[Tarantool-patches] [PATCH 2/8] replication: introduce replication_sync_quorum cfg
Serge Petrenko
sergepetrenko at tarantool.org
Tue Jun 9 15:20:14 MSK 2020
From: Vladislav Shpilevoy <v.shpilevoy at tarantool.org>
Synchronous transactions are supposed to be replicated on a
specified number of replicas before committed on master. The
number of replicas can be specified using replication_sync_quorum
option. It is 1 by default, so sync transactions work like
asynchronous when not configured anyhow. 1 means successful WAL
write on master is enough for commit.
Part of #4844
---
src/box/box.cc | 27 +++++++++++++++++++++++++++
src/box/box.h | 1 +
src/box/lua/cfg.cc | 9 +++++++++
src/box/lua/load_cfg.lua | 5 +++++
src/box/replication.cc | 1 +
src/box/replication.h | 6 ++++++
test/app-tap/init_script.result | 1 +
test/box/admin.result | 2 ++
test/box/cfg.result | 4 ++++
9 files changed, 56 insertions(+)
diff --git a/src/box/box.cc b/src/box/box.cc
index 96557651b..9b67aeb1f 100644
--- a/src/box/box.cc
+++ b/src/box/box.cc
@@ -476,6 +476,19 @@ box_check_replication_sync_lag(void)
return lag;
}
+static int
+box_check_replication_sync_quorum(void)
+{
+ int quorum = cfg_geti("replication_sync_quorum");
+ if (quorum <= 0 || quorum > VCLOCK_MAX) {
+ diag_set(ClientError, ER_CFG, "replication_sync_quorum",
+ "the value must be greater than and less than "
+ "maximal number of replicas");
+ return -1;
+ }
+ return quorum;
+}
+
static double
box_check_replication_sync_timeout(void)
{
@@ -658,6 +671,8 @@ box_check_config()
box_check_replication_connect_timeout();
box_check_replication_connect_quorum();
box_check_replication_sync_lag();
+ if (box_check_replication_sync_quorum() < 0)
+ diag_raise();
box_check_replication_sync_timeout();
box_check_readahead(cfg_geti("readahead"));
box_check_checkpoint_count(cfg_geti("checkpoint_count"));
@@ -777,6 +792,16 @@ box_set_replication_sync_lag(void)
replication_sync_lag = box_check_replication_sync_lag();
}
+int
+box_set_replication_sync_quorum(void)
+{
+ int value = box_check_replication_sync_quorum();
+ if (value < 0)
+ return -1;
+ replication_sync_quorum = value;
+ return 0;
+}
+
void
box_set_replication_sync_timeout(void)
{
@@ -2417,6 +2442,8 @@ box_cfg_xc(void)
box_set_replication_connect_timeout();
box_set_replication_connect_quorum();
box_set_replication_sync_lag();
+ if (box_set_replication_sync_quorum() != 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 557542a83..e81ff9ea4 100644
--- a/src/box/box.h
+++ b/src/box/box.h
@@ -243,6 +243,7 @@ void box_set_replication_timeout(void);
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_sync_quorum(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 a5b15e527..67dc89242 100644
--- a/src/box/lua/cfg.cc
+++ b/src/box/lua/cfg.cc
@@ -313,6 +313,14 @@ lbox_cfg_set_replication_sync_lag(struct lua_State *L)
return 0;
}
+static int
+lbox_cfg_set_replication_sync_quorum(struct lua_State *L)
+{
+ if (box_set_replication_sync_quorum() != 0)
+ luaT_error(L);
+ return 0;
+}
+
static int
lbox_cfg_set_replication_sync_timeout(struct lua_State *L)
{
@@ -370,6 +378,7 @@ box_lua_cfg_init(struct lua_State *L)
{"cfg_set_replication_connect_quorum", lbox_cfg_set_replication_connect_quorum},
{"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_sync_quorum", lbox_cfg_set_replication_sync_quorum},
{"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 7dc40a47f..182c1dfdd 100644
--- a/src/box/lua/load_cfg.lua
+++ b/src/box/lua/load_cfg.lua
@@ -88,6 +88,7 @@ local default_cfg = {
worker_pool_threads = 4,
replication_timeout = 1,
replication_sync_lag = 10,
+ replication_sync_quorum = 1,
replication_sync_timeout = 300,
replication_connect_timeout = 30,
replication_connect_quorum = nil, -- connect all
@@ -163,6 +164,7 @@ local template_cfg = {
worker_pool_threads = 'number',
replication_timeout = 'number',
replication_sync_lag = 'number',
+ replication_sync_quorum = 'number',
replication_sync_timeout = 'number',
replication_connect_timeout = 'number',
replication_connect_quorum = 'number',
@@ -274,6 +276,7 @@ local dynamic_cfg = {
replication_connect_timeout = private.cfg_set_replication_connect_timeout,
replication_connect_quorum = private.cfg_set_replication_connect_quorum,
replication_sync_lag = private.cfg_set_replication_sync_lag,
+ replication_sync_quorum = private.cfg_set_replication_sync_quorum,
replication_sync_timeout = private.cfg_set_replication_sync_timeout,
replication_skip_conflict = private.cfg_set_replication_skip_conflict,
replication_anon = private.cfg_set_replication_anon,
@@ -307,6 +310,7 @@ local dynamic_cfg_order = {
-- rule - apply before replication itself.
