Last Update: 2017-02-10 07:35:34 -0800

Connecting to a database

All Sequel activity begins with connecting to a database, which creates a Sequel::Database object. The Database object is used to create datasets and execute queries. Sequel provides a powerful and flexible mechanism for connecting to databases. There are two main ways to establish database connections:

  1. Using the Sequel.connect method

  2. Using the specialized adapter method (Sequel.sqlite, Sequel.postgres, etc.)

The connection options needed depend on the adapter being used, though most adapters share the same basic connection options.

If you are only connecting to a single database, it is recommended that you store the database object in a constant named DB. This is not required, but it is the convention that most Sequel code uses.

Using the Sequel.connect method

The connect method usually takes a well-formed URI, which is parsed into connection options needed to open the database connection. The scheme/protocol part of the URI is used to determine the adapter to use:

DB = Sequel.connect('postgres://user:password@localhost/blog') # Uses the postgres adapter

You can use URI query parameters to specify options:

DB = Sequel.connect('postgres://localhost/blog?user=user&password=password')

You can also pass an additional option hash with the connection string:

DB = Sequel.connect('postgres://localhost/blog', :user=>'user', :password=>'password')

You can also just use an options hash without a connection string. If you do this, you must provide the adapter to use:

DB = Sequel.connect(:adapter=>'postgres', :host=>'localhost', :database=>'blog', :user=>'user', :password=>'password')

All of the above statements are equivalent.

Using the specialized adapter method

The specialized adapter method is similar to Sequel.connect with an options hash, except that it automatically populates the :adapter option and assumes the first argument is the :database option, unless the first argument is a hash. So the following statements are equivalent to the previous statements.

DB = Sequel.postgres('blog', :host=>'localhost', :user=>'user', :password=>'password')
DB = Sequel.postgres(:host=>'localhost', :user=>'user', :password=>'password', :database=>'blog')

Note that using an adapter method forces the use of the specified adapter, not a database type, even though some adapters have the same name as the database type. So if you want to connect to SQLite, for example, you can do so using the sqlite, do, jdbc, and swift adapters. If you want to connect to SQLite on JRuby using the jdbc adapter, you should not use Sequel.sqlite for example, as that uses the C-based sqlite3 gem. Instead, the Sequel.jdbc would be appropriate (though as mentioned below, using Sequel.connect is recommended instead of Sequel.jdbc).

Passing a block to either method

Both the Sequel.connect method and the specialized adapter methods take a block. If you provide a block to the method, Sequel will create a Database object and pass it as an argument to the block. When the block returns, Sequel will disconnect the database connection. For example:

Sequel.connect('sqlite://blog.db'){|db| puts db[:users].count}

Note that if you do not pass a block to Sequel.connect, Sequel will automatically retain a reference to the object in the Sequel::DATABASES array. So calling Sequel.connect multiple times (say once per request), can result in a memory leak. For any application where database access is needed for a long period of time, it's best to store the result of Sequel.connection in a constant, as recommended above.

Using the Sequel.connect method

General connection options

These options are shared by all adapters unless otherwise noted.


The adapter to use


The name of the database to which to connect


The hostname of the database server to which to connect


An array of SQL loggers to log to


The password for the user account


A hash with symbol keys and hash or proc values, used with master/slave/partitioned database configurations


Whether to use a single-threaded (non-thread safe) connection pool


Whether to test that a valid database connection can be made (false by default)


The user account name to use logging in

The following options can be specified and are passed to the database's internal connection pool.


A callable object called after each new connection is made, with the connection object (and server argument if the callable accepts 2 arguments), useful for customizations that you want to apply to all connections (default: nil).


The maximum size of the connection pool (default: 4 connections on most databases)


The number of seconds to sleep before trying to acquire a connection again (default: 0.001 seconds)


The number of seconds to wait if a connection cannot be acquired before raising an error (default: 5 seconds)

Adapter specific connection options

The following sections explain the options and behavior specific to each adapter. If the library the adapter requires is different from the name of the adapter scheme, it is listed specifically, otherwise you can assume that is requires the library with the same name.


Requires: win32ole

The ADO adapter provides connectivity to ADO databases in Windows. It relies on WIN32OLE library, so it isn't usable on other operating systems (except possibly through WINE, but that's unlikely).

The following options are supported:


Sets the time in seconds to wait while attempting to execute a command before cancelling the attempt and generating an error. Specifically, it sets the ADO CommandTimeout property. If this property is not set, the default of 30 seconds is used.


The driver to use in the ADO connection string. If not provided, a default of “SQL Server” is used.


The full ADO connection string. If this is provided, the usual options are ignored.


Sets the Provider of this ADO connection (for example, “SQLOLEDB”). If you don't specify a provider, the default one used by WIN32OLE has major problems, such as creating a new native database connection for every query, which breaks things such as transactions and temporary tables.

Pay special attention to the :provider option, as without specifying a provider, many things will be broken. The SQLNCLI10 and SQLNCLI11 providers work well if you are connecting to Microsoft SQL Server, but it is not the default as it depends on those providers being installed.

