How to fine-tune our MySQL server?

Question

MySQL is not my thing, yet I need to fine-tune one of our servers.

Here are the requirements/specs:

The MySQL server has only one significant database
We only have one "type" of application connected to it, and not many instances at the same time are connected to it : at most 15. (these applications are XMPP bots)
These application have a non-blocking IO, which means that they never "wait" on the DB server and continue dealing with incoming requests while the DB queries are being processed. It implies that sometime one instance of this application can have several (a lot!) connections to the database server (specially if some queries are slow)
- All the queries are using indices
- Our host machine only runs MySQL. It's a Xen instance (@slicehost) with 2GB of RAM.
- We use InnoDB table because we need some basic transactions, but we could probably switch to MyISAM if this had a real impact on performance.

As it is configured right now, our MySQL server slowly starts to eat the all the available memory (we use collectd, here is a graph). At some point (after a few days/weeks), it stops performing queries (it stopped this night for 2 hours, and I had to restart the MySQL server : see 2nd image) :

(sorry, new usrs can't post images, and only 1 hyperlink :/)

Weekly : http://i27.tinypic.com/6ticyv.jpg
Today : i31.tinypic.com/ir53yg.png

Here is our current my.cnf

#
# The MySQL database server configuration file.
#
# This will be passed to all mysql clients
# It has been reported that passwords should be enclosed with ticks/quotes
# escpecially if they contain "#" chars...
# Remember to edit /etc/mysql/debian.cnf when changing the socket location.
[client]
port        = 3306
socket      = /var/run/mysqld/mysqld.sock

# Here is entries for some specific programs
# The following values assume you have at least 32M ram

# This was formally known as [safe_mysqld]. Both versions are currently parsed.
[mysqld_safe]
socket      = /var/run/mysqld/mysqld.sock
nice        = 0

[mysqld]
#
# * Basic Settings
#

#
# * IMPORTANT
#   If you make changes to these settings and your system uses apparmor, you may
#   also need to also adjust /etc/apparmor.d/usr.sbin.mysqld.
#

user        = mysql
pid-file            = /var/run/mysqld/mysqld.pid
socket      = /var/run/mysqld/mysqld.sock
port        = 3306
basedir     = /usr
datadir     = /var/lib/mysql
tmpdir      = /tmp
language            = /usr/share/mysql/english
skip-external-locking
#
# Instead of skip-networking the default is now to listen only on
# localhost which is more compatible and is not less secure.
# yann changed this on a friday balbla
#bind-address       = 127.0.0.1
bind-address        = 0.0.0.0
#
# * Fine Tuning
#
key_buffer      = 16M
max_allowed_packet  = 16M
thread_stack        = 128K
thread_cache_size   = 8
# This replaces the startup script and checks MyISAM tables if needed
# the first time they are touched
myisam-recover      = BACKUP
max_connections        = 2000
#table_cache            = 64
#thread_concurrency     = 10
#
# * Query Cache Configuration
#
query_cache_limit       = 1M
query_cache_size        = 16M
#
# * Logging and Replication
#
# Both location gets rotated by the cronjob.
# Be aware that this log type is a performance killer.
# log       = /var/log/mysql/mysql.log
#
# Error logging goes to syslog. This is a Debian improvement :)
#
# Here you can see queries with especially long duration
log_slow_queries    = /var/log/mysql/mysql-slow.log
long_query_time = 3
log-queries-not-using-indexes

#
# The following can be used as easy to replay backup logs or for replication.
# note: if you are setting up a replication slave, see README.Debian about
#       other settings you may need to change.
#server-id      = 1
#log_bin            = /var/log/mysql/mysql-bin.log
expire_logs_days    = 10
max_binlog_size         = 100M
#binlog_do_db       = include_database_name
#binlog_ignore_db   = include_database_name
#
# * BerkeleyDB
#
# Using BerkeleyDB is now discouraged as its support will cease in 5.1.12.
skip-bdb
#
# * InnoDB
#
# InnoDB is enabled by default with a 10MB datafile in /var/lib/mysql/.
# Read the manual for more InnoDB related options. There are many!
# You might want to disable InnoDB to shrink the mysqld process by circa 100MB.
#skip-innodb

