一、描述
MySQL 5.7版本主從複製，批量時候顯示延遲上萬秒。
二、現象
1、io使用率高
#iostat -dxm 1 1000
Device: rrqm/s wrqm/s r/s w/s rMB/s wMB/s avgrq-sz avgqu-sz await svctm %util
scd0 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
vda 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
vdb 0.00 96.00 0.00 2596.00 0.00 8.54 6.74 1.33 0.51 0.37 95.30
vdc 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
vdd 0.00 0.00 0.00 11.00 0.00 0.06 11.64 0.00 0.09 0.09 0.10
vde 0.00 0.00 0.00 7.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
vdf 0.00 0.00 0.00 511.00 0.00 0.00 0.00 0.05 0.09 0.09 4.60
vdg 0.00 0.00 0.00 511.00 0.00 0.00 0.00 0.05 0.09 0.09 4.80
dm-0 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
dm-1 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
dm-2 0.00 0.00 0.00 34.00 0.00 0.23 13.65 0.02 0.59 0.38 1.30
dm-3 0.00 0.00 0.00 2144.00 0.00 8.38 8.00 1.40 0.65 0.45 97.20
dm-4 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
dm-5 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
2、dm3是relay log 和binlog分區
$ ls -l /dev/mapper
total 0
lrwxrwxrwx 1 root root 7 Jul 23 23:20 backup-backup -> ../dm-0
crw-rw---- 1 root root 10, 58 Jul 23 23:20 control
lrwxrwxrwx 1 root root 7 Jul 23 23:20 VG00-lv_root -> ../dm-4
lrwxrwxrwx 1 root root 7 Jul 23 23:20 zxmysql-zxdba -> ../dm-1
lrwxrwxrwx 1 root root 7 Jul 23 23:20 zxmysql-zxlog -> ../dm-3
3、slave狀態
mysql> show slave status \G;
*************************** 1. row ***************************
Slave_IO_State: Queueing master event to the relay log
略.........................................
Connect_Retry: 60
Master_Log_File: mysql-bin.011494
Read_Master_Log_Pos: 21037034
Relay_Log_File: relay-log.001904
Relay_Log_Pos: 3154097
Relay_Master_Log_File: mysql-bin.011494
Slave_IO_Running: Yes
Slave_SQL_Running: Yes
Replicate_Do_DB:
Replicate_Ignore_DB:
Replicate_Do_Table:
Replicate_Ignore_Table:
Replicate_Wild_Do_Table:
Replicate_Wild_Ignore_Table:
Last_Errno: 0
Last_Error:
Skip_Counter: 0
Exec_Master_Log_Pos: 3153884
Relay_Log_Space: 21037535
Until_Condition: None
Until_Log_File:
Until_Log_Pos: 0
Master_SSL_Allowed: No
Master_SSL_CA_File:
Master_SSL_CA_Path:
Master_SSL_Cert:
Master_SSL_Cipher:
Master_SSL_Key:
Seconds_Behind_Master: 471
Master_SSL_Verify_Server_Cert: No
Last_IO_Errno: 0
Last_IO_Error:
Last_SQL_Errno: 0
Last_SQL_Error:
Replicate_Ignore_Server_Ids:
Master_Server_Id: 400011
Master_UUID: 0f8507ea-6da1-11e8-8646-005056873c4a
Master_Info_File: mysql.slave_master_info
SQL_Delay: 0
SQL_Remaining_Delay: NULL
Slave_SQL_Running_State: Reading event from the relay log
Master_Retry_Count: 86400
Master_Bind:
Last_IO_Error_Timestamp:
Last_SQL_Error_Timestamp:
Master_SSL_Crl:
Master_SSL_Crlpath:
Retrieved_Gtid_Set: 0f8507ea-6da1-11e8-8646-005056873c4a:14137114-19288497
Executed_Gtid_Set: 0f8507ea-6da1-11e8-8646-005056873c4a:1-19288446
Auto_Position: 1
Replicate_Rewrite_DB:
Channel_Name:
Master_TLS_Version:
1 row in set (0.01 sec)
ERROR:
No query specified
三、分析
通過以上現象發現備庫io使用率過高，超過90%。io過高的磁盤爲日誌盤，存放relay log和binlog。io thead一致在寫relay log，調用fdatasync寫磁盤。這裏涉及到一個參數sync_relay_log，默認值爲10000，查看當前系統參數值爲1.
四、解決方案
優化io thread線程和sql thread線程。sync_relay_log使用默認值，使用mts優化sql thread。
stop slave;
set global slave_parallel_type=logical_clock;
set global slave_parallel_workers=8;
set global sync_master_info=10000;
set global sync_relay_log=10000;
set global sync_relay_log_info=10000;
start slave;