On Thu, Jul 19, 2012 at 5:31 PM,  <traw...@apache.org> wrote:
> Author: trawick
> Date: Thu Jul 19 21:31:52 2012
> New Revision: 1363557
>
> URL: http://svn.apache.org/viewvc?rev=1363557&view=rev
> Log:
> mpm_event, mpm_worker: Remain active amidst prevalent child process
> resource shortages.
>
> This is a somewhat different direction than r168182 ("transient thread
> creation errors shouldn't take down the whole server").
>
> r168182: If APEXIT_CHILDSICK is received and there aren't any
>          active children at the time, exit.
>
> Now:     If APEXIT_CHILDSICK is received and we never successfully
>          initialized a child, exit.
>
> The issue seen with the r168182 handling is that it is rather easy
> to be left with no active child processes (which causes the server
> to exit completely) during a resource shortage that lasts for some
> measurable period of time, as contrasted with a resource shortage
> that results in only a handful of allocation failures.
>
> Now the server will remain active, though as long as the resource
> shortage exists children may continually fail and the parent will
> try once per second to create a replacement.  The existing logic
> to reduce the spawn rate after such errors will prevent the
> parent from trying to create children more rapidly.

This is change in strategies is worth some (other) thought.  In some
situations it may be trading one basket case for another, but there
are other situations  where a resource exhaustion attack on the system
(or some other self-inflicted fiasco with similar results) is resolved
but with httpd notably absent afterwards until manual intervention.

