* [Tarantool-patches] [PATCH v32 2/2] qsync: order access to the limbo terms
2022-03-04 6:01 [Tarantool-patches] [PATCH v32 0/2] qsync: implement packet filtering (part 1) Cyrill Gorcunov via Tarantool-patches
2022-03-04 6:01 ` [Tarantool-patches] [PATCH v32 1/2] latch: add latch_is_locked helper Cyrill Gorcunov via Tarantool-patches
@ 2022-03-04 6:01 ` Cyrill Gorcunov via Tarantool-patches
2022-03-04 7:11 ` [Tarantool-patches] [PATCH v32 0/2] qsync: implement packet filtering (part 1) Serge Petrenko via Tarantool-patches
2 siblings, 0 replies; 5+ messages in thread
From: Cyrill Gorcunov via Tarantool-patches @ 2022-03-04 6:01 UTC (permalink / raw)
To: tml; +Cc: Vladislav Shpilevoy
Limbo terms tracking is shared between appliers and when
one of appliers is waiting for write to complete inside
journal_write() routine, an other may need to access read
term value to figure out if promote request is valid to
apply. Due to cooperative multitasking access to the terms
is not consistent so we need to be sure that other fibers
read up to date terms (ie written to the WAL).
For this sake we use a latching mechanism, when one fiber
takes a lock for updating other readers are waiting until
the operation is complete.
For example here is a call graph of two appliers
applier 1
---------
applier_apply_tx
(promote term = 3
current max term = 2)
applier_synchro_filter_tx
apply_synchro_row
journal_write
(sleeping)
at this moment another applier comes in with obsolete
data and term 2
applier 2
---------
applier_apply_tx
(term 2)
applier_synchro_filter_tx
txn_limbo_is_replica_outdated -> false
journal_write (sleep)
applier 1
---------
journal wakes up
apply_synchro_row_cb
set max term to 3
So the applier 2 didn't notice that term 3 is already seen
and wrote obsolete data. With locking the applier 2 will
wait until applier 1 has finished its write.
We introduce the following helpers:
1) txn_limbo_begin: which takes a lock
2) txn_limbo_commit and txn_limbo_rollback which simply release
the lock but have different names for better semantics
3) txn_limbo_process is a general function which uses x_begin
and x_commit helper internally
4) txn_limbo_apply to do a real job over processing the
request, it implies that txn_limbo_begin been called
Testing such in-flight condition won't be easy so we introduce
"box.info.synchro.queue.busy" field to report if limbo is
currently latched and processing a sync request.
@TarantoolBot document
Title: synchronous replication changes
`box.info.synchro.queue` gets a new `busy` field. It is set to
`true` when there is a synchronous transaction is processing
but not yet complete. Thus any other incoming synchronous
transactions will be delayed until active one is finished.
Part-of #6036
Acked-by: Serge Petrenko <sergepetrenko@tarantool.org>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
---
.../unreleased/gh-6036-applier-fix-race.md | 5 +
src/box/applier.cc | 6 +-
src/box/box.cc | 8 +-
src/box/lua/info.c | 4 +-
src/box/txn_limbo.c | 13 +-
src/box/txn_limbo.h | 44 +++-
.../gh_6036_qsync_order_test.lua | 194 ++++++++++++++++++
test/replication-luatest/suite.ini | 1 +
8 files changed, 265 insertions(+), 10 deletions(-)
create mode 100644 changelogs/unreleased/gh-6036-applier-fix-race.md
create mode 100644 test/replication-luatest/gh_6036_qsync_order_test.lua
diff --git a/changelogs/unreleased/gh-6036-applier-fix-race.md b/changelogs/unreleased/gh-6036-applier-fix-race.md
new file mode 100644
index 000000000..98f65447c
--- /dev/null
+++ b/changelogs/unreleased/gh-6036-applier-fix-race.md
@@ -0,0 +1,5 @@
+## bugfix/replication
+
+* Fixed potential obsolete data write in synchronious replication
+ due to race in accessing terms while disk write operation is in
+ progress and not yet completed.
