Preface
Semi-synchronous replication is supported since MySQL 5.5 and then enhanced gradually in 5.6 & 5.7.It's also called enhanced semi-synchronous replicaiton in MySQL 5.7.The new method of ACK rule in 5.7(after_sync) make the consistency be strongly guaranteed.Maybe that is the reason why it gets the name.Today I'm gonna demonstrate the effect of sevaral variables of semi-sync.
Introduce
Semi-synchronized replicatin can be installed as a plugin.There're a lot of variables with prefix of "repl_semi_sync" after you install the "rpl_semi_sync_master" plugin on your MySQL server.Meanwhile,there're a series of global status about it as well.Let's see the details about it.
Framework
| Hostname | IP/Port | Identity | OS Version | MySQL Version | GTID Mode | Binlog Format | Semi-Sync status |
| zlm2 | 192.168.1.101/3306 | master | CentOS 7.0 | 5.7.21 | on | row | on |
| zlm3 | 192.168.1.102/3306 | slave1 | CentOS 7.0 | 5.7.21 | on | row | on |
| zlm4 | 192.168.1.103/3306 | slave2 | CentOS 7.0 | 5.7.21 | on | row | on |
Example
Check the variables relevant to semi-sync replication.
1 //Master 2 (zlm@192.168.1.101 3306)[(none)]>show variables like 'rpl_semi%'; 3 +-------------------------------------------+------------+ 4 | Variable_name | Value | 5 +-------------------------------------------+------------+ 6 | rpl_semi_sync_master_enabled | OFF | 7 | rpl_semi_sync_master_timeout | 10000 | 8 | rpl_semi_sync_master_trace_level | 32 | 9 | rpl_semi_sync_master_wait_for_slave_count | 2 | 10 | rpl_semi_sync_master_wait_no_slave | ON | 11 | rpl_semi_sync_master_wait_point | AFTER_SYNC | 12 +-------------------------------------------+------------+ 13 6 rows in set (0.00 sec) 14 15 //rpl_semi_sync_master_enabled --Specify whether to use the feature of semi-sync(master side,default "off"). 16 //rpl_semi_sync_master_timeout --Specify the threshold of timeout between master and slave(default value is "10000" means 10s). 17 //rpl_semi_sync_master_trace_level --Specify the debug trace level on master(default "32",others "1","16","64"). 18 //rpl_semi_sync_master_wait_for_slave_count --Specify how many slaves the master should wait when timeout happens(default "1"). 19 //rpl_semi_sync_master_wait_no_slave --Specify whether the master switches to normal replication(default "on").When setting "on",it permits the amount of slave can be less than the value of "rpl_semi_sync_master_wait_for_slave_count" you specified and prevents it becoming normal replication. 20 //rpl_semi_sync_master_wait_point --Specify the ACK mode(default "after-sync").It's the most important variable of semi-sync. 21 22 //Slave1 23 (zlm@192.168.1.102 3306)[(none)]>show variables like 'rpl_semi%'; 24 +---------------------------------+-------+ 25 | Variable_name | Value | 26 +---------------------------------+-------+ 27 | rpl_semi_sync_slave_enabled | OFF | 28 | rpl_semi_sync_slave_trace_level | 32 | 29 +---------------------------------+-------+ 30 2 rows in set (0.00 sec) 31 32 //There're only two variables after install the rpl_semi_sync_slave plugin. 33 //rpl_semi_sync_slave_enabled -- Specify whether use the feature of semi-sync(slave side,default "off"). 34 //rpl_semi_sync_slave_trace_level -- Specify the debug trace level on slave(default "32").
Enable semi-sync replication on both master and slaves.
