我已经将数据库的表从InnoDB转换为TokuDB,并且我注意到使用TokuDB时,读取操作太多CPU。为什么是这样?MySQL TokuDB引擎使用太多的CPU
更具体地说,具有TokuDB表的服务器是具有InnoDB的服务器的从属服务器,其是PXC的一部分。奴隶只使用普通的percona服务器而不是PXC。但奴隶似乎使用太多的CPU,我不知道为什么?
下面是我的my.cnf配置:
[client]
port = 3306
socket = /var/run/mysqld/mysqld.sock
[mysqld_safe]
thp-setting=never
socket = /var/run/mysqld/mysqld.sock
nice = 0
flush_caches
numa_interleave
core-file-size = unlimited
open_files_limit = 1024
[mysqld]
back_log = 65535
bind-address = 0.0.0.0
binlog_format = ROW
character_set_server = utf8
collation_server = utf8_general_ci
core_file
basedir = /usr
datadir = /var/lib/mysql
#default_storage_engine = InnoDB
enforce-gtid-consistency = 1
expand_fast_index_creation = 1
expire_logs_days = 7
gtid_mode = ON
innodb_autoinc_lock_mode = 2
innodb_buffer_pool_instances = 1
innodb_buffer_pool_populate = 1
innodb_buffer_pool_size = 512M
innodb_data_file_path = ibdata1:64M;ibdata2:64M:autoextend
innodb_file_format = Barracuda
innodb_file_per_table
innodb_force_recovery = 1
innodb_flush_log_at_trx_commit = 2
innodb_flush_method = O_DIRECT
innodb_io_capacity = 1600
innodb_large_prefix
innodb_locks_unsafe_for_binlog = 1
innodb_log_file_size = 64M
innodb_print_all_deadlocks = 1
innodb_read_io_threads = 64
innodb_stats_on_metadata = FALSE
innodb_support_xa = FALSE
innodb_write_io_threads = 64
lc-messages-dir = /usr/share/mysql
log-bin = mysqld-bin
log-queries-not-using-indexes
log-slave-updates
long_query_time = 1
master_info_repository = TABLE
max_allowed_packet = 64M
max_connect_errors = 4294967295
max_connections = 2500
max_user_connections = 2550
min_examined_row_limit = 1000
open_files_limit = 1024
port = 3306
relay_log_info_repository = TABLE
relay-log-recovery = TRUE
relay-log-recovery = 1
skip-external-locking
skip-name-resolve
slave_parallel_workers = 8
slow_query_log = 1
slow_query_log_timestamp_always = 1
socket = /var/run/mysqld/mysqld.sock
table_open_cache = 4096
thread_cache = 1024
tmpdir = /srv/tmp
transaction_isolation = REPEATABLE-READ
updatable_views_with_limit = 0
user = mysql
wait_timeout = 60
server-id = 2
# TokuDB fine tuning
default_storage_engine = TokuDB
tokudb_analyze_time = 5
#tokudb_cache_size = 6G
tokudb_directio = 1
tokudb_commit_sync = 0
tokudb_fsync_log_period = 1000
tokudb_load_save_space =1
tokudb_alter_print_error=0
tokudb_block_size = 4MB
tokudb_bulk_fetch = 1
tokudb_disable_slow_alter = 1
tokudb_last_lock_timeout = empty
tokudb_row_format = tokudb_quicklz
#tokudb_data_dir = /var/lib/tokudb
[mysqldump]
quick
quote-names
max_allowed_packet = 16M
[mysql]
#no-auto-rehash # faster start of mysql but no tab completition
[isamchk]
key_buffer = 16M
!includedir /etc/mysql/conf.d/
正在被我们的监控系统xymon报告以下复制邮件时tokudb_cache_size当最初设置为内存总量的80%。
2016-02-25 16:42:04 9604 [Warning] Neither --relay-log nor --relay-log-index were used; so replication may break when this MySQL server acts as a slave and has his hostname changed!! Please use '--relay-log=db-kdb-slave-6-relay-bin' to avoid this problem.