# Fine tunig added by JG on 06/03 based on http://www.mysqlperformanceblog.com/2007/11/01/innodb-performance-optimization-basics/
innodb_buffer_pool_size  = 1G
#innodb_log_file_size     = 256M
innodb_log_buffer_size   = 4M
innodb_flush_log_at_trx_commit = 2
innodb_thread_concurrency      = 8
innodb_flush_method            = O_DIRECT
innodb_file_per_table
transaction-isolation          = READ-COMMITTED
innodb_table_locks             = 0         

#
# * Federated
#
# The FEDERATED storage engine is disabled since 5.0.67 by default in the .cnf files
# shipped with MySQL distributions (my-huge.cnf, my-medium.cnf, and so forth).
#
skip-federated
#
# * Security Features
#
# Read the manual, too, if you want chroot!
# chroot = /var/lib/mysql/
#
# For generating SSL certificates I recommend the OpenSSL GUI "tinyca".
#
# ssl-ca=/etc/mysql/cacert.pem
# ssl-cert=/etc/mysql/server-cert.pem
# ssl-key=/etc/mysql/server-key.pem



[mysqldump]
quick
quote-names
max_allowed_packet  = 16M

[mysql]
#no-auto-rehash # faster start of mysql but no tab completition

[isamchk]
key_buffer      = 16M

#
# * NDB Cluster
#
# See /usr/share/doc/mysql-server-*/README.Debian for more information.
#
# The following configuration is read by the NDB Data Nodes (ndbd processes)
# not from the NDB Management Nodes (ndb_mgmd processes).
#
# [MYSQL_CLUSTER]
# ndb-connectstring=127.0.0.1


#
# * IMPORTANT: Additional settings that can override those from this file!
#   The files must end with '.cnf', otherwise they'll be ignored.
#
!includedir /etc/mysql/conf.d/

Here is a dump of slow queries:

$ mysqldumpslow /var/log/mysql/mysql-slow.log

Reading mysql slow query log from /var/log/mysql/mysql-slow.log
Count: 5  Time=3689348814741910528.00s (-1s)  Lock=0.00s (0s)  Rows=0.0 (0), superfeeder[superfeeder]@[172.21.1.158]
  SET insert_id=N;
  INSERT IGNORE INTO `feeds` (`url`) VALUES ('S')

Count: 41  Time=1349761761490942720.00s (-1s)  Lock=0.12s (5s)  Rows=253.0 (10373), superfeeder[superfeeder]@localhost
  SHOW GLOBAL STATUS

Count: 25  Time=737869762948382080.00s (-1s)  Lock=0.00s (0s)  Rows=18.1 (452), superfeeder[superfeeder]@[172.21.1.158]
  SELECT `feeds`.* FROM `feeds` WHERE (`fetch_session_id` = 'S')

Count: 12952  Time=1424239042133230.25s (-1s)  Lock=0.00s (1s)  Rows=0.0 (0), superfeeder[superfeeder]@[172.21.1.158]
  SET insert_id=N;
  INSERT IGNORE INTO `entries` (`chunks`, `time`, `feed_id`, `unique_id`, `link`, `chunk`) VALUES ('S', 'S', N, 'S', 'S', 'S')

Count: 29  Time=656.55s (19040s)  Lock=5.28s (153s)  Rows=0.8 (23), superfeeder[superfeeder]@[172.21.1.175]
  select salt,crypted_password from users where login='S'

Count: 39  Time=505.23s (19704s)  Lock=2.41s (94s)  Rows=0.0 (0), superfeeder[superfeeder]@[172.21.1.184]
  DELETE FROM `feeds` WHERE (url LIKE 'S')

Count: 2275  Time=502.50s (1143184s)  Lock=3.48s (7922s)  Rows=0.0 (0), superfeeder[superfeeder]@[172.21.1.158]
  UPDATE `feeds` SET `next_fetch` = 'S', `fetch_session_id` = 'S' WHERE (`next_fetch` < 'S') LIMIT N

