rpc: Reduce frame-timeout for glusterd connections

Reduce frame-timeout for glusterd connections from 30mins to 10 mins. 30mins is
too long when compared to cli timeout of 2mins. Changing to 10mins reduces the
disparity between cli and glusterd.

Also, fix glusterfs_submit_reply() so that a reply is sent even if serialize
failed.

BUG: 843003
Change-Id: Ie8d5ec16fbbb54318a5935a47065e66fd3338b87
Signed-off-by: Kaushal M <kaushal@redhat.com>
Reviewed-on: http://review.gluster.com/3812
Tested-by: Gluster Build System <jenkins@build.gluster.com>
Reviewed-by: Vijay Bellur <vbellur@redhat.com>

6 files changed, 58 insertions, 13 deletions

diff --git a/glusterfsd/src/glusterfsd-mgmt.c b/glusterfsd/src/glusterfsd-mgmt.c
index fb1619bc512..7f92299ff04 100644
--- a/glusterfsd/src/glusterfsd-mgmt.c
+++ b/glusterfsd/src/glusterfsd-mgmt.c
@@ -142,11 +142,10 @@ glusterfs_submit_reply (rpcsvc_request_t *req, void *arg,
         iob = glusterfs_serialize_reply (req, arg, &rsp, xdrproc);
         if (!iob) {
                 gf_log_callingfn (THIS->name, GF_LOG_ERROR, "Failed to serialize reply");
-                goto out;
+        } else {
+                iobref_add (iobref, iob);
         }
 
-        iobref_add (iobref, iob);
-
         ret = rpcsvc_submit_generic (req, &rsp, 1, payload, payloadcount,
                                      iobref);
 
diff --git a/rpc/rpc-lib/src/rpc-clnt.c b/rpc/rpc-lib/src/rpc-clnt.c
index 3cfd92a17a2..f011de5de0c 100644
--- a/rpc/rpc-lib/src/rpc-clnt.c
+++ b/rpc/rpc-lib/src/rpc-clnt.c
@@ -1679,7 +1679,8 @@ rpc_clnt_reconfig (struct rpc_clnt *rpc, struct rpc_clnt_config *config)
 }
 
 int
-rpc_clnt_transport_unix_options_build (dict_t **options, char *filepath)
+rpc_clnt_transport_unix_options_build (dict_t **options, char *filepath,
+                                       int frame_timeout)
 {
         dict_t                  *dict = NULL;
         char                    *fpath = NULL;
@@ -1718,6 +1719,12 @@ rpc_clnt_transport_unix_options_build (dict_t **options, char *filepath)
         if (ret)
                 goto out;
 
+        if (frame_timeout > 0) {
+                ret = dict_set_int32 (dict, "frame-timeout", frame_timeout);
+                if (ret)
+                        goto out;
+        }
+
         *options = dict;
 out:
         if (ret) {
diff --git a/rpc/rpc-lib/src/rpc-clnt.h b/rpc/rpc-lib/src/rpc-clnt.h
index e7335e38839..ca3c1d027a2 100644
--- a/rpc/rpc-lib/src/rpc-clnt.h
+++ b/rpc/rpc-lib/src/rpc-clnt.h
@@ -235,7 +235,8 @@ int rpcclnt_cbk_program_register (struct rpc_clnt *svc,
                                   rpcclnt_cb_program_t *program);
 
 int
-rpc_clnt_transport_unix_options_build (dict_t **options, char *filepath);
+rpc_clnt_transport_unix_options_build (dict_t **options, char *filepath,
+                                       int frame_timeout);
 
 void
 rpc_clnt_disable (struct rpc_clnt *rpc);
diff --git a/xlators/mgmt/glusterd/src/glusterd-handler.c b/xlators/mgmt/glusterd/src/glusterd-handler.c
index 2dcfd28148f..2e50b1c0e7d 100644
--- a/xlators/mgmt/glusterd/src/glusterd-handler.c
+++ b/xlators/mgmt/glusterd/src/glusterd-handler.c
@@ -2106,8 +2106,8 @@ glusterd_transport_keepalive_options_get (int *interval, int *time)
 }
 
 int
-glusterd_transport_inet_keepalive_options_build (dict_t **options,
-                                                 const char *hostname, int port)
+glusterd_transport_inet_options_build (dict_t **options, const char *hostname,
+                                       int port)
 {
         dict_t  *dict = NULL;
         int32_t interval = -1;
@@ -2119,10 +2119,25 @@ glusterd_transport_inet_keepalive_options_build (dict_t **options,
 
         if (!port)
                 port = GLUSTERD_DEFAULT_PORT;
+
+        /* Build default transport options */
         ret = rpc_transport_inet_options_build (&dict, hostname, port);
         if (ret)
                 goto out;
 
