /* this is a memory leak, but I'll fix it later. */
        apr_proc_t pid;

        ap_wait_or_timeout(&exitwhy, &status, &pid, pconf);

        /* XXX: if it takes longer than 1 second for all our children
         * to start up and get into IDLE state then we may spawn an
         * extra child
         */
        if (pid.pid != -1) {
            processed_status = ap_process_child_status(&pid, exitwhy, status);
            if (processed_status == APEXIT_CHILDFATAL) {
                mpm_state = AP_MPMQ_STOPPING;
                return 1;
            }

            /* non-fatal death... note that it's gone in the scoreboard. */
            child_slot = find_child_by_pid(&pid);
            if (child_slot >= 0) {
                (void) ap_update_child_status_from_indexes(child_slot, 0, SERVER_DEAD,
                                                           (request_rec *) NULL);
                if (processed_status == APEXIT_CHILDSICK) {
                    /* child detected a resource shortage (E[NM]FILE, ENOBUFS, etc)
                     * cut the fork rate to the minimum
                     */
                    idle_spawn_rate = 1;
                }
                else if (remaining_children_to_start
                    && child_slot < ap_daemons_limit) {
                    /* we're still doing a 1-for-1 replacement of dead
                     * children with new children
                     */
                    make_child(ap_server_conf, child_slot);
                    --remaining_children_to_start;
                }
#if APR_HAS_OTHER_CHILD
            }
            else if (apr_proc_other_child_alert(&pid, APR_OC_REASON_DEATH, status) == APR_SUCCESS) {
                /* handled */
#endif
            }
            else if (is_graceful) {
                /* Great, we've probably just lost a slot in the
                 * scoreboard.  Somehow we don't know about this
                 * child.
                 */
                ap_log_error(APLOG_MARK, APLOG_WARNING,
                            0, ap_server_conf,
                            "long lost child came home! (pid %ld)", (long)pid.pid);
            }
            /* Don't perform idle maintenance when a child dies,
             * only do it when there's a timeout.  Remember only a
             * finite number of children can die, and it's pretty
             * pathological for a lot to die suddenly.
             */
            continue;
        }
        else if (remaining_children_to_start) {
            /* we hit a 1 second timeout in which none of the previous
             * generation of children needed to be reaped... so assume
             * they're all done, and pick up the slack if any is left.
             */
            startup_children(remaining_children_to_start);
            remaining_children_to_start = 0;
            /* In any event we really shouldn't do the code below because
             * few of the servers we just started are in the IDLE state
             * yet, so we'd mistakenly create an extra server.
             */
            continue;
        }

        perform_idle_server_maintenance(pconf);
#ifdef TPF
        shutdown_pending = os_check_server(tpf_server_name);
        ap_check_signals();
        sleep(1);
#endif /*TPF */
    }
    } /* one_process */

    mpm_state = AP_MPMQ_STOPPING;

    if (shutdown_pending && !is_graceful) {
        /* Time to shut down:
         * Kill child processes, tell them to call child_exit, etc...
         */
        if (unixd_killpg(getpgrp(), SIGTERM) < 0) {
            ap_log_error(APLOG_MARK, APLOG_WARNING, errno, ap_server_conf, "killpg SIGTERM");
        }
        ap_reclaim_child_processes(1);          /* Start with SIGTERM */

        /* cleanup pid file on normal shutdown */
        {
            const char *pidfile = NULL;
            pidfile = ap_server_root_relative (pconf, ap_pid_fname);
            if ( pidfile != NULL && unlink(pidfile) == 0)
                ap_log_error(APLOG_MARK, APLOG_INFO,
                                0, ap_server_conf,
                                "removed PID file %s (pid=%ld)",
                                pidfile, (long)getpid());
        }

        ap_log_error(APLOG_MARK, APLOG_NOTICE, 0, ap_server_conf,
                    "caught SIGTERM, shutting down");

        return 1;
    } else if (shutdown_pending) {
        /* Time to perform a graceful shut down:
         * Reap the inactive children, and ask the active ones
         * to close their listeners, then wait until they are
         * all done to exit.
         */
        int active_children;
        apr_time_t cutoff = 0;