diff --git a/src/box/applier.cc b/src/box/applier.cc
index 0ac7ac3c7..99484eea2 100644
--- a/src/box/applier.cc
+++ b/src/box/applier.cc
@@ -959,7 +959,7 @@ apply_synchro_req_cb(struct journal_entry *entry)
applier_rollback_by_wal_io(entry->res);
} else {
replica_txn_wal_write_cb(synchro_entry->rcb);
- txn_limbo_process(&txn_limbo, synchro_entry->req);
+ txn_limbo_apply(&txn_limbo, synchro_entry->req);
trigger_run(&replicaset.applier.on_wal_write, NULL);
}
fiber_wakeup(synchro_entry->owner);
@@ -1004,14 +1004,18 @@ apply_synchro_req(uint32_t replica_id, struct xrow_header *row, struct synchro_r
* before trying to commit. But that requires extra steps from the
* transactions side, including the async ones.
*/
+ txn_limbo_begin(&txn_limbo);
if (journal_write(&entry.base) != 0)
goto err;
if (entry.base.res < 0) {
diag_set_journal_res(entry.base.res);
goto err;
}
+ txn_limbo_commit(&txn_limbo);
return 0;
+
err:
+ txn_limbo_rollback(&txn_limbo);
diag_log();
return -1;
}
diff --git a/src/box/box.cc b/src/box/box.cc
index 6a33203df..fd34ddb0c 100644
--- a/src/box/box.cc
+++ b/src/box/box.cc
@@ -1780,6 +1780,7 @@ box_issue_promote(uint32_t prev_leader_id, int64_t promote_lsn)
struct raft *raft = box_raft();
assert(raft->volatile_term == raft->term);
assert(promote_lsn >= 0);
+ txn_limbo_begin(&txn_limbo);
txn_limbo_write_promote(&txn_limbo, promote_lsn,
raft->term);
struct synchro_request req = {
@@ -1789,7 +1790,8 @@ box_issue_promote(uint32_t prev_leader_id, int64_t promote_lsn)
.lsn = promote_lsn,
.term = raft->term,
};
- txn_limbo_process(&txn_limbo, &req);
+ txn_limbo_apply(&txn_limbo, &req);
+ txn_limbo_commit(&txn_limbo);
assert(txn_limbo_is_empty(&txn_limbo));
}
@@ -1802,6 +1804,7 @@ box_issue_demote(uint32_t prev_leader_id, int64_t promote_lsn)
{
assert(box_raft()->volatile_term == box_raft()->term);
assert(promote_lsn >= 0);
+ txn_limbo_begin(&txn_limbo);
txn_limbo_write_demote(&txn_limbo, promote_lsn,
box_raft()->term);
struct synchro_request req = {
@@ -1811,7 +1814,8 @@ box_issue_demote(uint32_t prev_leader_id, int64_t promote_lsn)
.lsn = promote_lsn,
.term = box_raft()->term,
};
- txn_limbo_process(&txn_limbo, &req);
+ txn_limbo_apply(&txn_limbo, &req);
+ txn_limbo_commit(&txn_limbo);
assert(txn_limbo_is_empty(&txn_limbo));
}
diff --git a/src/box/lua/info.c b/src/box/lua/info.c
index b75a301eb..8d7bb1d4a 100644
--- a/src/box/lua/info.c
+++ b/src/box/lua/info.c
@@ -637,11 +637,13 @@ lbox_info_synchro(struct lua_State *L)
/* Queue information. */
struct txn_limbo *queue = &txn_limbo;
- lua_createtable(L, 0, 2);
+ lua_createtable(L, 0, 3);
lua_pushnumber(L, queue->len);
lua_setfield(L, -2, "len");
lua_pushnumber(L, queue->owner_id);
lua_setfield(L, -2, "owner");
+ lua_pushboolean(L, latch_is_locked(&queue->promote_latch));
+ lua_setfield(L, -2, "busy");
