[Tarantool-patches] [PATCH v2 8/8] replication: send latest effective promote in initial join
Serge Petrenko
sergepetrenko at tarantool.org
Fri Jun 18 00:07:42 MSK 2021
A joining instance may never receive the latest PROMOTE request, which
is the only source of information about the limbo owner. Send out the
latest limbo state (e.g. the latest applied PROMOTE request) together
with the initial join snapshot.
Follow-up #6034
---
src/box/applier.cc | 5 ++
src/box/relay.cc | 15 +++++
test/replication/replica_rejoin.result | 77 ++++++++++++++----------
test/replication/replica_rejoin.test.lua | 50 +++++++--------
4 files changed, 92 insertions(+), 55 deletions(-)
diff --git a/src/box/applier.cc b/src/box/applier.cc
index 10cea26a7..9d90a2384 100644
--- a/src/box/applier.cc
+++ b/src/box/applier.cc
@@ -458,6 +458,11 @@ applier_wait_snapshot(struct applier *applier)
xrow_decode_vclock_xc(&row, &replicaset.vclock);
}
break; /* end of stream */
+ } else if (iproto_type_is_promote_request(row.type)) {
+ struct synchro_request req;
+ if (xrow_decode_synchro(&row, &req) != 0)
+ diag_raise();
+ txn_limbo_process(&txn_limbo, &req);
} else if (iproto_type_is_error(row.type)) {
xrow_decode_error_xc(&row); /* rethrow error */
} else {
diff --git a/src/box/relay.cc b/src/box/relay.cc
index b47767769..e05b53d5d 100644
--- a/src/box/relay.cc
+++ b/src/box/relay.cc
@@ -399,12 +399,27 @@ relay_initial_join(int fd, uint64_t sync, struct vclock *vclock)
if (txn_limbo_wait_confirm(&txn_limbo) != 0)
diag_raise();
+ struct synchro_request req;
+ txn_limbo_checkpoint(&txn_limbo, &req);
+
/* Respond to the JOIN request with the current vclock. */
struct xrow_header row;
xrow_encode_vclock_xc(&row, vclock);
row.sync = sync;
coio_write_xrow(&relay->io, &row);
+ /*
+ * Send out the latest limbo state. Don't do that when limbo is unused,
+ * let the old instances join without trouble.
+ */
+ if (req.replica_id != REPLICA_ID_NIL) {
+ char body[XROW_SYNCHRO_BODY_LEN_MAX];
+ xrow_encode_synchro(&row, body, &req);
+ row.replica_id = req.replica_id;
+ row.sync = sync;
+ coio_write_xrow(&relay->io, &row);
+ }
+
/* Send read view to the replica. */
engine_join_xc(&ctx, &relay->stream);
}
diff --git a/test/replication/replica_rejoin.result b/test/replication/replica_rejoin.result
index 843333a19..e489c150a 100644
--- a/test/replication/replica_rejoin.result
+++ b/test/replication/replica_rejoin.result
@@ -7,10 +7,19 @@ test_run = env.new()
log = require('log')
---
...
-engine = test_run:get_cfg('engine')
+test_run:cmd("create server master with script='replication/master1.lua'")
---
+- true
...
-test_run:cleanup_cluster()
+test_run:cmd("start server master")
+---
+- true
+...
+test_run:switch("master")
+---
+- true
+...
+engine = test_run:get_cfg('engine')
---
...
--
@@ -43,7 +52,7 @@ _ = box.space.test:insert{3}
---
...
-- Join a replica, then stop it.
-test_run:cmd("create server replica with rpl_master=default, script='replication/replica_rejoin.lua'")
+test_run:cmd("create server replica with rpl_master=master, script='replication/replica_rejoin.lua'")
---
- true
...
@@ -65,7 +74,7 @@ box.space.test:select()
- [2]
- [3]
...
-test_run:cmd("switch default")
+test_run:cmd("switch master")
---
- true
...
@@ -75,7 +84,7 @@ test_run:cmd("stop server replica")
...
-- Restart the server to purge the replica from
-- the garbage collection state.
-test_run:cmd("restart server default")
+test_run:cmd("restart server master")
box.cfg{wal_cleanup_delay = 0}
---
...
@@ -146,7 +155,7 @@ box.space.test:select()
- [20]
- [30]
...
-test_run:cmd("switch default")
+test_run:cmd("switch master")
---
- true
...
@@ -154,7 +163,7 @@ test_run:cmd("switch default")
for i = 10, 30, 10 do box.space.test:update(i, {{'!', 1, i}}) end
---
...
-vclock = test_run:get_vclock('default')
+vclock = test_run:get_vclock('master')
---
...
vclock[0] = nil
@@ -191,7 +200,7 @@ box.space.test:replace{1, 2, 3} -- bumps LSN on the replica
---
- [1, 2, 3]
...
