Re: [Tarantool-patches] [PATCH 2/8] replication: introduce replication_sync_quorum cfg

[Vladislav Shpilevoy <v.shpilevoy@tarantool.org>]

I appended a new commit on top of this one on the
branch. In the quorum commit I renamed the
option to

    replication_synchro_quorum

====================
commit fc19662ec528c5217c7b611ae16d417497d9fe35
Author: Vladislav Shpilevoy <v.shpilevoy@tarantool.org>
Date:   Tue Jun 16 00:47:24 2020 +0200

    replication: introduce replication_synchro_timeout cfg
    
    [TO BE SQUASHED INTO THE PREVIOUS COMMIT]
    
    Part of #4844
    Part of #5073

diff --git a/src/box/box.cc b/src/box/box.cc
index c7a5f2e3c..9db55e05a 100644
--- a/src/box/box.cc
+++ b/src/box/box.cc
@@ -489,6 +489,18 @@ box_check_replication_synchro_quorum(void)
 	return quorum;
 }
 
+static double
+box_check_replication_synchro_timeout(void)
+{
+	double timeout = cfg_getd("replication_synchro_timeout");
+	if (timeout <= 0) {
+		diag_set(ClientError, ER_CFG, "replication_synchro_timeout",
+			 "the value must be greater than zero");
+		return -1;
+	}
+	return timeout;
+}
+
 static double
 box_check_replication_sync_timeout(void)
 {
@@ -673,6 +685,8 @@ box_check_config()
 	box_check_replication_sync_lag();
 	if (box_check_replication_synchro_quorum() < 0)
 		diag_raise();
+	if (box_check_replication_synchro_timeout() < 0)
+		diag_raise();
 	box_check_replication_sync_timeout();
 	box_check_readahead(cfg_geti("readahead"));
 	box_check_checkpoint_count(cfg_geti("checkpoint_count"));
@@ -802,6 +816,16 @@ box_set_replication_synchro_quorum(void)
 	return 0;
 }
 
+int
+box_set_replication_synchro_timeout(void)
+{
+	double value = box_check_replication_synchro_timeout();
+	if (value < 0)
+		return -1;
+	replication_synchro_timeout = value;
+	return 0;
+}
+
 void
 box_set_replication_sync_timeout(void)
 {
@@ -2444,6 +2468,8 @@ box_cfg_xc(void)
 	box_set_replication_sync_lag();
 	if (box_set_replication_synchro_quorum() != 0)
 		diag_raise();
+	if (box_set_replication_synchro_timeout() != 0)
+		diag_raise();
 	box_set_replication_sync_timeout();
 	box_set_replication_skip_conflict();
 	box_set_replication_anon();
diff --git a/src/box/box.h b/src/box/box.h
index 24802d0f1..f9789154e 100644
--- a/src/box/box.h
+++ b/src/box/box.h
@@ -244,6 +244,7 @@ void box_set_replication_connect_timeout(void);
 void box_set_replication_connect_quorum(void);
 void box_set_replication_sync_lag(void);
 int box_set_replication_synchro_quorum(void);
+int box_set_replication_synchro_timeout(void);
 void box_set_replication_sync_timeout(void);
 void box_set_replication_skip_conflict(void);
 void box_set_replication_anon(void);
diff --git a/src/box/lua/cfg.cc b/src/box/lua/cfg.cc
index 01e8958cd..d481155cd 100644
--- a/src/box/lua/cfg.cc
+++ b/src/box/lua/cfg.cc
@@ -321,6 +321,14 @@ lbox_cfg_set_replication_synchro_quorum(struct lua_State *L)
 	return 0;
 }
 