        /* Stop listening */
        ap_close_listeners();

        /* kill off the idle ones */
        ap_mpm_pod_killpg(pod, ap_max_daemons_limit);

        /* Send SIGUSR1 to the active children */
        active_children = 0;
        for (index = 0; index < ap_daemons_limit; ++index) {
            if (ap_scoreboard_image->servers[index][0].status != SERVER_DEAD) {
                /* Ask each child to close its listeners. */
                ap_mpm_safe_kill(MPM_CHILD_PID(index), AP_SIG_GRACEFUL);
                active_children++;
            }
        }

        /* Allow each child which actually finished to exit */
        ap_relieve_child_processes();

        /* cleanup pid file */
        {
            const char *pidfile = NULL;
            pidfile = ap_server_root_relative (pconf, ap_pid_fname);
            if ( pidfile != NULL && unlink(pidfile) == 0)
                ap_log_error(APLOG_MARK, APLOG_INFO,
                                0, ap_server_conf,
                                "removed PID file %s (pid=%ld)",
                                pidfile, (long)getpid());
        }

        ap_log_error(APLOG_MARK, APLOG_NOTICE, 0, ap_server_conf,
           "caught " AP_SIG_GRACEFUL_STOP_STRING ", shutting down gracefully");

        if (ap_graceful_shutdown_timeout) {
            cutoff = apr_time_now() +
                     apr_time_from_sec(ap_graceful_shutdown_timeout);
        }

        /* Don't really exit until each child has finished */
        shutdown_pending = 0;
        do {
            /* Pause for a second */
            sleep(1);

            /* Relieve any children which have now exited */
            ap_relieve_child_processes();

            active_children = 0;
            for (index = 0; index < ap_daemons_limit; ++index) {
                if (ap_mpm_safe_kill(MPM_CHILD_PID(index), 0) == APR_SUCCESS) {
                    active_children = 1;
                    /* Having just one child is enough to stay around */
                    break;
                }
            }
        } while (!shutdown_pending && active_children &&
                 (!ap_graceful_shutdown_timeout || apr_time_now() < cutoff));

        /* We might be here because we received SIGTERM, either
         * way, try and make sure that all of our processes are
         * really dead.
         */
        unixd_killpg(getpgrp(), SIGTERM);

        return 1;
    }

    /* we've been told to restart */
    apr_signal(SIGHUP, SIG_IGN);
    apr_signal(AP_SIG_GRACEFUL, SIG_IGN);
    if (one_process) {
        /* not worth thinking about */
        return 1;
    }

    /* advance to the next generation */
    /* XXX: we really need to make sure this new generation number isn't in
     * use by any of the children.
     */
    ++ap_my_generation;
    ap_scoreboard_image->global->running_generation = ap_my_generation;

    if (is_graceful) {
        ap_log_error(APLOG_MARK, APLOG_NOTICE, 0, ap_server_conf,
                    "Graceful restart requested, doing restart");

        /* kill off the idle ones */
        ap_mpm_pod_killpg(pod, ap_max_daemons_limit);

        /* This is mostly for debugging... so that we know what is still
         * gracefully dealing with existing request.  This will break
         * in a very nasty way if we ever have the scoreboard totally
         * file-based (no shared memory)
         */
        for (index = 0; index < ap_daemons_limit; ++index) {
            if (ap_scoreboard_image->servers[index][0].status != SERVER_DEAD) {
                ap_scoreboard_image->servers[index][0].status = SERVER_GRACEFUL;
                /* Ask each child to close its listeners.
                 *
                 * NOTE: we use the scoreboard, because if we send SIGUSR1
                 * to every process in the group, this may include CGI's,
                 * piped loggers, etc. They almost certainly won't handle
                 * it gracefully.
                 */
                ap_mpm_safe_kill(ap_scoreboard_image->parent[index].pid, AP_SIG_GRACEFUL);
            }
        }
    }
    else {
        /* Kill 'em off */
        if (unixd_killpg(getpgrp(), SIGHUP) < 0) {
            ap_log_error(APLOG_MARK, APLOG_WARNING, errno, ap_server_conf, "killpg SIGHUP");
        }
        ap_reclaim_child_processes(0);          /* Not when just starting up */
        ap_log_error(APLOG_MARK, APLOG_NOTICE, 0, ap_server_conf,
                    "SIGHUP received.  Attempting to restart");
    }