Count: 1  Time=443.00s (443s)  Lock=0.00s (0s)  Rows=0.0 (0), superfeeder[superfeeder]@[172.21.1.184]
  UPDATE `feeds` SET `next_fetch` = 'S' WHERE (`feeds`.`url` IN (NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL))

Count: 14  Time=289.43s (4052s)  Lock=0.71s (10s)  Rows=0.0 (0), superfeeder[superfeeder]@[172.21.1.184]
  UPDATE `feeds` SET `next_fetch` = 'S' WHERE (`feeds`.`url` IN ('S','S'))

Count: 2  Time=256.00s (512s)  Lock=0.00s (0s)  Rows=0.0 (0), superfeeder[superfeeder]@[172.21.1.184]
  UPDATE `feeds` SET `next_fetch` = 'S' WHERE (`feeds`.`url` IN (NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL))

Count: 1  Time=237.00s (237s)  Lock=0.00s (0s)  Rows=0.0 (0), superfeeder[superfeeder]@[172.21.1.184]
  UPDATE `feeds` SET `next_fetch` = 'S' WHERE (`feeds`.`url` IN ('S'))

Count: 24  Time=191.58s (4598s)  Lock=1.12s (27s)  Rows=0.0 (0), superfeeder[superfeeder]@[172.21.1.184]
  UPDATE `feeds` SET `next_fetch` = 'S' WHERE (`feeds`.`id` = 'S')

Count: 5  Time=144.20s (721s)  Lock=0.00s (0s)  Rows=0.0 (0), superfeeder[superfeeder]@[172.21.1.184]
  UPDATE `feeds` SET `next_fetch` = 'S' WHERE (`feeds`.`url` IN (NULL,NULL,NULL))

Count: 1  Time=101.00s (101s)  Lock=1.00s (1s)  Rows=1.0 (1), superfeeder[superfeeder]@[172.21.1.158]
  SELECT * FROM `users` WHERE (`login` = 'S') LIMIT N

Count: 79  Time=35.51s (2805s)  Lock=2.52s (199s)  Rows=0.2 (12), superfeeder[superfeeder]@[172.21.1.184]
  SELECT `feeds`.id FROM `feeds` WHERE (`feeds`.`url` = BINARY 'S' AND `feeds`.id <> N)  LIMIT N

Count: 1  Time=28.00s (28s)  Lock=0.00s (0s)  Rows=0.0 (0), superfeeder[superfeeder]@[172.21.1.184]
  UPDATE `feeds` SET `last_maintenance_at` = 'S', `updated_at` = 'S' WHERE `id` = N

Count: 51  Time=23.51s (1199s)  Lock=0.12s (6s)  Rows=19.2 (981), superfeeder[superfeeder]@2hosts
  SELECT version FROM schema_migrations

Count: 5  Time=20.60s (103s)  Lock=0.00s (0s)  Rows=0.0 (0), superfeeder[superfeeder]@[172.21.1.184]
  BEGIN

Count: 65  Time=15.86s (1031s)  Lock=0.00s (0s)  Rows=0.0 (0), superfeeder[superfeeder]@[172.21.1.158]
  UPDATE `feeds` SET `last_error_message` = 'S', `period` = 'S', `last_sup_update_id` = NULL, `updated_at` = 'S', `modified` = 'S', `fetch_session_id` = 'S', `streamed` = 'S', `last_parse` = 'S', `etag` = 'S', `last_entry_time` = 'S', `min_period` = 'S', `url` = 'S', `id` = 'S', `feed_type` = NULL, `sup_id` = NULL, `sup_url_id` = NULL, `next_fetch` = 'S', `hashed_content` = 'S', `last_maintenance_at` = 'S', `last_ping` = NULL, `last_http_code` = 'S', `active` = 'S', `last_fetch` = 'S', `created_at` = 'S', `max_period` = 'S' WHERE (`id` = N)

Count: 23  Time=11.52s (265s)  Lock=0.00s (0s)  Rows=231.0 (5313), superfeeder[superfeeder]@2hosts
  #