1 //Master 2 (zlm@192.168.1.101 3306)[(none)]>set rpl_semi_sync_master_enabled=on; 3 ERROR 1229 (HY000): Variable 'rpl_semi_sync_master_enabled' is a GLOBAL variable and should be set with SET GLOBAL 4 (zlm@192.168.1.101 3306)[(none)]>set global rpl_semi_sync_master_enabled=on; 5 Query OK, 0 rows affected (0.00 sec) 6 7 (zlm@192.168.1.101 3306)[(none)]>show variables like 'rpl_semi_sync_master_enabled'; 8 +------------------------------+-------+ 9 | Variable_name | Value | 10 +------------------------------+-------+ 11 | rpl_semi_sync_master_enabled | ON | 12 +------------------------------+-------+ 13 1 row in set (0.01 sec) 14 15 //Slave1 16 (zlm@192.168.1.102 3306)[(none)]>set @@global.rpl_semi_sync_slave_enabled=on; 17 Query OK, 0 rows affected (0.00 sec) 18 19 (zlm@192.168.1.102 3306)[(none)]>show variables like 'rpl_semi_sync%'; 20 +---------------------------------+-------+ 21 | Variable_name | Value | 22 +---------------------------------+-------+ 23 | rpl_semi_sync_slave_enabled | ON | 24 | rpl_semi_sync_slave_trace_level | 32 | 25 +---------------------------------+-------+ 26 2 rows in set (0.00 sec) 27 28 //Slave2 29 (zlm@192.168.1.103 3306)[(none)]>set @@global.rpl_semi_sync_slave_enabled=on; 30 Query OK, 0 rows affected (0.00 sec) 31 32 (zlm@192.168.1.103 3306)[(none)]>show variables like 'rpl_semi_sync%'; 33 +---------------------------------+-------+ 34 | Variable_name | Value | 35 +---------------------------------+-------+ 36 | rpl_semi_sync_slave_enabled | ON | 37 | rpl_semi_sync_slave_trace_level | 32 | 38 +---------------------------------+-------+ 39 2 rows in set (0.01 sec) 40 41 //Master 42 (zlm@192.168.1.101 3306)[(none)]>show global status like 'Rpl_semi_sync_master_clients'; 43 +------------------------------+-------+ 44 | Variable_name | Value | 45 +------------------------------+-------+ 46 | Rpl_semi_sync_master_clients | 2 | //It means there're two slaves using semi-sync replicaiton follow master now. 47 +------------------------------+-------+ 48 1 row in set (0.00 sec)
Execute sysbench to generate some trasactions.
1 [root@zlm2 09:45:08 ~/sysbench-1.0/src/lua] 2 #sysbench oltp_insert.lua --mysql-host=192.168.1.101 --mysql-port=3306 --mysql-user=zlm --mysql-password=zlmzlm --mysql-db=sysbench --tables=5 --table-size=10000 --mysql-storage-engine=innodb prepare 3 sysbench 1.0.15 (using bundled LuaJIT 2.1.0-beta2) 4 5 Creating table 'sbtest1'... 6 Inserting 10000 records into 'sbtest1' 7 Creating a secondary index on 'sbtest1'... 8 Creating table 'sbtest2'... 9 Inserting 10000 records into 'sbtest2' 10 Creating a secondary index on 'sbtest2'... 11 Creating table 'sbtest3'... 12 Inserting 10000 records into 'sbtest3' 13 Creating a secondary index on 'sbtest3'... 14 Creating table 'sbtest4'... 15 Inserting 10000 records into 'sbtest4' 16 Creating a secondary index on 'sbtest4'... 17 Creating table 'sbtest5'... 18 Inserting 10000 records into 'sbtest5' 19 Creating a secondary index on 'sbtest5'... 20 21 [root@zlm2 09:45:22 ~/sysbench-1.0/src/lua] 22 #sysbench oltp_insert.lua --mysql-host=192.168.1.101 --mysql-port=3306 --mysql-user=zlm --mysql-password=zlmzlm --mysql-db=sysbench --threads=3 --time=1200 --report-interval=30 --rand-type=uniform run 23 sysbench 1.0.15 (using bundled LuaJIT 2.1.0-beta2) 24 25 Running the test with following options: 26 Number of threads: 3 27 Report intermediate results every 30 second(s) 28 Initializing random number generator from current time 29 30 31 Initializing worker threads... 32 33 Threads started! 34 35 [ 30s ] thds: 3 tps: 931.95 qps: 931.95 (r/w/o: 0.00/931.95/0.00) lat (ms,95%): 6.09 err/s: 0.00 reconn/s: 0.00 36 [ 60s ] thds: 3 tps: 971.62 qps: 971.62 (r/w/o: 0.00/971.62/0.00) lat (ms,95%): 5.18 err/s: 0.00 reconn/s: 0.00 37 [ 90s ] thds: 3 tps: 949.33 qps: 949.33 (r/w/o: 0.00/949.33/0.00) lat (ms,95%): 4.41 err/s: 0.00 reconn/s: 0.00
Check the global status of semi-sync on master.