replication_timeout = 150,
replication_sync_lag = 150,
+ replication_sync_quorum = 150,
replication_sync_timeout = 150,
replication_connect_timeout = 150,
replication_connect_quorum = 150,
@@ -342,6 +346,7 @@ local dynamic_cfg_skip_at_load = {
replication_connect_timeout = true,
replication_connect_quorum = true,
replication_sync_lag = true,
+ replication_sync_quorum = true,
replication_sync_timeout = true,
replication_skip_conflict = true,
replication_anon = true,
diff --git a/src/box/replication.cc b/src/box/replication.cc
index 273a7cb66..f2c18fa21 100644
--- a/src/box/replication.cc
+++ b/src/box/replication.cc
@@ -51,6 +51,7 @@ double replication_timeout = 1.0; /* seconds */
double replication_connect_timeout = 30.0; /* seconds */
int replication_connect_quorum = REPLICATION_CONNECT_QUORUM_ALL;
double replication_sync_lag = 10.0; /* seconds */
+int replication_sync_quorum = 1;
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 93a25c8a7..3624019fb 100644
--- a/src/box/replication.h
+++ b/src/box/replication.h
@@ -125,6 +125,12 @@ extern int replication_connect_quorum;
*/
extern double replication_sync_lag;
+/**
+ * Minimal number of replicas to confirm a synchronous transaction
+ * is applied on to be able to finish its commit.
+ */
+extern int replication_sync_quorum;
+
/**
* 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 7c4454285..7d2a610fa 100644
--- a/test/app-tap/init_script.result
+++ b/test/app-tap/init_script.result
@@ -29,6 +29,7 @@ replication_anon:false
replication_connect_timeout:30
replication_skip_conflict:false
replication_sync_lag:10
+replication_sync_quorum:1
replication_sync_timeout:300
replication_timeout:1
slab_alloc_factor:1.05
diff --git a/test/box/admin.result b/test/box/admin.result
index d94da8c5d..c4f849f2d 100644
--- a/test/box/admin.result
+++ b/test/box/admin.result
@@ -79,6 +79,8 @@ cfg_filter(box.cfg)
- false
- - replication_sync_lag
- 10
+ - - replication_sync_quorum
+ - 1
- - replication_sync_timeout
- 300
- - replication_timeout
diff --git a/test/box/cfg.result b/test/box/cfg.result
index b41d54599..c511458a8 100644
--- a/test/box/cfg.result
+++ b/test/box/cfg.result
@@ -67,6 +67,8 @@ cfg_filter(box.cfg)
| - false
| - - replication_sync_lag
| - 10
+ | - - replication_sync_quorum
+ | - 1
| - - replication_sync_timeout
| - 300
| - - replication_timeout
@@ -170,6 +172,8 @@ cfg_filter(box.cfg)
| - false
| - - replication_sync_lag
| - 10
+ | - - replication_sync_quorum
+ | - 1
| - - replication_sync_timeout
| - 300
| - - replication_timeout
--
2.24.3 (Apple Git-128)
More information about the Tarantool-patches
mailing list