Count: 132  Time=10.53s (1390s)  Lock=0.02s (2s)  Rows=0.0 (0), superfeeder[superfeeder]@[172.21.1.158]
  UPDATE `feeds` SET `last_error_message` = 'S', `period` = 'S', `last_sup_update_id` = NULL, `updated_at` = 'S', `modified` = 'S', `fetch_session_id` = 'S', `streamed` = 'S', `last_parse` = 'S', `etag` = 'S', `last_entry_time` = 'S', `min_period` = 'S', `url` = 'S', `id` = 'S', `feed_type` = NULL, `sup_id` = NULL, `sup_url_id` = NULL, `next_fetch` = 'S', `hashed_content` = 'S', `last_maintenance_at` = 'S', `last_ping` = NULL, `last_http_code` = 'S', `active` = 'S', `last_fetch` = 'S', `created_at` = NULL, `max_period` = 'S' WHERE (`id` = N)

Count: 62  Time=9.81s (608s)  Lock=0.00s (0s)  Rows=0.0 (0), superfeeder[superfeeder]@[172.21.1.184]
  ROLLBACK

Count: 151  Time=8.94s (1350s)  Lock=0.00s (0s)  Rows=0.0 (0), superfeeder[superfeeder]@2hosts
  DELETE FROM `entries` WHERE (`time` < 'S')

Count: 25  Time=8.76s (219s)  Lock=0.00s (0s)  Rows=1.0 (24), superfeeder[superfeeder]@[172.21.1.158]
  SELECT * FROM `feeds` WHERE (`url` = 'S') LIMIT N

Count: 2  Time=8.50s (17s)  Lock=0.00s (0s)  Rows=0.0 (0), superfeeder[superfeeder]@[172.21.1.158]
  set SQL_AUTO_IS_NULL=N

Count: 8802  Time=8.44s (74319s)  Lock=0.00s (0s)  Rows=0.0 (0), superfeeder[superfeeder]@[172.21.1.158]
  INSERT IGNORE INTO `entries` (`chunks`, `time`, `feed_id`, `unique_id`, `link`, `chunk`) VALUES ('S', 'S', N, 'S', 'S', 'S')

Count: 1  Time=8.00s (8s)  Lock=0.00s (0s)  Rows=0.0 (0), superfeeder[superfeeder]@[172.21.1.158]
  INSERT IGNORE INTO `subscriptions` (`user_id`, `feed_id`) VALUES (N, N)

Count: 38  Time=7.92s (301s)  Lock=0.00s (0s)  Rows=1.0 (38), superfeeder[superfeeder]@[172.21.1.184]
  SELECT count(DISTINCT `users`.id) AS count_users_id FROM `users`  INNER JOIN `subscriptions` ON `users`.id = `subscriptions`.user_id    WHERE ((`subscriptions`.feed_id = N))

Count: 9  Time=7.67s (69s)  Lock=0.00s (0s)  Rows=0.0 (0), superfeeder[superfeeder]@[172.21.1.158]
  INSERT IGNORE INTO `feeds` (`url`) VALUES ('S')

Count: 244  Time=7.20s (1756s)  Lock=0.00s (0s)  Rows=0.0 (0), superfeeder[superfeeder]@[172.21.1.158]
  UPDATE `feeds` SET `last_error_message` = 'S', `period` = N, `last_sup_update_id` = NULL, `updated_at` = 'S', `modified` = 'S', `fetch_session_id` = 'S', `streamed` = 'S', `last_parse` = 'S', `etag` = 'S', `last_entry_time` = 'S', `min_period` = 'S', `url` = 'S', `id` = 'S', `feed_type` = NULL, `sup_id` = NULL, `sup_url_id` = NULL, `next_fetch` = 'S', `hashed_content` = 'S', `last_maintenance_at` = 'S', `last_ping` = NULL, `last_http_code` = N, `active` = 'S', `last_fetch` = 'S', `created_at` = 'S', `max_period` = 'S' WHERE (`id` = N)