    return 0;
}

/* This really should be a post_config hook, but the error log is already
 * redirected by that point, so we need to do this in the open_logs phase.
 */
static int prefork_open_logs(apr_pool_t *p, apr_pool_t *plog, apr_pool_t *ptemp, server_rec *s)
{
    apr_status_t rv;

    pconf = p;
    ap_server_conf = s;

    if ((num_listensocks = ap_setup_listeners(ap_server_conf)) < 1) {
        ap_log_error(APLOG_MARK, APLOG_ALERT|APLOG_STARTUP, 0,
                     NULL, "no listening sockets available, shutting down");
        return DONE;
    }

    if ((rv = ap_mpm_pod_open(pconf, &pod))) {
        ap_log_error(APLOG_MARK, APLOG_CRIT|APLOG_STARTUP, rv, NULL,
                "Could not open pipe-of-death.");
        return DONE;
    }
    return OK;
}
Example #3
0
void ap_reclaim_child_processes(int terminate)
{
    apr_time_t waittime = 1024 * 16;
    int i;
    extra_process_t *cur_extra;
    int not_dead_yet;
    int max_daemons;
    apr_time_t starttime = apr_time_now();
    /* this table of actions and elapsed times tells what action is taken
     * at which elapsed time from starting the reclaim
     */
    struct {
        action_t action;
        apr_time_t action_time;
    } action_table[] = {
        {DO_NOTHING, 0}, /* dummy entry for iterations where we reap
                          * children but take no action against
                          * stragglers
                          */
        {SEND_SIGTERM, apr_time_from_sec(3)},
        {SEND_SIGTERM, apr_time_from_sec(5)},
        {SEND_SIGTERM, apr_time_from_sec(7)},
        {SEND_SIGKILL, apr_time_from_sec(9)},
        {GIVEUP,       apr_time_from_sec(10)}
    };
    int cur_action;      /* index of action we decided to take this
                          * iteration
                          */
    int next_action = 1; /* index of first real action */

    ap_mpm_query(AP_MPMQ_MAX_DAEMON_USED, &max_daemons);

    do {
        apr_sleep(waittime);
        /* don't let waittime get longer than 1 second; otherwise, we don't
         * react quickly to the last child exiting, and taking action can
         * be delayed
         */
        waittime = waittime * 4;
        if (waittime > apr_time_from_sec(1)) {
            waittime = apr_time_from_sec(1);
        }

        /* see what action to take, if any */
        if (action_table[next_action].action_time <= apr_time_now() - starttime) {
            cur_action = next_action;
            ++next_action;
        }
        else {
            cur_action = 0; /* nothing to do */
        }

        /* now see who is done */
        not_dead_yet = 0;
        for (i = 0; i < max_daemons; ++i) {
            pid_t pid = MPM_CHILD_PID(i);

            if (pid == 0) {
                continue; /* not every scoreboard entry is in use */
            }

            if (reclaim_one_pid(pid, action_table[cur_action].action)) {
                MPM_NOTE_CHILD_KILLED(i);
            }
            else {
                ++not_dead_yet;
            }
        }

        cur_extra = extras;
        while (cur_extra) {
            extra_process_t *next = cur_extra->next;

            if (reclaim_one_pid(cur_extra->pid, action_table[cur_action].action)) {
                AP_DEBUG_ASSERT(1 == ap_unregister_extra_mpm_process(cur_extra->pid));
            }
            else {
                ++not_dead_yet;
            }
            cur_extra = next;
        }
#if APR_HAS_OTHER_CHILD
        apr_proc_other_child_refresh_all(APR_OC_REASON_RESTART);
#endif

    } while (not_dead_yet > 0 &&
             action_table[cur_action].action != GIVEUP);
}