>
> Modified:
>     httpd/httpd/trunk/CHANGES
>     httpd/httpd/trunk/docs/log-message-tags/next-number
>     httpd/httpd/trunk/server/mpm/event/event.c
>     httpd/httpd/trunk/server/mpm/worker/worker.c
>
> Modified: httpd/httpd/trunk/CHANGES
> URL: 
> http://svn.apache.org/viewvc/httpd/httpd/trunk/CHANGES?rev=1363557&r1=1363556&r2=1363557&view=diff
> ==============================================================================
> --- httpd/httpd/trunk/CHANGES [utf-8] (original)
> +++ httpd/httpd/trunk/CHANGES [utf-8] Thu Jul 19 21:31:52 2012
> @@ -1,6 +1,9 @@
>                                                           -*- coding: utf-8 
> -*-
>  Changes with Apache 2.5.0
>
> +  *) mpm_event, mpm_worker: Remain active amidst prevalent child process
> +     resource shortages.  [Jeff Trawick]
> +
>    *) mpm_event, mpm_worker: Fix cases where the spawn rate wasn't reduced
>       after child process resource shortages.  [Jeff Trawick]
>
>
> Modified: httpd/httpd/trunk/docs/log-message-tags/next-number
> URL: 
> http://svn.apache.org/viewvc/httpd/httpd/trunk/docs/log-message-tags/next-number?rev=1363557&r1=1363556&r2=1363557&view=diff
> ==============================================================================
> --- httpd/httpd/trunk/docs/log-message-tags/next-number (original)
> +++ httpd/httpd/trunk/docs/log-message-tags/next-number Thu Jul 19 21:31:52 
> 2012
> @@ -1 +1 @@
> -2324
> +2326
>
> Modified: httpd/httpd/trunk/server/mpm/event/event.c
> URL: 
> http://svn.apache.org/viewvc/httpd/httpd/trunk/server/mpm/event/event.c?rev=1363557&r1=1363556&r2=1363557&view=diff
> ==============================================================================
> --- httpd/httpd/trunk/server/mpm/event/event.c (original)
> +++ httpd/httpd/trunk/server/mpm/event/event.c Thu Jul 19 21:31:52 2012
> @@ -183,6 +183,7 @@ static apr_uint32_t lingering_count = 0;
>  static apr_uint32_t suspended_count = 0;    /* Number of suspended 
> connections */
>  static apr_uint32_t clogged_count = 0;      /* Number of threads processing 
> ssl conns */
>  static int resource_shortage = 0;
> +static int had_healthy_child = 0;
>  static fd_queue_t *worker_queue;
>  static fd_queue_info_t *worker_queue_info;
>  static int mpm_state = AP_MPMQ_STARTING;
> @@ -2403,6 +2404,7 @@ static void perform_idle_server_maintena
>          int any_dying_threads = 0;
>          int any_dead_threads = 0;
>          int all_dead_threads = 1;
> +        int child_threads_active = 0;
>
>          if (i >= retained->max_daemons_limit
>              && totally_free_length == retained->idle_spawn_rate)
> @@ -2438,6 +2440,7 @@ static void perform_idle_server_maintena
>                  }
>                  if (status >= SERVER_READY && status < SERVER_GRACEFUL) {
>                      ++active_thread_count;
> +                    ++child_threads_active;
>                  }
>              }
>          }
> @@ -2464,6 +2467,9 @@ static void perform_idle_server_maintena
>              }
>              ++free_length;
>          }
> +        else if (child_threads_active == threads_per_child) {
> +            had_healthy_child = 1;
> +        }
>          /* XXX if (!ps->quiescing)     is probably more reliable  GLA */
>          if (!any_dying_threads) {
>              last_non_dead = i;
> @@ -2472,21 +2478,23 @@ static void perform_idle_server_maintena
>      }
>
>      if (retained->sick_child_detected) {
> -        if (active_thread_count > 0) {
> -            /* some child processes appear to be working.  don't kill the
> -             * whole server.
> +        if (had_healthy_child) {
> +            /* Assume this is a transient error, even though it may not be.  
> Leave
> +             * the server up in case it is able to serve some requests or the
> +             * problem will be resolved.
>               */
>              retained->sick_child_detected = 0;
>          }
>          else {
> -            /* looks like a basket case.  give up.
> +            /* looks like a basket case, as no child ever fully initialized; 
> give up.
>               */
>              shutdown_pending = 1;
>              child_fatal = 1;
>              ap_log_error(APLOG_MARK, APLOG_ALERT, 0,
> -                         ap_server_conf, APLOGNO(00483)
> -                         "No active workers found..."
> -                         " Apache is exiting!");
> +                         ap_server_conf, APLOGNO(02324)
> +                         "A resource shortage or other unrecoverable failure 
> "
> +                         "was encountered before any child process 
> initialized "
> +                         "successfully... httpd is exiting!");
>              /* the child already logged the failure details */
>              return;
>          }
>
> Modified: httpd/httpd/trunk/server/mpm/worker/worker.c
> URL: 
> http://svn.apache.org/viewvc/httpd/httpd/trunk/server/mpm/worker/worker.c?rev=1363557&r1=1363556&r2=1363557&view=diff
> ==============================================================================
> --- httpd/httpd/trunk/server/mpm/worker/worker.c (original)
> +++ httpd/httpd/trunk/server/mpm/worker/worker.c Thu Jul 19 21:31:52 2012
> @@ -129,6 +129,7 @@ static int listener_may_exit = 0;
>  static int requests_this_child;
>  static int num_listensocks = 0;
>  static int resource_shortage = 0;
> +static int had_healthy_child = 0;
>  static fd_queue_t *worker_queue;
>  static fd_queue_info_t *worker_queue_info;
>  static int mpm_state = AP_MPMQ_STARTING;
> @@ -1473,6 +1474,7 @@ static void perform_idle_server_maintena
>          int any_dying_threads = 0;
>          int any_dead_threads = 0;
>          int all_dead_threads = 1;
> +        int child_threads_active = 0;
>
>          if (i >= retained->max_daemons_limit && totally_free_length == 
> retained->idle_spawn_rate)
>              /* short cut if all active processes have been examined and
> @@ -1507,6 +1509,7 @@ static void perform_idle_server_maintena
>                  }
>                  if (status >= SERVER_READY && status < SERVER_GRACEFUL) {
>                      ++active_thread_count;
> +                    ++child_threads_active;
>                  }
>              }
>          }
> @@ -1532,6 +1535,9 @@ static void perform_idle_server_maintena
>              }
>              ++free_length;
>          }
> +        else if (child_threads_active == threads_per_child) {
> +            had_healthy_child = 1;
> +        }
>          /* XXX if (!ps->quiescing)     is probably more reliable  GLA */
>          if (!any_dying_threads) {
>              last_non_dead = i;
> @@ -1540,21 +1546,23 @@ static void perform_idle_server_maintena
>      }
>
>      if (retained->sick_child_detected) {
> -        if (active_thread_count > 0) {
> -            /* some child processes appear to be working.  don't kill the
> -             * whole server.
> +        if (had_healthy_child) {
> +            /* Assume this is a transient error, even though it may not be.  
> Leave
> +             * the server up in case it is able to serve some requests or the
> +             * problem will be resolved.
>               */
>              retained->sick_child_detected = 0;
>          }
>          else {
> -            /* looks like a basket case.  give up.
> +            /* looks like a basket case, as no child ever fully initialized; 
> give up.
>               */
>              shutdown_pending = 1;
>              child_fatal = 1;
>              ap_log_error(APLOG_MARK, APLOG_ALERT, 0,
> -                         ap_server_conf, APLOGNO(00285)
> -                         "No active workers found..."
> -                         " Apache is exiting!");
> +                         ap_server_conf, APLOGNO(02325)
> +                         "A resource shortage or other unrecoverable failure 
> "
> +                         "was encountered before any child process 
> initialized "
> +                         "successfully... httpd is exiting!");
>              /* the child already logged the failure details */
>              return;
>          }
>
>



-- 
Born in Roswell... married an alien...
http://emptyhammock.com/

Reply via email to