Count: 336  Time=6.85s (2301s)  Lock=0.00s (0s)  Rows=0.0 (0), superfeeder[superfeeder]@[172.21.1.158]
  UPDATE `feeds` SET `last_error_message` = 'S', `period` = N, `last_sup_update_id` = NULL, `updated_at` = 'S', `modified` = 'S', `fetch_session_id` = 'S', `streamed` = 'S', `last_parse` = 'S', `etag` = 'S', `last_entry_time` = 'S', `min_period` = 'S', `url` = 'S', `id` = 'S', `feed_type` = NULL, `sup_id` = NULL, `sup_url_id` = NULL, `next_fetch` = 'S', `hashed_content` = 'S', `last_maintenance_at` = 'S', `last_ping` = NULL, `last_http_code` = N, `active` = 'S', `last_fetch` = 'S', `created_at` = NULL, `max_period` = 'S' WHERE (`id` = N)

Count: 16  Time=6.38s (102s)  Lock=0.00s (0s)  Rows=0.0 (0), superfeeder[superfeeder]@[172.21.1.158]
  UPDATE `feeds` SET `last_error_message` = 'S', `period` = N, `last_sup_update_id` = NULL, `updated_at` = NULL, `modified` = 'S', `fetch_session_id` = 'S', `streamed` = 'S', `last_parse` = 'S', `etag` = 'S', `last_entry_time` = 'S', `min_period` = 'S', `url` = 'S', `id` = 'S', `feed_type` = NULL, `sup_id` = NULL, `sup_url_id` = NULL, `next_fetch` = 'S', `hashed_content` = 'S', `last_maintenance_at` = 'S', `last_ping` = NULL, `last_http_code` = N, `active` = 'S', `last_fetch` = 'S', `created_at` = NULL, `max_period` = 'S' WHERE (`id` = N)

Count: 122  Time=5.91s (721s)  Lock=0.00s (0s)  Rows=1.0 (119), superfeeder[superfeeder]@[172.21.1.158]
  SELECT DISTINCT `users`.* FROM `users` INNER JOIN `subscriptions` ON (`subscriptions`.`user_id` = `users`.`id`) WHERE (`subscriptions`.`feed_id` = N)

Count: 299  Time=5.78s (1727s)  Lock=0.00s (0s)  Rows=1.0 (299), superfeeder[superfeeder]@[172.21.1.158]
  SELECT * FROM `feeds` WHERE (`id` = 'S')

Count: 21  Time=5.48s (115s)  Lock=0.00s (0s)  Rows=1.0 (21), superfeeder[superfeeder]@[172.21.1.158]
  SELECT * FROM `subscriptions` WHERE ((`user_id` = N) AND (`feed_id` = N)) LIMIT N

Count: 27  Time=5.37s (145s)  Lock=0.00s (0s)  Rows=0.0 (0), superfeeder[superfeeder]@[172.21.1.158]
  UPDATE `feeds` SET `last_error_message` = 'S', `period` = 'S', `last_sup_update_id` = NULL, `updated_at` = NULL, `modified` = 'S', `fetch_session_id` = 'S', `streamed` = 'S', `last_parse` = 'S', `etag` = 'S', `last_entry_time` = 'S', `min_period` = 'S', `url` = 'S', `id` = 'S', `feed_type` = NULL, `sup_id` = NULL, `sup_url_id` = NULL, `next_fetch` = 'S', `hashed_content` = 'S', `last_maintenance_at` = 'S', `last_ping` = NULL, `last_http_code` = 'S', `active` = 'S', `last_fetch` = 'S', `created_at` = NULL, `max_period` = 'S' WHERE (`id` = N)

Count: 9  Time=4.33s (39s)  Lock=0.00s (0s)  Rows=0.0 (0), superfeeder[superfeeder]@[172.21.1.158]
  UPDATE `feeds` SET `last_error_message` = 'S', `period` = 'S', `last_sup_update_id` = NULL, `updated_at` = NULL, `modified` = 'S', `fetch_session_id` = 'S', `streamed` = 'S', `last_parse` = 'S', `etag` = 'S', `last_entry_time` = 'S', `min_period` = 'S', `url` = 'S', `id` = 'S', `feed_type` = NULL, `sup_id` = NULL, `sup_url_id` = NULL, `next_fetch` = 'S', `hashed_content` = 'S', `last_maintenance_at` = 'S', `last_ping` = NULL, `last_http_code` = NULL, `active` = 'S', `last_fetch` = 'S', `created_at` = NULL, `max_period` = 'S' WHERE (`id` = N)

