我有我的服务器上真正简单的INSERT/UPDATES问题。不时需要比几秒钟更长时间才能完成这样的查询的:更新/插入时不时需要几秒钟
2.1062s - INSERT INTO `transaction` SET `idUser` = 72, `currency` = 50, `amount` = '10', `action` = 'buyCoins';
11.785s - UPDATE `user` SET `cash` = 10, `crystal` = 10, `expPoints` = 10, `energy` = 10 WHERE idUser = 72;
0.6296s - UPDATE `user` SET `lastEnergyUpdate` = CURRENT_TIMESTAMP WHERE idUser = 72;
看起来问题不依赖于特定的表。我没有那些桌子上的TRIGGERS。
表定义:
CREATE TABLE `user` (
`idUser` int(10) unsigned NOT NULL AUTO_INCREMENT,
`expPoints` int(10) NOT NULL DEFAULT '0',
`cash` int(10) NOT NULL DEFAULT '1000',
`crystal` int(10) NOT NULL DEFAULT '10',
`energy` int(4) NOT NULL DEFAULT '0',
`name` varchar(50) DEFAULT NULL,
`surname` varchar(50) DEFAULT NULL,
`age` int(4) unsigned DEFAULT NULL,
`sex` enum('men','women','unknown') DEFAULT NULL,
`lastEnergyUpdate` timestamp NULL DEFAULT NULL,
`lastLogin` timestamp NOT NULL DEFAULT '0000-00-00 00:00:00',
`insertDate` timestamp NOT NULL DEFAULT CURRENT_TIMESTAMP,
PRIMARY KEY (`idUser`),
UNIQUE KEY `serviceUnique` (`serviceName`,`serviceId`)
) ENGINE=InnoDB AUTO_INCREMENT=5333 DEFAULT CHARSET=utf8
CREATE TABLE `transaction` (
`idTransaction` int(10) NOT NULL AUTO_INCREMENT,
`idUser` int(10) unsigned NOT NULL,
`currency` enum('crystal','partnerCurrency','cash') DEFAULT NULL,
`amount` int(5) NOT NULL,
`action` enum('unlockPlace','buyExtra','collectReleased') NOT NULL,
`insertDate` timestamp NOT NULL DEFAULT CURRENT_TIMESTAMP,
PRIMARY KEY (`idTransaction`),
KEY `fk_transaction_user1` (`idUser`),
CONSTRAINT `fk_transaction_user1` FOREIGN KEY (`idUser`) REFERENCES `user` (`idUser`) ON DELETE CASCADE ON UPDATE CASCADE
) ENGINE=InnoDB AUTO_INCREMENT=156329 DEFAULT CHARSET=utf8
在同一台服务器我有多个数据库(〜100),但没有大的。所有数据库的转储大约为300MB。
Mysqltunner输出:
>> MySQLTuner 1.0.1 - Major Hayden <[email protected]>
>> Bug reports, feature requests, and downloads at http://mysqltuner.com/
>> Run with '--help' for additional options and output filtering
-------- General Statistics --------------------------------------------------
[--] Skipped version check for MySQLTuner script
[OK] Currently running supported MySQL version 5.1.66-0ubuntu0.11.10.2-log
[OK] Operating on 64-bit architecture
-------- Storage Engine Statistics -------------------------------------------
[--] Status: -Archive -BDB -Federated +InnoDB -ISAM -NDBCluster
[--] Data in MyISAM tables: 138M (Tables: 267)
[--] Data in InnoDB tables: 170M (Tables: 327)
[--] Data in MEMORY tables: 0B (Tables: 1)
[!!] Total fragmented tables: 329
-------- Performance Metrics -------------------------------------------------
[--] Up for: 20h 45m 57s (558K q [7.468 qps], 58K conn, TX: 685M, RX: 98M)
[--] Reads/Writes: 66%/34%
[--] Total buffers: 1.1G global + 6.0M per thread (150 max threads)
[OK] Maximum possible memory usage: 2.0G (12% of installed RAM)
[OK] Slow queries: 0% (54/558K)
[OK] Highest usage of available connections: 6% (10/150)
[OK] Key buffer size/total MyISAM indexes: 16.0M/8.8M
[OK] Key buffer hit rate: 99.9% (245K cached/258 reads)
[OK] Query cache efficiency: 51.5% (176K cached/342K selects)
[OK] Query cache prunes per day: 0
[OK] Sorts requiring temporary tables: 6% (1K temp sorts/19K sorts)
[!!] Temporary tables created on disk: 34% (2K on disk/8K total)
[OK] Thread cache hit rate: 99% (10 created/58K connections)
[!!] Table cache hit rate: 16% (786 open/4K opened)
[OK] Open file limit used: 32% (714/2K)
[OK] Table locks acquired immediately: 99% (329K immediate/329K locks)
[OK] InnoDB data size/buffer pool: 170.3M/512.0M
-------- Recommendations -----------------------------------------------------
General recommendations:
Run OPTIMIZE TABLE to defragment tables for better performance
MySQL started within last 24 hours - recommendations may be inaccurate
Temporary table size is already large - reduce result set size
Reduce your SELECT DISTINCT queries without LIMIT clauses
Increase table_cache gradually to avoid file descriptor limits
Variables to adjust:
table_cache (> 1024)
当然这只是为了那个〜1%的情况查询(99%正常工作),然后HDD真的布西(13% - 8个核服务器上的20%WA )
我应该不断增加table_cache吗?任何其他想法发生了什么?我该如何改进它?
我的MySQL服务器是5.1.66。我试图升级到5.5.x,但这并没有帮助我,所以我将它降级。
看看慢查询日志,如果有一个查询是锁定表,并因此阻止你的upsert。 – fancyPants
您的表缓存命中率很低(很好的实际预期是95%以上),缓存大小应该会增加。创建的临时磁盘表的数量很高,但这可能只是因为您正在运行的查询;有些将始终需要临时磁盘表。使用主键顺序插入/更新数据时,InnoDB可能会很烦人。表中有多少行?外键可能是问题的一部分,因为它需要在插入/更新时进行检查。这两个更新查询是不是有一些原因? –
@tambom - 我检查了慢查询日志,但我发现只有那些更新+来自不同数据库的一些查询(但现在我只关心这个数据库) – Skowron