+        /* Set frame-timeout to 10mins. Default timeout of 30 mins is too long
+         * when compared to 2 mins for cli timeout. This ensures users don't
+         * wait too long after cli timesout before being able to resume normal
+         * operations
+         */
+        ret = dict_set_int32 (dict, "frame-timeout", 600);
+        if (ret) {
+                gf_log ("glusterd", GF_LOG_ERROR,
+                        "Failed to set frame-timeout");
+                goto out;
+        }
+
+        /* Set keepalive options */
         glusterd_transport_keepalive_options_get (&interval, &time);
 
         if ((interval > 0) || (time > 0))
@@ -2168,8 +2183,7 @@ glusterd_friend_add (const char *hoststr, int port,
 
         peerctx->peerinfo = *friend;
 
-        ret = glusterd_transport_inet_keepalive_options_build (&options,
-                                                         hoststr, port);
+        ret = glusterd_transport_inet_options_build (&options, hoststr, port);
         if (ret)
                 goto out;
 
diff --git a/xlators/mgmt/glusterd/src/glusterd-rebalance.c b/xlators/mgmt/glusterd/src/glusterd-rebalance.c
index 010ff599cf9..b63f4643bee 100644
--- a/xlators/mgmt/glusterd/src/glusterd-rebalance.c
+++ b/xlators/mgmt/glusterd/src/glusterd-rebalance.c
@@ -326,7 +326,13 @@ glusterd_handle_defrag_start (glusterd_volinfo_t *volinfo, char *op_errstr,
         }
 
         sleep (5);
-        ret = rpc_clnt_transport_unix_options_build (&options, sockfile);
+
+        /* Setting frame-timeout to 10mins (600seconds).
+         * Unix domain sockets ensures that the connection is reliable. The
+         * default timeout of 30mins used for unreliable network connections is
+         * too long for unix domain socket connections.
+         */
+        ret = rpc_clnt_transport_unix_options_build (&options, sockfile, 600);
         if (ret) {
                 gf_log (THIS->name, GF_LOG_ERROR, "Unix options build failed");
                 goto out;
@@ -370,7 +376,13 @@ glusterd_rebalance_rpc_create (glusterd_volinfo_t *volinfo,
         LOCK_INIT (&defrag->lock);
 
         GLUSTERD_GET_DEFRAG_SOCK_FILE (sockfile, volinfo, priv);
-        ret = rpc_clnt_transport_unix_options_build (&options, sockfile);
+
+        /* Setting frame-timeout to 10mins (600seconds).
+         * Unix domain sockets ensures that the connection is reliable. The
+         * default timeout of 30mins used for unreliable network connections is
+         * too long for unix domain socket connections.
+         */
+        ret = rpc_clnt_transport_unix_options_build (&options, sockfile, 600);
         if (ret) {
                 gf_log (THIS->name, GF_LOG_ERROR, "Unix options build failed");
                 goto out;
diff --git a/xlators/mgmt/glusterd/src/glusterd-utils.c b/xlators/mgmt/glusterd/src/glusterd-utils.c
index ce3d0739719..3593683f266 100644
--- a/xlators/mgmt/glusterd/src/glusterd-utils.c
+++ b/xlators/mgmt/glusterd/src/glusterd-utils.c
@@ -1091,7 +1091,14 @@ glusterd_brick_connect (glusterd_volinfo_t  *volinfo,
                 glusterd_set_brick_socket_filepath (volinfo, brickinfo,
                                                     socketpath,
                                                     sizeof (socketpath));
-                ret = rpc_clnt_transport_unix_options_build (&options, socketpath);
+
+                /* Setting frame-timeout to 10mins (600seconds).
+                 * Unix domain sockets ensures that the connection is reliable.
+                 * The default timeout of 30mins used for unreliable network
+                 * connections is too long for unix domain socket connections.
+                 */
+                ret = rpc_clnt_transport_unix_options_build (&options,
+                                                             socketpath, 600);
                 if (ret)
                         goto out;
                 ret = glusterd_rpc_create (&rpc, options,
@@ -2787,8 +2794,13 @@ glusterd_nodesvc_connect (char *server, char *socketpath) {
         rpc = glusterd_nodesvc_get_rpc (server);
 
         if (rpc == NULL) {
+                /* Setting frame-timeout to 10mins (600seconds).
+                 * Unix domain sockets ensures that the connection is reliable.
+                 * The default timeout of 30mins used for unreliable network
+                 * connections is too long for unix domain socket connections.
+                 */
                 ret = rpc_clnt_transport_unix_options_build (&options,
-                                                             socketpath);
+                                                             socketpath, 600);
                 if (ret)
                         goto out;
                 ret = glusterd_rpc_create (&rpc, options,