Example connections:

# SQL Server
# Access 2007
Sequel.ado(:conn_string=>'Provider=Microsoft.ACE.OLEDB.12.0;Data Source=drive:\path\filename.accdb')
# Access 2000
Sequel.ado(:conn_string=>'Provider=Microsoft.Jet.OLEDB.4.0;Data Source=drive:\path\filename.mdb')
# Excel 2000 (for table names, use a dollar after the sheet name, e.g. Sheet1$)
Sequel.ado(:conn_string=>'Provider=Microsoft.Jet.OLEDB.4.0;Data Source=drive:\path\filename.xls;Extended Properties=Excel 8.0;')


Amalgalite is an ruby extension that provides self contained access to SQLite, so you don't need to install SQLite separately. As amalgalite is a file backed database, the :host, :user, and :password options are not used.


The name of the database file


The busy timeout period given in milliseconds

Without a database argument, assumes a memory database, so you can do:


Handles paths in the connection string similar to the SQLite adapter, so see the sqlite section below for details.


cubrid is a ruby extension for accessing a CUBRID database. Currently, the ruby cubrid gem is in fairly rough state, with broken transaction support and some other issues, but most things work.


Requires: data_objects

The DataObjects adapter supports PostgreSQL, MySQL, and SQLite. One possible advantage of using DataObjects is that it does the typecasting in C, which may be faster than the other adapters.

Similar to the JDBC adapter, the DO adapter only cares about connection strings, which can either be the String argument given to Sequel.connect directly or contained in a :uri or :url option. The DO adapter passes through the connection string directly to DataObjects, it does no processing of it (other than removing the do: prefix).

Connection string examples:



requires 'ibm_db'

This connects to DB2 using IBM_DB. This is the recommended adapter if you are using a C-based ruby to connect to DB2.


Requires: java

Houses Sequel's JDBC support when running on JRuby. Support for individual database types is done using subadapters. There are currently subadapters for PostgreSQL, MySQL, SQLite, H2, HSQLDB, Derby, Oracle, MSSQL, JTDS, AS400, Progress, Firebird, Informix, and DB2. For PostgreSQL, MySQL, SQLite, H2, HSQLDB, Derby, AS400 and JTDS, this can use the `jdbc-*` gem, for the others you need to have the `.jar` in your CLASSPATH or load the Java class manually before calling Sequel.connect.

You just use the JDBC connection string directly, which can be specified via the string given to Sequel.connect or via the :uri, :url, or :database options. Sequel does no preprocessing of the string, it passes it directly to JDBC. So if you have problems getting a connection string to work, look up the JDBC documentation.

Note that when using a JDBC adapter, the best way to use Sequel is via Sequel.connect, NOT Sequel.jdbc. Use the JDBC connection string when connecting, which will be in a different format than the native connection string. The connection string should start with 'jdbc:'. For PostgreSQL, use 'jdbc:postgresql:', and for SQLite you do not need 2 preceding slashes for the database name (use no preceding slashes for a relative path, and one preceding slash for an absolute path).

Example connection strings:


You can also use JNDI connection strings:


The following additional options are supported:


If set to false, does not attempt to convert some Java types to ruby types. Setting to false roughly doubles performance when selecting large numbers of rows. Note that you can't provide this option inside the connection string (as that is passed directly to JDBC), you have to pass it as a separate option.


Specify the Java driver class to use to connect to the database. This only has an effect if the database type is not recognized from the connection string, and only helps cases where java.sql.DriverManager.getConnection does not return a connection.


Set the login timeout on the JDBC connection (in seconds).

There are a few issues with specific jdbc drivers:


jdbc-h2 versions greater than 1.3.175 have issues with ORDER BY not working correctly in some cases.


jdbc-postgres 9.4.1204 has issues using PostgreSQL-specific types in bound variables. Use jdbc-postgres 9.4.1200 or below.


Requires: mysqlplus (or mysql if mysqlplus is not available)

The MySQL adapter does not support the pure-ruby MySQL adapter that used to ship with ActiveRecord, it requires the native adapter.

The following additional options are supported:


If set to true, makes “WHERE primary_key IS NULL” select the last inserted id.


Same as :encoding, :encoding takes precedence.


Whether to compress data sent/received via the socket connection.


The default group to read from the in the MySQL config file.


If provided, sets the Mysql::OPT_LOCAL_INFILE option on the connection with the given value.


Specify the encoding/character set to use for the connection.


On MySQL 5.6.5+, this option is recognized and will include fractional seconds in time/timestamp values, as well as have the schema method create columns that can contain fractional seconds by deafult. This option is also supported on other adapters that connect to MySQL.


Can be used to specify a Unix socket file to connect to instead of a TCP host and port.


Set the sql_mode(s) for a given connection. Can be single symbol or string, or an array of symbols or strings (e.g. :sql_mode=>[:no_zero_date, :pipes_as_concat]).