lua_setfield(L, -2, "queue");
return 1;
diff --git a/src/box/txn_limbo.c b/src/box/txn_limbo.c
index 70447caaf..5e708a951 100644
--- a/src/box/txn_limbo.c
+++ b/src/box/txn_limbo.c
@@ -47,6 +47,7 @@ txn_limbo_create(struct txn_limbo *limbo)
vclock_create(&limbo->vclock);
vclock_create(&limbo->promote_term_map);
limbo->promote_greatest_term = 0;
+ latch_create(&limbo->promote_latch);
limbo->confirmed_lsn = 0;
limbo->rollback_count = 0;
limbo->is_in_rollback = false;
@@ -724,8 +725,10 @@ txn_limbo_wait_empty(struct txn_limbo *limbo, double timeout)
}
void
-txn_limbo_process(struct txn_limbo *limbo, const struct synchro_request *req)
+txn_limbo_apply(struct txn_limbo *limbo, const struct synchro_request *req)
{
+ assert(latch_is_locked(&limbo->promote_latch));
+
uint64_t term = req->term;
uint32_t origin = req->origin_id;
if (txn_limbo_replica_term(limbo, origin) < term) {
@@ -786,6 +789,14 @@ txn_limbo_process(struct txn_limbo *limbo, const struct synchro_request *req)
return;
}
+void
+txn_limbo_process(struct txn_limbo *limbo, const struct synchro_request *req)
+{
+ txn_limbo_begin(limbo);
+ txn_limbo_apply(limbo, req);
+ txn_limbo_commit(limbo);
+}
+
void
txn_limbo_on_parameters_change(struct txn_limbo *limbo)
{
diff --git a/src/box/txn_limbo.h b/src/box/txn_limbo.h
index 53e52f676..e2ac75c08 100644
--- a/src/box/txn_limbo.h
+++ b/src/box/txn_limbo.h
@@ -31,6 +31,7 @@
*/
#include "small/rlist.h"
#include "vclock/vclock.h"
+#include "latch.h"
#include <stdint.h>
@@ -147,6 +148,10 @@ struct txn_limbo {
* limbo and raft are in sync and the terms are the same.
*/
uint64_t promote_greatest_term;
+ /**
+ * To order access to the promote data.
+ */
+ struct latch promote_latch;
/**
* Maximal LSN gathered quorum and either already confirmed in WAL, or
* whose confirmation is in progress right now. Any attempt to confirm
@@ -226,11 +231,13 @@ txn_limbo_replica_term(const struct txn_limbo *limbo, uint32_t replica_id)
* data from it. The check is only valid when elections are enabled.
*/
static inline bool
-txn_limbo_is_replica_outdated(const struct txn_limbo *limbo,
- uint32_t replica_id)
+txn_limbo_is_replica_outdated(struct txn_limbo *limbo, uint32_t replica_id)
{
- return txn_limbo_replica_term(limbo, replica_id) <
- limbo->promote_greatest_term;
+ latch_lock(&limbo->promote_latch);
+ uint64_t v = vclock_get(&limbo->promote_term_map, replica_id);
+ bool res = v < limbo->promote_greatest_term;
+ latch_unlock(&limbo->promote_latch);
+ return res;
}
/**
@@ -300,7 +307,34 @@ txn_limbo_ack(struct txn_limbo *limbo, uint32_t replica_id, int64_t lsn);
int
txn_limbo_wait_complete(struct txn_limbo *limbo, struct txn_limbo_entry *entry);
-/** Execute a synchronous replication request. */
+/**
+ * Initiate execution of a synchronous replication request.