Count: 1  Time=4.00s (4s)  Lock=0.00s (0s)  Rows=1.0 (1), superfeeder[superfeeder]@[172.21.1.175]
  select id from users where login='S'

Count: 1  Time=3.00s (3s)  Lock=0.00s (0s)  Rows=22.0 (22), debian-sys-maint[debian-sys-maint]@localhost
  select concat("S",
  TABLE_SCHEMA, "S", TABLE_NAME, "S") 
  from information_schema.TABLES where ENGINE="S"

Count: 1056  Time=0.11s (111s)  Lock=0.00s (0s)  Rows=126.9 (133998), superfeeder[superfeeder]@[172.21.1.184]
  SELECT * FROM `feeds` WHERE (last_maintenance_at < 'S')

Count: 1049  Time=0.00s (1s)  Lock=0.00s (0s)  Rows=3.1 (3303), superfeeder[superfeeder]@[172.21.1.184]
  SELECT * FROM `users` WHERE (one_week_anniversary_sent = N AND activated_at < 'S')

Count: 21  Time=0.00s (0s)  Lock=0.00s (0s)  Rows=0.0 (0), 0users@0hosts
  administrator command: Ping

Count: 1  Time=0.00s (0s)  Lock=0.00s (0s)  Rows=0.0 (0), debian-sys-maint[debian-sys-maint]@localhost
  select count(*) into @discard from `information_schema`.`COLUMNS`

Count: 8  Time=0.00s (0s)  Lock=0.00s (0s)  Rows=30.0 (240), superfeeder[superfeeder]@[172.21.1.184]
  SELECT DISTINCT `feeds`.* FROM `feeds` INNER JOIN `subscriptions` ON `feeds`.id = `subscriptions`.feed_id WHERE ((`subscriptions`.user_id = N)) AND ((`subscriptions`.user_id = N))  LIMIT N, N

Count: 31  Time=0.00s (0s)  Lock=0.00s (0s)  Rows=1.0 (31), superfeeder[superfeeder]@2hosts
  SELECT count(*) AS count_all FROM `feeds`

Count: 1  Time=0.00s (0s)  Lock=0.00s (0s)  Rows=0.0 (0), debian-sys-maint[debian-sys-maint]@localhost
  select count(*) into @discard from `information_schema`.`TRIGGERS`

Count: 1  Time=0.00s (0s)  Lock=0.00s (0s)  Rows=0.0 (0), debian-sys-maint[debian-sys-maint]@localhost
  select count(*) into @discard from `information_schema`.`VIEWS`

Count: 52  Time=0.00s (0s)  Lock=0.00s (0s)  Rows=0.7 (34), superfeeder[superfeeder]@[172.21.1.184]
  SELECT * FROM `users` WHERE (`users`.`remember_token` = 'S')  LIMIT N

Count: 120  Time=0.00s (0s)  Lock=0.00s (0s)  Rows=1.0 (120), superfeeder[superfeeder]@2hosts
  SELECT * FROM `feeds`  ORDER BY feeds.id DESC LIMIT N

Count: 19  Time=0.00s (0s)  Lock=0.00s (0s)  Rows=15.7 (299), superfeeder[superfeeder]@2hosts
  SELECT count(*) AS count_all, last_http_code AS last_http_code FROM `feeds`  GROUP BY last_http_code

Count: 1  Time=0.00s (0s)  Lock=0.00s (0s)  Rows=0.0 (0), debian-sys-maint[debian-sys-maint]@localhost
  select count(*) into @discard from `information_schema`.`ROUTINES`

Count: 1  Time=0.00s (0s)  Lock=0.00s (0s)  Rows=1.0 (1), debian-sys-maint[debian-sys-maint]@localhost
  SELECT count(*) FROM mysql.user WHERE user='S' and password='S'

Table definition for feeds :

