From: Eric Covener <covener@apache.org>
Date: Tue, 17 Mar 2015 20:12:19 +0000 (+0000)
Subject: Retry ENOENT like ECONNREFUSED, but only near a server restart.
X-Git-Tag: 2.5.0-alpha~3372
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=c27569bfd3bc1b4562588fb8161e22788cea24a7;p=apache

Retry ENOENT like ECONNREFUSED, but only near a server restart.

PR57685

Submitted By: Edward Lu
Committed By: covener




git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/trunk@1667385 13f79535-47bb-0310-9956-ffa450edef68
---

diff --git a/CHANGES b/CHANGES
index b19045cead..93577cd773 100644
--- a/CHANGES
+++ b/CHANGES
@@ -6,6 +6,12 @@ Changes with Apache 2.5.0
      to a local URL-path with the INCLUDES filter active, introduced
      in 2.4.11. PR 57531. [Yann Ylavic]
 
+  *) mod_cgid: Within the first minute of a server start or restart, 
+     allow mod_cgid to retry connecting to its daemon process. Previously,
+     'No such file or directory: unable to connect to cgi daemon...' could
+     be logged without an actual retry. PR57685. 
+     [Edward Lu <Chaosed0 gmail.com>]
+     
   *) core: Cleanup the request soon/even if some output filter fails to
      handle the EOR bucket.  [Yann Ylavic]
 
diff --git a/modules/generators/mod_cgid.c b/modules/generators/mod_cgid.c
index 50cc883067..2a0e94db5a 100644
--- a/modules/generators/mod_cgid.c
+++ b/modules/generators/mod_cgid.c
@@ -170,6 +170,10 @@ static int is_scriptaliased(request_rec *r)
 #define DEFAULT_CONNECT_ATTEMPTS  15
 #endif
 
+#ifndef DEFAULT_CONNECT_STARTUP_DELAY
+#define DEFAULT_CONNECT_STARTUP_DELAY 60
+#endif
+
 typedef struct {
     const char *logname;
     long logbytes;
@@ -1203,18 +1207,26 @@ static int connect_to_daemon(int *sdptr, request_rec *r,
 {
     int sd;
     int connect_tries;
+    int connect_errno;
     apr_interval_time_t sliding_timer;
 
     connect_tries = 0;
     sliding_timer = 100000; /* 100 milliseconds */
     while (1) {
+        connect_errno = 0;
         ++connect_tries;
         if ((sd = socket(AF_UNIX, SOCK_STREAM, 0)) < 0) {
             return log_scripterror(r, conf, HTTP_INTERNAL_SERVER_ERROR, errno,
                                    APLOGNO(01255) "unable to create socket to cgi daemon");
         }
         if (connect(sd, (struct sockaddr *)server_addr, server_addr_len) < 0) {
-            if (errno == ECONNREFUSED && connect_tries < DEFAULT_CONNECT_ATTEMPTS) {
+            /* Save errno for later */
+            connect_errno = errno;
+            /* ECONNREFUSED means the listen queue is full; ENOENT means that
+             * the cgid server either hasn't started up yet, or we're pointing
+             * at the wrong socket file */
+            if ((errno == ECONNREFUSED || errno == ENOENT) && 
+                 connect_tries < DEFAULT_CONNECT_ATTEMPTS) {
                 ap_log_rerror(APLOG_MARK, APLOG_DEBUG, errno, r, APLOGNO(01256)
                               "connect #%d to cgi daemon failed, sleeping before retry",
                               connect_tries);
@@ -1235,9 +1247,20 @@ static int connect_to_daemon(int *sdptr, request_rec *r,
                                       close_unix_socket, apr_pool_cleanup_null);
             break; /* we got connected! */
         }
+
+        /* If we didn't find the socket but the server was not recently restarted,
+         * chances are there's something wrong with the cgid daemon
+         */
+        if (connect_errno == ENOENT &&
+            apr_time_sec(apr_time_now() - ap_scoreboard_image->global->restart_time) > 
+                DEFAULT_CONNECT_STARTUP_DELAY) {
+            return log_scripterror(r, conf, HTTP_SERVICE_UNAVAILABLE, connect_errno, 
+                                   apr_pstrcat(r->pool, APLOGNO(02831) "ScriptSock ", sockname, " does not exist", NULL));
+        }
+
         /* gotta try again, but make sure the cgid daemon is still around */
-        if (kill(daemon_pid, 0) != 0) {
-            return log_scripterror(r, conf, HTTP_SERVICE_UNAVAILABLE, errno, APLOGNO(01258)
+        if (connect_errno != ENOENT && kill(daemon_pid, 0) != 0) {
+            return log_scripterror(r, conf, HTTP_SERVICE_UNAVAILABLE, connect_errno, APLOGNO(01258)
                                    "cgid daemon is gone; is Apache terminating?");
         }
     }