+static int
+lbox_cfg_set_replication_synchro_timeout(struct lua_State *L)
+{
+	if (box_set_replication_synchro_timeout() != 0)
+		luaT_error(L);
+	return 0;
+}
+
 static int
 lbox_cfg_set_replication_sync_timeout(struct lua_State *L)
 {
@@ -379,6 +387,7 @@ box_lua_cfg_init(struct lua_State *L)
 		{"cfg_set_replication_connect_timeout", lbox_cfg_set_replication_connect_timeout},
 		{"cfg_set_replication_sync_lag", lbox_cfg_set_replication_sync_lag},
 		{"cfg_set_replication_synchro_quorum", lbox_cfg_set_replication_synchro_quorum},
+		{"cfg_set_replication_synchro_timeout", lbox_cfg_set_replication_synchro_timeout},
 		{"cfg_set_replication_sync_timeout", lbox_cfg_set_replication_sync_timeout},
 		{"cfg_set_replication_skip_conflict", lbox_cfg_set_replication_skip_conflict},
 		{"cfg_set_replication_anon", lbox_cfg_set_replication_anon},
diff --git a/src/box/lua/load_cfg.lua b/src/box/lua/load_cfg.lua
index 991e919e4..1155248a5 100644
--- a/src/box/lua/load_cfg.lua
+++ b/src/box/lua/load_cfg.lua
@@ -90,6 +90,7 @@ local default_cfg = {
     replication_sync_lag = 10,
     replication_sync_timeout = 300,
     replication_synchro_quorum = 1,
+    replication_synchro_timeout = 5,
     replication_connect_timeout = 30,
     replication_connect_quorum = nil, -- connect all
     replication_skip_conflict = false,
@@ -166,6 +167,7 @@ local template_cfg = {
     replication_sync_lag = 'number',
     replication_sync_timeout = 'number',
     replication_synchro_quorum = 'number',
+    replication_synchro_timeout = 'number',
     replication_connect_timeout = 'number',
     replication_connect_quorum = 'number',
     replication_skip_conflict = 'boolean',
@@ -278,6 +280,7 @@ local dynamic_cfg = {
     replication_sync_lag    = private.cfg_set_replication_sync_lag,
     replication_sync_timeout = private.cfg_set_replication_sync_timeout,
     replication_synchro_quorum = private.cfg_set_replication_synchro_quorum,
+    replication_synchro_timeout = private.cfg_set_replication_synchro_timeout,
     replication_skip_conflict = private.cfg_set_replication_skip_conflict,
     replication_anon        = private.cfg_set_replication_anon,
     instance_uuid           = check_instance_uuid,
@@ -312,6 +315,7 @@ local dynamic_cfg_order = {
     replication_sync_lag    = 150,
     replication_sync_timeout    = 150,
     replication_synchro_quorum  = 150,
+    replication_synchro_timeout = 150,
     replication_connect_timeout = 150,
     replication_connect_quorum  = 150,
     replication             = 200,
@@ -348,6 +352,7 @@ local dynamic_cfg_skip_at_load = {
     replication_sync_lag    = true,
     replication_sync_timeout = true,
     replication_synchro_quorum = true,
+    replication_synchro_timeout = true,
     replication_skip_conflict = true,
     replication_anon        = true,
     wal_dir_rescan_delay    = true,
diff --git a/src/box/replication.cc b/src/box/replication.cc
index 5b52f3864..01e9e876a 100644
--- a/src/box/replication.cc
+++ b/src/box/replication.cc
@@ -52,6 +52,7 @@ double replication_connect_timeout = 30.0; /* seconds */
 int replication_connect_quorum = REPLICATION_CONNECT_QUORUM_ALL;
 double replication_sync_lag = 10.0; /* seconds */
 int replication_synchro_quorum = 1;
+double replication_synchro_timeout = 5.0; /* seconds */
 double replication_sync_timeout = 300.0; /* seconds */
 bool replication_skip_conflict = false;
 bool replication_anon = false;
diff --git a/src/box/replication.h b/src/box/replication.h
index 05e3eb943..a081870f9 100644
--- a/src/box/replication.h
+++ b/src/box/replication.h
@@ -131,6 +131,12 @@ extern double replication_sync_lag;
  */
 extern int replication_synchro_quorum;
 
+/**
+ * Time in seconds which the master node is able to wait for ACKs
+ * for a synchronous transaction until it is rolled back.
+ */
+extern double replication_synchro_timeout;
+
 /**
  * Max time to wait for appliers to synchronize before entering
  * the orphan mode.
diff --git a/test/app-tap/init_script.result b/test/app-tap/init_script.result
index 2987b60b9..857f0c95f 100644
--- a/test/app-tap/init_script.result
+++ b/test/app-tap/init_script.result
@@ -31,6 +31,7 @@ replication_skip_conflict:false
 replication_sync_lag:10
 replication_sync_timeout:300
 replication_synchro_quorum:1
+replication_synchro_timeout:5
 replication_timeout:1
 slab_alloc_factor:1.05
 sql_cache_size:5242880
diff --git a/test/box/admin.result b/test/box/admin.result
index 35ecc7617..ab3e80a97 100644
--- a/test/box/admin.result
+++ b/test/box/admin.result
@@ -83,6 +83,8 @@ cfg_filter(box.cfg)
     - 300
   - - replication_synchro_quorum
     - 1
+  - - replication_synchro_timeout
+    - 5
   - - replication_timeout
     - 1
   - - slab_alloc_factor
diff --git a/test/box/cfg.result b/test/box/cfg.result
index cdc0773f2..bdd210b09 100644
--- a/test/box/cfg.result
+++ b/test/box/cfg.result
@@ -71,6 +71,8 @@ cfg_filter(box.cfg)
  |     - 300
  |   - - replication_synchro_quorum
  |     - 1
+ |   - - replication_synchro_timeout
+ |     - 5
  |   - - replication_timeout
  |     - 1
  |   - - slab_alloc_factor
@@ -176,6 +178,8 @@ cfg_filter(box.cfg)
  |     - 300
  |   - - replication_synchro_quorum
  |     - 1
+ |   - - replication_synchro_timeout
+ |     - 5
  |   - - replication_timeout
  |     - 1
  |   - - slab_alloc_factor