1 (zlm@192.168.1.101 3306)[(none)]>show global status like 'Rpl_semi_sync%'; 2 +--------------------------------------------+------------+ 3 | Variable_name | Value | 4 +--------------------------------------------+------------+ 5 | Rpl_semi_sync_master_clients | 2 | 6 | Rpl_semi_sync_master_net_avg_wait_time | 0 | 7 | Rpl_semi_sync_master_net_wait_time | 0 | 8 | Rpl_semi_sync_master_net_waits | 3949144 | 9 | Rpl_semi_sync_master_no_times | 9 | 10 | Rpl_semi_sync_master_no_tx | 1788275 | 11 | Rpl_semi_sync_master_status | ON | 12 | Rpl_semi_sync_master_timefunc_failures | 0 | 13 | Rpl_semi_sync_master_tx_avg_wait_time | 1859 | 14 | Rpl_semi_sync_master_tx_wait_time | 1354150254 | 15 | Rpl_semi_sync_master_tx_waits | 728148 | 16 | Rpl_semi_sync_master_wait_pos_backtraverse | 0 | 17 | Rpl_semi_sync_master_wait_sessions | 0 | 18 | Rpl_semi_sync_master_yes_tx | 795124 | 19 +--------------------------------------------+------------+ 20 14 rows in set (0.00 sec)
Set the variable "rpl_semi_sync_master_timeout" to 60000.
1 (zlm@192.168.1.101 3306)[(none)]>set @@global.rpl_semi_sync_master_timeout=60000; 2 Query OK, 0 rows affected (0.00 sec) 3 4 (zlm@192.168.1.101 3306)[(none)]>show global variables like 'rpl_semi_sync_master_timeout'; 5 +------------------------------+-------+ 6 | Variable_name | Value | 7 +------------------------------+-------+ 8 | rpl_semi_sync_master_timeout | 60000 | 9 +------------------------------+-------+ 10 1 row in set (0.00 sec)
Stop slave on slave1 and record the execution time.
1 (zlm@192.168.1.102 3306)[(none)]>system date 2 Mon Jul 16 11:20:37 CEST 2018 3 (zlm@192.168.1.102 3306)[(none)]>stop slave; //The stopping time was 11:20:40 around. 4 Query OK, 0 rows affected (0.84 sec)
Check the global status of semi-sync and error log on master.
1 (zlm@192.168.1.101 3306)[(none)]>show global status like 'Rpl_semi_sync%'; 2 +--------------------------------------------+------------+ 3 | Variable_name | Value | 4 +--------------------------------------------+------------+ 5 | Rpl_semi_sync_master_clients | 1 | //The amount of client turned into "1". 6 | Rpl_semi_sync_master_net_avg_wait_time | 0 | 7 | Rpl_semi_sync_master_net_wait_time | 0 | 8 | Rpl_semi_sync_master_net_waits | 3975124 | 9 | Rpl_semi_sync_master_no_times | 9 | 10 | Rpl_semi_sync_master_no_tx | 1788275 | 11 | Rpl_semi_sync_master_status | ON | //Although the amount of client has been less than "slave_count" but it still work in semi-sync mode. 12 | Rpl_semi_sync_master_timefunc_failures | 0 | 13 | Rpl_semi_sync_master_tx_avg_wait_time | 1860 | 14 | Rpl_semi_sync_master_tx_wait_time | 1368916900 | 15 | Rpl_semi_sync_master_tx_waits | 735759 | 16 | Rpl_semi_sync_master_wait_pos_backtraverse | 0 | 17 | Rpl_semi_sync_master_wait_sessions | 1 | 18 | Rpl_semi_sync_master_yes_tx | 803390 | 19 +--------------------------------------------+------------+ 20 14 rows in set (0.00 sec) 21 22 2018-07-16T11:20:39.628398+01:00 0 [ERROR] mysqld: Got an error reading communication packets //It's the time I stopped the slave on slave1. 23 2018-07-16T11:20:40.355498+01:00 54 [ERROR] Semi-sync master failed on net_flush() before waiting for slave reply 24 2018-07-16T11:20:40.355532+01:00 54 [Note] Stop semi-sync binlog_dump to slave (server_id: 1023306) 25 2018-07-16T11:20:41.358003+01:00 54 [Note] Aborted connection 54 to db: 'unconnected' user: 'repl' host: 'zlm3' (Found net error) 26 2018-07-16T11:20:43.367327+01:00 0 [Note] InnoDB: page_cleaner: 1000ms intended loop took 6128ms. The settings might not be optimal. (flushed=700 and evicted=0, during the time.) 27 2018-07-16T11:20:59.683904+01:00 0 [Note] InnoDB: page_cleaner: 1000ms intended loop took 15315ms. The settings might not be optimal. (flushed=2000 and evicted=0, during the time.) 28 2018-07-16T11:21:05.709621+01:00 0 [Note] InnoDB: page_cleaner: 1000ms intended loop took 5025ms. The settings might not be optimal. (flushed=731 and evicted=0, during the time.) 29 2018-07-16T11:21:40.353597+01:00 57 [Warning] Timeout waiting for reply of binlog (file: mysql-bin.000008, pos: 61052572), semi-sync up to file mysql-bin.000008, position 61052134. 