-test_run:cmd("switch default")
+test_run:cmd("switch master")
---
- true
...
@@ -199,7 +208,7 @@ test_run:cmd("stop server replica")
---
- true
...
-test_run:cmd("restart server default")
+test_run:cmd("restart server master")
box.cfg{wal_cleanup_delay = 0}
---
...
@@ -253,7 +262,7 @@ box.space.test:select()
-- from the replica.
--
-- Bootstrap a new replica.
-test_run:cmd("switch default")
+test_run:cmd("switch master")
---
- true
...
@@ -295,7 +304,7 @@ box.cfg{replication = ''}
---
...
-- Bump vclock on the master.
-test_run:cmd("switch default")
+test_run:cmd("switch master")
---
- true
...
@@ -317,15 +326,15 @@ vclock = test_run:get_vclock('replica')
vclock[0] = nil
---
...
-_ = test_run:wait_vclock('default', vclock)
+_ = test_run:wait_vclock('master', vclock)
---
...
-- Restart the master and force garbage collection.
-test_run:cmd("switch default")
+test_run:cmd("switch master")
---
- true
...
-test_run:cmd("restart server default")
+test_run:cmd("restart server master")
box.cfg{wal_cleanup_delay = 0}
---
...
@@ -373,7 +382,7 @@ vclock = test_run:get_vclock('replica')
vclock[0] = nil
---
...
-_ = test_run:wait_vclock('default', vclock)
+_ = test_run:wait_vclock('master', vclock)
---
...
-- Restart the replica. It should successfully rebootstrap.
@@ -396,38 +405,42 @@ test_run:cmd("switch default")
---
- true
...
-box.cfg{replication = ''}
+test_run:cmd("stop server replica")
---
+- true
...
-test_run:cmd("stop server replica")
+test_run:cmd("delete server replica")
---
- true
...
-test_run:cmd("cleanup server replica")
+test_run:cmd("stop server master")
---
- true
...
-test_run:cmd("delete server replica")
+test_run:cmd("delete server master")
---
- true
...
-test_run:cleanup_cluster()
+--
+-- gh-4107: rebootstrap fails if the replica was deleted from
+-- the cluster on the master.
+--
+test_run:cmd("create server master with script='replication/master1.lua'")
---
+- true
...
-box.space.test:drop()
+test_run:cmd("start server master")
---
+- true
...
-box.schema.user.revoke('guest', 'replication')
+test_run:switch("master")
---
+- true
...
---
--- gh-4107: rebootstrap fails if the replica was deleted from
--- the cluster on the master.
---
box.schema.user.grant('guest', 'replication')
---
...
-test_run:cmd("create server replica with rpl_master=default, script='replication/replica_uuid.lua'")
+test_run:cmd("create server replica with rpl_master=master, script='replication/replica_uuid.lua'")
---
- true
...
@@ -462,11 +475,11 @@ box.space._cluster:get(2) ~= nil
---
- true
...
-test_run:cmd("stop server replica")
+test_run:switch("default")
---
- true
...
-test_run:cmd("cleanup server replica")
+test_run:cmd("stop server replica")
---
- true
...
@@ -474,9 +487,11 @@ test_run:cmd("delete server replica")
---
- true
...
-box.schema.user.revoke('guest', 'replication')
+test_run:cmd("stop server master")
---
+- true
...
-test_run:cleanup_cluster()
+test_run:cmd("delete server master")
---
+- true
...
diff --git a/test/replication/replica_rejoin.test.lua b/test/replication/replica_rejoin.test.lua
index c3ba9bf3f..2563177cf 100644
--- a/test/replication/replica_rejoin.test.lua
+++ b/test/replication/replica_rejoin.test.lua
@@ -1,9 +1,11 @@
env = require('test_run')
test_run = env.new()
log = require('log')
-engine = test_run:get_cfg('engine')
-test_run:cleanup_cluster()
+test_run:cmd("create server master with script='replication/master1.lua'")
+test_run:cmd("start server master")
+test_run:switch("master")
+engine = test_run:get_cfg('engine')
--
-- gh-5806: this replica_rejoin test relies on the wal cleanup fiber
@@ -23,17 +25,17 @@ _ = box.space.test:insert{2}
_ = box.space.test:insert{3}
-- Join a replica, then stop it.
-test_run:cmd("create server replica with rpl_master=default, script='replication/replica_rejoin.lua'")
+test_run:cmd("create server replica with rpl_master=master, script='replication/replica_rejoin.lua'")
test_run:cmd("start server replica")
test_run:cmd("switch replica")
box.info.replication[1].upstream.status == 'follow' or log.error(box.info)
box.space.test:select()
-test_run:cmd("switch default")
+test_run:cmd("switch master")
test_run:cmd("stop server replica")
-- Restart the server to purge the replica from
-- the garbage collection state.