+ */
+static inline void
+txn_limbo_begin(struct txn_limbo *limbo)
+{
+ latch_lock(&limbo->promote_latch);
+}
+
+/** Commit a synchronous replication request. */
+static inline void
+txn_limbo_commit(struct txn_limbo *limbo)
+{
+ latch_unlock(&limbo->promote_latch);
+}
+
+/** Rollback a synchronous replication request. */
+static inline void
+txn_limbo_rollback(struct txn_limbo *limbo)
+{
+ latch_unlock(&limbo->promote_latch);
+}
+
+/** Apply a synchronous replication request after processing stage. */
+void
+txn_limbo_apply(struct txn_limbo *limbo, const struct synchro_request *req);
+
+/** Process a synchronous replication request. */
void
txn_limbo_process(struct txn_limbo *limbo, const struct synchro_request *req);
diff --git a/test/replication-luatest/gh_6036_qsync_order_test.lua b/test/replication-luatest/gh_6036_qsync_order_test.lua
new file mode 100644
index 000000000..8f0bb204f
--- /dev/null
+++ b/test/replication-luatest/gh_6036_qsync_order_test.lua
@@ -0,0 +1,194 @@
+local t = require('luatest')
+local cluster = require('test.luatest_helpers.cluster')
+local server = require('test.luatest_helpers.server')
+local fiber = require('fiber')
+
+local g = t.group('gh-6036')
+
+g.before_all(function(cg)
+ cg.cluster = cluster:new({})
+
+ cg.box_cfg = {
+ replication = {
+ server.build_instance_uri('r1'),
+ server.build_instance_uri('r2'),
+ },
+ replication_timeout = 0.1,
+ replication_connect_quorum = 1,
+ election_mode = 'manual',
+ election_timeout = 0.1,
+ replication_synchro_quorum = 1,
+ replication_synchro_timeout = 0.1,
+ log_level = 6,
+ }
+
+ cg.r1 = cg.cluster:build_and_add_server({ alias = 'r1', box_cfg = cg.box_cfg })
+ cg.r2 = cg.cluster:build_and_add_server({ alias = 'r2', box_cfg = cg.box_cfg })
+ cg.cluster:start()
+end)
+
+g.after_all(function(cg)
+ cg.cluster:drop()
+ cg.cluster.servers = nil
+end)
+
+local function update_replication(...)
+ return (box.cfg{ replication = { ... } })
+end
+
+--
+-- The test requires 3rd replica to graft in.
+g.before_test("test_qsync_order", function(cg)
+ cg.box_cfg.replication[3] = server.build_instance_uri("r3")
+ cg.r3 = cg.cluster:build_and_add_server({ alias = 'r3', box_cfg = cg.box_cfg })
+ cg.r3:start()
+ cg.r1:exec(update_replication, cg.box_cfg.replication)
+ cg.r2:exec(update_replication, cg.box_cfg.replication)
+end)
+
+g.test_qsync_order = function(cg)
+ cg.cluster:wait_fullmesh()
+
+ --
+ -- Create a synchro space on the r1 node and make
+ -- sure the write processed just fine.
+ cg.r1:exec(function()
+ box.ctl.promote()
+ box.ctl.wait_rw()
+ local s = box.schema.create_space('test', {is_sync = true})
+ s:create_index('pk')
+ s:insert{1}
+ end)
+
+ local vclock = cg.r1:get_vclock()
+ vclock[0] = nil
+ cg.r2:wait_vclock(vclock)
+ cg.r3:wait_vclock(vclock)
+
+ t.assert_equals(cg.r1:exec(function() return box.space.test:select() end), {{1}})
+ t.assert_equals(cg.r2:exec(function() return box.space.test:select() end), {{1}})
+ t.assert_equals(cg.r3:exec(function() return box.space.test:select() end), {{1}})
+
+ --
+ -- Drop connection between r1 and r2.
+ cg.r1:exec(update_replication, {
+ server.build_instance_uri("r1"),
+ server.build_instance_uri("r3"),
+ })
+
+ --
+ -- Drop connection between r2 and r1.
+ cg.r2:exec(update_replication, {
+ server.build_instance_uri("r2"),
+ server.build_instance_uri("r3"),
+ })
+
+ --
+ -- Here we have the following scheme
+ --
+ -- r3 (WAL delay)
+ -- / \
+ -- r1 r2
+ --
+
+ --
+ -- Initiate disk delay in a bit tricky way: the next write will
+ -- fall into forever sleep.