Sets the wait_timeout for the connection, defaults to 1 month.


Set the timeout in seconds for reading back results to a query.


Set the timeout in seconds before a connection attempt is abandoned.


This is a newer MySQL adapter that does typecasting in C, so it is often faster than the mysql adapter. The options given are passed to, see the mysql2 documentation for details on what options are supported.


The ODBC adapter allows you to connect to any database with the appropriate ODBC drivers installed.

The :database option given ODBC database should be the DSN (Descriptive Service Name) from the ODBC configuration.

Sequel.odbc('mydb', :user => "user", :password => "password")

The :host and :port options are not respected. The following additional options are supported:


Can be specified as 'mssql', 'progress', or 'db2' to use SQL syntax specific to those databases.


Can be given an ODBC connection string, and will use ODBC::Database#drvconnect to do the connection. Typical usage would be: Sequel.odbc(:drvconnect=>'driver={...};...')


Requires: oci8

The following additional options are supported:


Set to true to use Sequel's conventions to guess the sequence to use for datasets. False by default.


The number of rows to prefetch. Defaults to 100, a larger number can be specified and may improve performance when retrieving a large number of rows.


The Oracle privilege level.


Requires: pg (or postgres if pg is not available)

The Sequel postgres adapter works with the pg, postgres, and postgres-pr ruby libraries. The pg library is the best supported, as it supports real bound variables and prepared statements. If the pg library is being used, Sequel will also attempt to load the sequel_pg library, which is a C extension that optimizes performance when Sequel is used with pg. All users of Sequel who use pg are encouraged to install sequel_pg. For users who want to use postgres-pr to avoid issues with C extensions, it is recommended to use jeremyevans-postgres-pr, which fixes many issues in the upstream postgres-pr gem, and is regularly tested with Sequel.

The following additional options are supported:


Same as :encoding, :encoding takes precedence


Whether infinite timestamps/dates should be converted on retrieval. By default, no conversion is done, so an error is raised if you attempt to retrieve an infinite timestamp/date. You can set this to :nil to convert to nil, :string to leave as a string, or :float to convert to an infinite float.


Set the number of seconds to wait for a connection (default 20, only respected if using the pg library).


A hash of options to pass to the underlying driver (only respected if using the pg library)


Set the client_encoding to the given string


A proc that be called with the PGresult objects that have notice or warning messages. The default notice receiver just prints the messages to stderr, but this can be used to handle notice/warning messages differently. Only respected if using the pg library).


Set to 'disable', 'allow', 'prefer', 'require' to choose how to treat SSL (only respected if using the pg library)


Set to the schema search_path. This can either be a single string containing the schemas separated by commas (for use via a URL: postgres:///?search_path=schema1,schema2), or it can be an array of strings (for use via an option: Sequel.postgres(:search_path=>['schema1', 'schema2'])).


This can be set to false to not force the ISO date format. Sequel forces it by default to allow for an optimization.


The sqlanywhere driver works off connection strings, so a connection string is built based on the url/options hash provided. The following additional options are respected:


specify the CommLinks connection string option


specify the connection string to use, ignoring all other options


specify the ConnectionName connection string option


specify the CharSet connection string option


Requires: sqlite3

As SQLite is a file-based database, the :host and :port options are ignored, and the :database option should be a path to the file.


# In Memory databases:

# Relative Path

# Absolute Path

The following additional options are supported:


open database in read-only mode


the busy timeout to use in milliseconds (default: 5000).

Note that SQLite memory databases are restricted to a single connection by default. This is because SQLite does not allow multiple connections to a single memory database. For this reason, Sequel sets the maximum number of connections in the connection pool to 1 by default when an SQLite memory database is used. Attempts to force the use of more than 1 connection can result in weird behavior, since the connections will be to separate memory databases.


swift is a ruby 1.9+ library, so you'll need to be running ruby 1.9+. It can connect to SQLite, MySQL, and PostgreSQL, and you must specify which database using the db_type option.

You need to install one of the swift db adapters

  • swift-db-sqlite3

  • swift-db-mysql

  • swift-db-postgres




Requires: tiny_tds

The connection options are passed directly to tiny_tds, except that the tiny_tds :username option is set to the Sequel :user option. If you want to use an entry in the freetds.conf file, you should specify the :dataserver option with that name as the value. Some other options that you may want to set are :login_timeout, :timeout, :tds_version, :azure, :appname, and :encoding, see the tiny_tds README for details.

Other Sequel specific options:


Override the server version to use (9000000 = SQL Server 2005). This also works on any other adapter that connects to Microsoft SQL Server.


Override the default TEXTSIZE setting for this connection. The FreeTDS default is small (around 64000 bytes), but can be set up to around 2GB. This should be specified as an integer. If you plan on setting large text or blob values via tinytds, you should use this option or modify your freetds.conf file.

The Sequel tinytds adapter requires tiny_tds >= 0.4.5, and if you are using FreeTDS 0.91, you must at least be using 0.91rc2 (0.91rc1 does not work).