-test_run:cmd("restart server default")
+test_run:cmd("restart server master")
box.cfg{wal_cleanup_delay = 0}
-- Make some checkpoints to remove old xlogs.
@@ -58,11 +60,11 @@ box.info.replication[2].downstream.vclock ~= nil or log.error(box.info)
test_run:cmd("switch replica")
box.info.replication[1].upstream.status == 'follow' or log.error(box.info)
box.space.test:select()
-test_run:cmd("switch default")
+test_run:cmd("switch master")
-- Make sure the replica follows new changes.
for i = 10, 30, 10 do box.space.test:update(i, {{'!', 1, i}}) end
-vclock = test_run:get_vclock('default')
+vclock = test_run:get_vclock('master')
vclock[0] = nil
_ = test_run:wait_vclock('replica', vclock)
test_run:cmd("switch replica")
@@ -76,9 +78,9 @@ box.space.test:select()
-- Check that rebootstrap is NOT initiated unless the replica
-- is strictly behind the master.
box.space.test:replace{1, 2, 3} -- bumps LSN on the replica
-test_run:cmd("switch default")
+test_run:cmd("switch master")
test_run:cmd("stop server replica")
-test_run:cmd("restart server default")
+test_run:cmd("restart server master")
box.cfg{wal_cleanup_delay = 0}
checkpoint_count = box.cfg.checkpoint_count
box.cfg{checkpoint_count = 1}
@@ -99,7 +101,7 @@ box.space.test:select()
--
-- Bootstrap a new replica.
-test_run:cmd("switch default")
+test_run:cmd("switch master")
test_run:cmd("stop server replica")
test_run:cmd("cleanup server replica")
test_run:cleanup_cluster()
@@ -113,17 +115,17 @@ box.cfg{replication = replica_listen}
test_run:cmd("switch replica")
box.cfg{replication = ''}
-- Bump vclock on the master.
-test_run:cmd("switch default")
+test_run:cmd("switch master")
box.space.test:replace{1}
-- Bump vclock on the replica.
test_run:cmd("switch replica")
for i = 1, 10 do box.space.test:replace{2} end
vclock = test_run:get_vclock('replica')
vclock[0] = nil
-_ = test_run:wait_vclock('default', vclock)
+_ = test_run:wait_vclock('master', vclock)
-- Restart the master and force garbage collection.
-test_run:cmd("switch default")
-test_run:cmd("restart server default")
+test_run:cmd("switch master")
+test_run:cmd("restart server master")
box.cfg{wal_cleanup_delay = 0}
replica_listen = test_run:cmd("eval replica 'return box.cfg.listen'")
replica_listen ~= nil
@@ -139,7 +141,7 @@ test_run:cmd("switch replica")
for i = 1, 10 do box.space.test:replace{2} end
vclock = test_run:get_vclock('replica')
vclock[0] = nil
-_ = test_run:wait_vclock('default', vclock)
+_ = test_run:wait_vclock('master', vclock)
-- Restart the replica. It should successfully rebootstrap.
test_run:cmd("restart server replica with args='true'")
box.space.test:select()
@@ -148,20 +150,20 @@ box.space.test:replace{2}
-- Cleanup.
test_run:cmd("switch default")
-box.cfg{replication = ''}
test_run:cmd("stop server replica")
-test_run:cmd("cleanup server replica")
test_run:cmd("delete server replica")
-test_run:cleanup_cluster()
-box.space.test:drop()
-box.schema.user.revoke('guest', 'replication')
+test_run:cmd("stop server master")
+test_run:cmd("delete server master")
--
-- gh-4107: rebootstrap fails if the replica was deleted from
-- the cluster on the master.
--
+test_run:cmd("create server master with script='replication/master1.lua'")
+test_run:cmd("start server master")
+test_run:switch("master")
box.schema.user.grant('guest', 'replication')
-test_run:cmd("create server replica with rpl_master=default, script='replication/replica_uuid.lua'")
+test_run:cmd("create server replica with rpl_master=master, script='replication/replica_uuid.lua'")
start_cmd = string.format("start server replica with args='%s'", require('uuid').new())
box.space._cluster:get(2) == nil
test_run:cmd(start_cmd)
@@ -170,8 +172,8 @@ test_run:cmd("cleanup server replica")
box.space._cluster:delete(2) ~= nil
test_run:cmd(start_cmd)
box.space._cluster:get(2) ~= nil
+test_run:switch("default")
test_run:cmd("stop server replica")
-test_run:cmd("cleanup server replica")
test_run:cmd("delete server replica")
-box.schema.user.revoke('guest', 'replication')
-test_run:cleanup_cluster()
+test_run:cmd("stop server master")
+test_run:cmd("delete server master")
--
2.30.1 (Apple Git-130)
More information about the Tarantool-patches
mailing list