+ cg.r3:exec(function()
+ box.error.injection.set('ERRINJ_WAL_DELAY', true)
+ end)
+
+ --
+ -- Make r2 been a leader and start writting data, the PROMOTE
+ -- request get queued on r3 and not yet processed, same time
+ -- the INSERT won't complete either waiting for the PROMOTE
+ -- completion first. Note that we enter r3 as well just to be
+ -- sure the PROMOTE has reached it via queue state test.
+ cg.r2:exec(function()
+ box.ctl.promote()
+ box.ctl.wait_rw()
+ end)
+ t.helpers.retrying({}, function()
+ t.assert(cg.r3:exec(function()
+ return box.info.synchro.queue.busy == true
+ end))
+ end)
+ cg.r2:exec(function()
+ box.space.test:insert{2}
+ end)
+
+ --
+ -- The r1 node has no clue that there is a new leader and continue
+ -- writing data with obsolete term. Since r3 is delayed now
+ -- the INSERT won't proceed yet but get queued.
+ cg.r1:exec(function()
+ box.space.test:insert{3}
+ end)
+
+ --
+ -- Finally enable r3 back. Make sure the data from new r2 leader get
+ -- writing while old leader's data ignored.
+ cg.r3:exec(function()
+ box.error.injection.set('ERRINJ_WAL_DELAY', false)
+ end)
+ t.helpers.retrying({}, function()
+ t.assert(cg.r3:exec(function()
+ return box.space.test:get{2} ~= nil
+ end))
+ end)
+
+ t.assert_equals(cg.r3:exec(function() return box.space.test:select() end), {{1},{2}})
+end
+
+--
+-- Drop the r3 replica, since it is no longer needed for this test.
+g.after_test("test_qsync_order", function(cg)
+ cg.box_cfg.replication[3] = nil
+ cg.r1:exec(update_replication, cg.box_cfg.replication)
+ cg.r2:exec(update_replication, cg.box_cfg.replication)
+ cg.r3:stop()
+ cg.r3:cleanup()
+ cg.r3 = nil
+end)
+
+g.test_promote_order = function(cg)
+ --
+ -- Make sure that while we're processing PROMOTE no other records
+ -- get sneaked in via applier code from other replicas. For this
+ -- sake initiate voting and stop inside wal thread just before
+ -- PROMOTE get written. Another replica sends us new record and
+ -- it should be dropped.
+ cg.r1:exec(function()
+ box.ctl.promote()
+ box.ctl.wait_rw()
+ end)
+ local vclock = cg.r1:get_vclock()
+ vclock[0] = nil
+ cg.r2:wait_vclock(vclock)
+
+ --
+ -- Drop connection between r1 and the rest of the cluster.
+ -- Otherwise r1 might become Raft follower before attempting
+ -- insert{4}.
+ cg.r1:exec(function() box.cfg{replication=""} end)
+ cg.r2:exec(function()
+ box.error.injection.set('ERRINJ_WAL_DELAY_COUNTDOWN', 2)
+ require('fiber').create(function() box.ctl.promote() end)
+ end)
+ t.helpers.retrying({}, function()
+ t.assert(cg.r2:exec(function()
+ return box.info.synchro.queue.busy
+ end))
+ end)
+ t.assert(cg.r1:exec(function() return box.info.ro == false end))
+ cg.r1:exec(function()
+ box.space.test:insert{4}
+ end)
+ cg.r2:exec(function()
+ require('luatest').assert(box.info.synchro.queue.busy == true)
+ box.error.injection.set('ERRINJ_WAL_DELAY', false)
+ box.ctl.wait_rw()
+ end)
+
+ t.assert_equals(cg.r2:exec(function() return box.space.test:select() end), {{1},{2}})
+end
diff --git a/test/replication-luatest/suite.ini b/test/replication-luatest/suite.ini
index 374f1b87a..07ec93a52 100644
--- a/test/replication-luatest/suite.ini
+++ b/test/replication-luatest/suite.ini
@@ -2,3 +2,4 @@
core = luatest
description = replication luatests
is_parallel = True
+release_disabled = gh_6036_qsync_order_test.lua
--
2.35.1
^ permalink raw reply [flat|nested] 5+ messages in thread