30 2018-07-16T11:21:40.353630+01:00 57 [Note] Semi-sync replication switched OFF. //Until the timeout was accumulated to 60s(11:21:40),the semi-sync mode reverts "OFF" what means the master work as asynchronous replication. 31 32 (zlm@192.168.1.101 3306)[(none)]>show global status like 'Rpl_semi_sync%'; 33 +--------------------------------------------+------------+ 34 | Variable_name | Value | 35 +--------------------------------------------+------------+ 36 | Rpl_semi_sync_master_clients | 1 | 37 | Rpl_semi_sync_master_net_avg_wait_time | 0 | 38 | Rpl_semi_sync_master_net_wait_time | 0 | 39 | Rpl_semi_sync_master_net_waits | 4003361 | 40 | Rpl_semi_sync_master_no_times | 10 | 41 | Rpl_semi_sync_master_no_tx | 1826027 | 42 | Rpl_semi_sync_master_status | OFF | //Now the status became "OFF". 43 | Rpl_semi_sync_master_timefunc_failures | 0 | 44 | Rpl_semi_sync_master_tx_avg_wait_time | 1860 | 45 | Rpl_semi_sync_master_tx_wait_time | 1368916900 | 46 | Rpl_semi_sync_master_tx_waits | 735759 | 47 | Rpl_semi_sync_master_wait_pos_backtraverse | 0 | 48 | Rpl_semi_sync_master_wait_sessions | 0 | 49 | Rpl_semi_sync_master_yes_tx | 803390 | 50 +--------------------------------------------+------------+ 51 14 rows in set (0.00 sec)
Change the variable "rpl_semi_sync_master_wait_no_slave" ot off.
1 (zlm@192.168.1.101 3306)[(none)]>set @@global.rpl_semi_sync_master_wait_no_slave=off; 2 Query OK, 0 rows affected (0.00 sec) 3 4 (zlm@192.168.1.101 3306)[(none)]>show global variables like 'rpl_semi%'; 5 +-------------------------------------------+------------+ 6 | Variable_name | Value | 7 +-------------------------------------------+------------+ 8 | rpl_semi_sync_master_enabled | ON | 9 | rpl_semi_sync_master_timeout | 60000 | 10 | rpl_semi_sync_master_trace_level | 32 | 11 | rpl_semi_sync_master_wait_for_slave_count | 2 | 12 | rpl_semi_sync_master_wait_no_slave | OFF | 13 | rpl_semi_sync_master_wait_point | AFTER_SYNC | 14 +-------------------------------------------+------------+ 15 6 rows in set (0.00 sec)
Start slave on slave1 to restore the initial status of semi-sync replication.
1 //Slave 2 (zlm@192.168.1.102 3306)[(none)]>start slave; 3 Query OK, 0 rows affected (0.01 sec) 4 5 //Error log of master 6 2018-07-16T11:33:22.727473+01:00 60 [Note] Start binlog_dump to master_thread_id(60) slave_server(1023306), pos(, 4) 7 2018-07-16T11:33:31.005135+01:00 0 [Note] InnoDB: page_cleaner: 1000ms intended loop took 8431ms. The settings might not be optimal. (flushed=737 and evicted=0, during the time.) 8 2018-07-16T11:33:31.838932+01:00 60 [Note] Start semi-sync binlog_dump to slave (server_id: 1023306), pos(, 4) 9 2018-07-16T11:33:50.117239+01:00 0 [Note] InnoDB: page_cleaner: 1000ms intended loop took 6667ms. The settings might not be optimal. (flushed=741 and evicted=0, during the time.) 10 2018-07-16T11:33:59.977585+01:00 0 [Note] Semi-sync replication switched ON at (mysql-bin.000008, 206778238)
Stop slave on slave1 again and record the execution time.
1 (zlm@192.168.1.102 3306)[(none)]>system date 2 Mon Jul 16 11:35:55 CEST 2018 3 (zlm@192.168.1.102 3306)[(none)]>stop slave; //The stopping time was 11:35:58 around. 4 Query OK, 0 rows affected (1.60 sec)
Check the global status of semi-sync and error log on master one more time.