+---------------------+--------------+------+-----+---------------------+----------------+
| Field               | Type         | Null | Key | Default             | Extra          |
+---------------------+--------------+------+-----+---------------------+----------------+
| id                  | int(11)      | NO   | PRI | NULL                | auto_increment | 
| url                 | varchar(255) | YES  | UNI | NULL                |                | 
| last_parse          | datetime     | YES  |     | 2009-08-10 14:51:46 |                | 
| etag                | varchar(255) | YES  |     | etag                |                | 
| modified            | datetime     | YES  |     | 2009-08-10 14:51:46 |                | 
| active              | tinyint(1)   | YES  | MUL | 1                   |                | 
| last_fetch          | datetime     | YES  |     | 2009-08-10 14:51:46 |                | 
| next_fetch          | datetime     | YES  | MUL | 2009-08-10 14:51:46 |                | 
| fetch_session_id    | varchar(255) | YES  | MUL |                     |                | 
| period              | int(11)      | YES  |     | 240                 |                | 
| hashed_content      | varchar(255) | YES  |     |                     |                | 
| streamed            | tinyint(1)   | YES  |     | 0                   |                | 
| sup_id              | varchar(255) | YES  | MUL | NULL                |                | 
| last_sup_update_id  | varchar(255) | YES  |     | NULL                |                | 
| last_entry_time     | datetime     | YES  |     | 2009-08-10 14:51:46 |                | 
| last_ping           | datetime     | YES  |     | NULL                |                | 
| last_http_code      | int(11)      | YES  |     | NULL                |                | 
| last_error_message  | varchar(255) | YES  |     |                     |                | 
| sup_url_id          | int(11)      | YES  | MUL | NULL                |                | 
| created_at          | datetime     | YES  |     | NULL                |                | 
| updated_at          | datetime     | YES  |     | NULL                |                | 
| last_maintenance_at | datetime     | YES  |     | 2008-08-10 21:51:50 |                | 
| min_period          | int(11)      | YES  |     | 60                  |                | 
| max_period          | int(11)      | YES  |     | 900                 |                | 
+---------------------+--------------+------+-----+---------------------+----------------+

+-------+------------+--------------------------------------+--------------+------------------+-----------+-------------+----------+--------+------+------------+---------+
| Table | Non_unique | Key_name                             | Seq_in_index | Column_name      | Collation | Cardinality | Sub_part | Packed | Null | Index_type | Comment |
+-------+------------+--------------------------------------+--------------+------------------+-----------+-------------+----------+--------+------+------------+---------+
| feeds |          0 | PRIMARY                              |            1 | id               | A         |         166 |     NULL | NULL   |      | BTREE      |         | 
| feeds |          0 | index_feeds_on_url                   |            1 | url              | A         |         166 |     NULL | NULL   | YES  | BTREE      |         | 
| feeds |          1 | index_feeds_on_next_fetch_and_active |            1 | next_fetch       | A         |           1 |     NULL | NULL   | YES  | BTREE      |         | 
| feeds |          1 | index_feeds_on_next_fetch_and_active |            2 | active           | A         |           1 |     NULL | NULL   | YES  | BTREE      |         | 
| feeds |          1 | index_feeds_on_sup_id                |            1 | sup_id           | A         |           1 |     NULL | NULL   | YES  | BTREE      |         | 
| feeds |          1 | index_feeds_on_sup_url_id            |            1 | sup_url_id       | A         |           1 |     NULL | NULL   | YES  | BTREE      |         | 
| feeds |          1 | index_feeds_on_fetch_session_id      |            1 | fetch_session_id | A         |           1 |     NULL | NULL   | YES  | BTREE      |         | 
+-------+------------+--------------------------------------+--------------+------------------+-----------+-------------+----------+--------+------+------------+---------+

What does the error log say? – shantanuo Jan 24 '11 at 02:55 — shantanuo, Jan 24 '11 at 02:55

score 12 · Accepted Answer · answered Aug 12 '09 at 16:19

12

You probably shouldn't consider MyISAM, INNODB will work for you. MyISAM is maybe faster when it comes to SELECT but (for example) it locks your full table on updates.

As for INNODB:

generally, always consider more RAM before you go into sharding (size of the DB =~ RAM)
take a look at the following variables:
- innodb_buffer_pool_size (we use roughly 60-70% of our memory)
- innodb_log_file_size
- innodb_log_buffer_size
- innodb_flush_log_at_trx_commit
- innodb_thread_concurrency
- innodb_flush_method=O_DIRECT
- innodb_file_per_table
switch from innodb to xtradb (same API)
use the percona builds (they contain performance patches from Google, etc.)