2016-02-25 16:42:05 9604 [Warning] Recovery from master pos 552554502 and file mysqld-bin.001163. Previous relay log pos and relay log file had been set to 552554714, ./db-kdb-slave-6-relay-bin.002933 respectively.
2016-02-25 16:42:05 9604 [Warning] Storing MySQL user name or password information in the master info repository is not secure and is therefore not recommended. Please consider using the USER and PASSWORD connection options for START SLAVE; see the 'START SLAVE Syntax' in the MySQL Manual for more information.
------有关运行InnoDB的主服务器和PXC的一部分-----------更多信息
## Results from top
top - 10:05:12 up 14 days, 7:56, 2 users, load average: 2.16, 2.31, 2.39
Tasks: 413 total, 1 running, 412 sleeping, 0 stopped, 0 zombie
%Cpu(s): 8.9 us, 0.6 sy, 0.0 ni, 89.9 id, 0.3 wa, 0.0 hi, 0.2 si, 0.0 st
KiB Mem: 65704012 total, 63553216 used, 2150796 free, 169832 buffers
KiB Swap: 975868 total, 809892 used, 165976 free. 16304268 cached Mem
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
2485 mysql 20 0 60.146g 0.045t 2.612g S 314.9 73.3 27762:43 mysqld
## disk info
[email protected]:~$ df -h
Filesystem Size Used Avail Use% Mounted on
udev 32G 8.0K 32G 1% /dev
tmpfs 6.3G 1.2M 6.3G 1% /run
/dev/sda2 274G 2.1G 258G 1%/
none 4.0K 0 4.0K 0% /sys/fs/cgroup
none 5.0M 0 5.0M 0% /run/lock
none 32G 0 32G 0% /run/shm
none 100M 0 100M 0% /run/user
/dev/nvme0n1p1 1.1T 542G 503G 52% /srv
na1:/vol/yphome 4.5T 3.7T 875G 82% /net/account
## Memory info
[email protected]:~$ free -g
total used free shared buffers cached
Mem: 62 60 2 0 0 15
-/+ buffers/cache: 44 17
Swap: 0 0 0
[email protected]:~$
## Database info
+--------------------+----------------------+
| Data Base Name | Data Base Size in MB |
+--------------------+----------------------+
| information_schema | 0.00976563 |
| dberp | 347143.32031250 |
| mysql | 2.11562061 |
| performance_schema | 0.00000000 |
+--------------------+----------------------+
4 rows in set (0.13 sec)
+--------------------+----------------------+------------------+
| Data Base Name | Data Base Size in MB | Free Space in MB |
+--------------------+----------------------+------------------+
| information_schema | 0.00976563 | 0.00000000 |
| dberp | 347143.32031250 | 6270.00000000 |
| mysql | 2.11562061 | 4.00199127 |
| performance_schema | 0.00000000 | 0.00000000 |
+--------------------+----------------------+------------------+
4 rows in set (0.03 sec)
@mcallaghan,谢谢你的回应。我对TokuDB的理解是缓存中的数据是未压缩的,并且在磁盘中是压缩的。我的理解是,使用TokuDB时,数据大小是否不适合RAM,并不重要,因为这不会降低性能。所以如果我们有缓存的数据(包含未压缩的数据),为什么TokuDB需要解压数据? tokudb_cache_size最初设置为80%,我们仍然有很高的CPU使用率,再加上一些复制问题。当从服务器有复制问题时,我添加了发出的通知(检查原始文章)。 –
您正在混淆了很多TokuDB概念。底线,压缩和解压缩以CPU为代价节省IO,因此使用TokuDB时CPU总是更高。如果你想看到它更多的苹果到苹果在您的主表上启用InnoDB压缩。要获得更具体的帮助,您需要更详细地解释您的具体工作量。 – tmcallaghan
我在原始文章中提供了有关主服务器的更多信息。主服务器上的所有表都使用压缩= COMPRESSED的InnoDB表。所有的表都有charset ='utf8mb4',并整理了utf8mb4_general_ci。 –