1 (zlm@192.168.1.101 3306)[(none)]>show global status like 'Rpl_semi_sync%'; 2 +--------------------------------------------+------------+ 3 | Variable_name | Value | 4 +--------------------------------------------+------------+ 5 | Rpl_semi_sync_master_clients | 2 | 6 | Rpl_semi_sync_master_net_avg_wait_time | 0 | 7 | Rpl_semi_sync_master_net_wait_time | 0 | 8 | Rpl_semi_sync_master_net_waits | 4313674 | 9 | Rpl_semi_sync_master_no_times | 11 | 10 | Rpl_semi_sync_master_no_tx | 2088478 | 11 | Rpl_semi_sync_master_status | ON | 12 | Rpl_semi_sync_master_timefunc_failures | 0 | 13 | Rpl_semi_sync_master_tx_avg_wait_time | 1874 | 14 | Rpl_semi_sync_master_tx_wait_time | 1444213447 | 15 | Rpl_semi_sync_master_tx_waits | 770393 | 16 | Rpl_semi_sync_master_wait_pos_backtraverse | 0 | 17 | Rpl_semi_sync_master_wait_sessions | 1 | 18 | Rpl_semi_sync_master_yes_tx | 840887 | 19 +--------------------------------------------+------------+ 20 14 rows in set (0.00 sec) 21 22 2018-07-16T11:35:58.920988+01:00 0 [ERROR] mysqld: Got an error reading communication packets 23 2018-07-16T11:36:00.321218+01:00 0 [Note] InnoDB: page_cleaner: 1000ms intended loop took 14988ms. The settings might not be optimal. (flushed=2000 and evicted=0, during the time.) 24 2018-07-16T11:36:11.150976+01:00 0 [Note] InnoDB: page_cleaner: 1000ms intended loop took 8718ms. The settings might not be optimal. (flushed=1124 and evicted=0, during the time.) 25 2018-07-16T11:36:28.973672+01:00 60 [Note] Stop semi-sync binlog_dump to slave (server_id: 1023306) 26 2018-07-16T11:36:29.012288+01:00 60 [Note] Semi-sync replication switched OFF. //The master didn't wait the timeout occur and revert to aynchronous replication directly within about merely 30s. 27 28 (zlm@192.168.1.101 3306)[(none)]>show global status like 'Rpl_semi_sync%'; 29 +--------------------------------------------+------------+ 30 | Variable_name | Value | 31 +--------------------------------------------+------------+ 32 | Rpl_semi_sync_master_clients | 1 | 33 | Rpl_semi_sync_master_net_avg_wait_time | 0 | 34 | Rpl_semi_sync_master_net_wait_time | 0 | 35 | Rpl_semi_sync_master_net_waits | 4336577 | 36 | Rpl_semi_sync_master_no_times | 12 | 37 | Rpl_semi_sync_master_no_tx | 2121402 | 38 | Rpl_semi_sync_master_status | OFF | 39 | Rpl_semi_sync_master_timefunc_failures | 0 | 40 | Rpl_semi_sync_master_tx_avg_wait_time | 1913 | 41 | Rpl_semi_sync_master_tx_wait_time | 1474253267 | 42 | Rpl_semi_sync_master_tx_waits | 770394 | 43 | Rpl_semi_sync_master_wait_pos_backtraverse | 0 | 44 | Rpl_semi_sync_master_wait_sessions | 0 | 45 | Rpl_semi_sync_master_yes_tx | 840887 | 46 +--------------------------------------------+------------+ 47 14 rows in set (0.00 sec)
Summay
- The behavior of semi-sync replication depends on the variables that you specified.
- If the value of "rpl_semi_sync_master_wait_no_slave" is "on",the status of semi-sync won't change until it reaches the value of timeout,even though the amount of slaves are less than it in "rpl_semi_sync_master_slave_count".
- If the value of "rpl_semi_sync_master_wait_no_slave" is "off",master won't wait for timeout before it reverts to normal replication when slave decreases to less than the value above.
- In order to observe the result conveniently,I increase the value of "rpl_semi_sync_master_timeout" to 60s.
- In most high consistent scenarios,the value can be tremendous enough to avoid degenerating to asynchronous replication.
- If the degeneration is due to network failure such as instant glitch.The semi-sync status will be ok after the network becomes normal.
- I'm afraid the best practice is to set "rpl_semi_sync_master_wait_no_slave" to "on" and the "rpl_semi_sync_master_timeout" to a comparative value.