Great reads:

On a side note:

a 2 GB slice is just not enough to run this
further more I found the storage on slicehost to be rather slow (io is a factor)
in the cloud it may make sense to shard earlier (cause of the RAM limit)
I'd run all queries through EXPLAIN to make sure the index is really being used

answered Aug 12 '09 at 16:19

Till

1,019
6
14

Great response! Right now, our database size is 61M... so if I believe your statement DB =~ RAM; we have wayyy enough RAM! I think storage is indeed slow (because it is virtualized I guess). I'll read your links. Thanks! – Julien Genestoux Aug 12 '09 at 16:29
(and yeah, I have ran all queries through EXPLAIN and they all use indices) – Julien Genestoux Aug 12 '09 at 16:31
Hey Julien -- first off, you vote up answers around here. ;-) Secondly, just having RAM is of no use if MySQL isn't using it. Make sure you double-check the settings so it uses them. – Till Aug 12 '09 at 16:43
Yeah, well that is exactly where I need my help ;) I think MySQL uses it. (and actually too much!, because our server seems to run out of RAM after some point). The DB is size if 61M, how come it "eats" 2GB of RAM? – Julien Genestoux Aug 12 '09 at 18:09
Can you add the table definition for your "worst" query? – Till Aug 12 '09 at 18:33
(e.g. the feeds table) – Till Aug 12 '09 at 18:34
Yeah, I added it in the post. – Julien Genestoux Aug 13 '09 at 06:53

score 2 · Answer 2 · answered Aug 12 '09 at 15:59

2

I've found MySQLTuner to be very effective in the past -- it can make sensible suggestions for configuration changes based on the usage of your server. It's based on the tuning-primer script, which is also worth a try.

answered Aug 12 '09 at 15:59

Dan

737
5
11

score 1 · Answer 3 · answered Aug 12 '09 at 22:46

1

Does your app release the connections to the pool when it finishes its stuff ?

answered Aug 12 '09 at 22:46

Nope, because it's never done with them actually. It's not like a webapp where the app "dies" between connections. – Julien Genestoux Aug 13 '09 at 06:52

score 0 · Answer 4 · answered Aug 12 '09 at 15:43

0

If you can replace the where clauses with specific IDs (assuming there aren't many) then that could speed those updates up.

Using something like sphinx search which you can easily cluster to find items in advance will help. I've heard it can return searches faster than mysql knows it has indices.

http://www.sphinxsearch.com/

answered Aug 12 '09 at 15:43

Mark L

568
2
9
19

Sorry, but I don't see the point? We're not doing any search on our database... – Julien Genestoux Aug 12 '09 at 15:54
You are: .. WHERE (`feeds`.`url` IN (NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL,NULL)) .. .. WHERE (`feeds`.`url` IN ('S','S')) .. Those could be more expensive than finding the integer id's via sphinx first and using sphinx you can push the load off to another server easily with worrying about master-slave setups and bin log delays. – Mark L Aug 12 '09 at 19:16
feeds.url is a index. Is the primary key faster than an index? – Julien Genestoux Aug 12 '09 at 22:10
If it were an integer instead of (I'm assuming) a VARCHAR(10) or something like that then yeah, the integer would use less space and mysql could find it in the index table quicker. – Mark L Aug 13 '09 at 06:29
But wouldn't that be really marginal? – Julien Genestoux Aug 13 '09 at 06:54

score 0 · Answer 5 · answered Aug 12 '09 at 16:00

0

Check again the queries that are taking so much time. Are the tables properly indexed?

Also You can run MySQL Tuner to fine tuning your mysql settings.

answered Aug 12 '09 at 16:00

hdanniel

4,253
22
25

Yeah, the happen when the server blocks, and again, I've checked all queries : they all use indices. ( + the slowest queries are also INSERT queries... which doesn't have anything to do with indices, AFAIK) – Julien Genestoux Aug 12 '09 at 16:15

How to fine-tune our MySQL server?

5 Answers5

Linked