glusterd: fix for starting brick on new port (#2090)

The Errno set by the runner code was not correct when the bind() fails
to assign an already occupied port in the __socket_server_bind().

Fix:
Updated the code to return the correct errno from the
__socket_server_bind() if the bind() fails due to EADDRINUSE error. And,
use the returned errno from runner_run() to retry allocating a new port
to the brick process.

Fixes: #1101

Change-Id: If124337f41344a04f050754e402490529ef4ecdc
Signed-off-by: nik-redhat nladha@redhat.com

1 files changed, 20 insertions, 12 deletions

diff --git a/rpc/rpc-transport/socket/src/socket.c b/rpc/rpc-transport/socket/src/socket.c
index ed8b473be2..b04847c223 100644
--- a/rpc/rpc-transport/socket/src/socket.c
+++ b/rpc/rpc-transport/socket/src/socket.c
@@ -948,6 +948,13 @@ __socket_server_bind(rpc_transport_t *this)
                        this->myinfo.identifier, strerror(errno));
                 if (errno == EADDRINUSE) {
                     gf_log(this->name, GF_LOG_ERROR, "Port is already in use");
+                    ret = -EADDRINUSE;
+
+                    /* TODO: Remove this delay. It was added only to address
+                     * failures in the regression test suite, In a real
+                     * situation, if a port is in use when the process starts,
+                     * it most likely will remain in use during 10 seconds as
+                     * well.*/
                     sleep(1);
                     retries--;
                 } else {
@@ -957,19 +964,9 @@ __socket_server_bind(rpc_transport_t *this)
                 break;
             }
         }
-    } else {
-        ret = bind(priv->sock, (struct sockaddr *)&this->myinfo.sockaddr,
-                   this->myinfo.sockaddr_len);
+        if (ret < 0)
+            goto out;
 
-        if (ret != 0) {
-            gf_log(this->name, GF_LOG_ERROR, "binding to %s failed: %s",
-                   this->myinfo.identifier, strerror(errno));
-            if (errno == EADDRINUSE) {
-                gf_log(this->name, GF_LOG_ERROR, "Port is already in use");
-            }
-        }
-    }
-    if (AF_UNIX != SA(&this->myinfo.sockaddr)->sa_family) {
         if (getsockname(priv->sock, SA(&this->myinfo.sockaddr),
                         &this->myinfo.sockaddr_len) != 0) {
             gf_log(this->name, GF_LOG_WARNING,
@@ -985,6 +982,17 @@ __socket_server_bind(rpc_transport_t *this)
                    "process started listening on port (%d)",
                    cmd_args->brick_port);
         }
+    } else {
+        ret = bind(priv->sock, (struct sockaddr *)&this->myinfo.sockaddr,
+                   this->myinfo.sockaddr_len);
+
+        if (ret != 0) {
+            gf_log(this->name, GF_LOG_ERROR, "binding to %s failed: %s",
+                   this->myinfo.identifier, strerror(errno));
+            if (errno == EADDRINUSE) {
+                gf_log(this->name, GF_LOG_ERROR, "Port is already in use");
+            }
+        }
     }
 
 out: