Requires Go 1.17 or above.
Install with go install github.com/microsoft/go-mssqldb@latest
.
The recommended connection string uses a URL format:
sqlserver://username:password@host/instance?param1=value¶m2=value
Other supported formats are listed below.
user id
- enter the SQL Server Authentication user id or the Windows Authentication user id in the DOMAIN\User format. On Windows, if user id is empty or missing Single-Sign-On is used. The user domain sensitive to the case which is defined in the connection string.password
database
connection timeout
- in seconds (default is 0 for no timeout), set to 0 for no timeout. Recommended to set to 0 and use context to manage query and connection timeouts.dial timeout
- in seconds (default is 15 times the number of registered protocols), set to 0 for no timeout.encrypt
strict
- Data sent between client and server is encrypted E2E using TDS8.disable
- Data send between client and server is not encrypted.false
/optional
/no
/0
/f
- Data sent between client and server is not encrypted beyond the login packet. (Default)true
/mandatory
/yes
/1
/t
- Data sent between client and server is encrypted.
app name
- The application name (default is go-mssqldb)authenticator
- Can be used to specify use of a registered authentication provider. (e.g. ntlm, winsspi (on windows) or krb5 (on linux))
server
- host or host\instance (default localhost)port
- specifies the host\instance port (default 1433). If instance name is provided but no port, the driver will use SQL Server Browser to discover the port.
keepAlive
- in seconds; 0 to disable (default is 30)failoverpartner
- host or host\instance (default is no partner).failoverport
- used only when there is no instance in failoverpartner (default 1433)packet size
- in bytes; 512 to 32767 (default is 4096)- Encrypted connections have a maximum packet size of 16383 bytes
- Further information on usage: https://docs.microsoft.com/en-us/sql/database-engine/configure-windows/configure-the-network-packet-size-server-configuration-option
log
- logging flags (default0
/no logging,255
for full logging)1
log errors2
log messages4
log rows affected8
trace sql statements16
log statement parameters32
log transaction begin/end64
additional debug logs128
log retries
TrustServerCertificate
- false - Server certificate is checked. Default is false if encrypt is specified.
- true - Server certificate is not checked. Default is true if encrypt is not specified. If trust server certificate is true, driver accepts any certificate presented by the server and any host name in that certificate. In this mode, TLS is susceptible to man-in-the-middle attacks. This should be used only for testing.
certificate
- The file that contains the public key certificate of the CA that signed the SQL Server certificate. The specified certificate overrides the go platform specific CA certificates. Currently, certificates of PEM type are supported.hostNameInCertificate
- Specifies the Common Name (CN) in the server certificate. Default value is the server host.tlsmin
- Specifies the minimum TLS version for negotiating encryption with the server. Recognized values are1.0
,1.1
,1.2
,1.3
. If not set to a recognized value the default value for thetls
package will be used. The default is currently1.2
.ServerSPN
- The kerberos SPN (Service Principal Name) for the server. Default is MSSQLSvc/host:port.Workstation ID
- The workstation name (default is the host name)ApplicationIntent
- Can be given the valueReadOnly
to initiate a read-only connection to an Availability Group listener. Thedatabase
must be specified when connecting withApplication Intent
set toReadOnly
.protocol
- forces use of a protocol. Make sure the corresponding package is imported.columnencryption
orcolumn encryption setting
- a boolean value indicating whether Always Encrypted should be enabled on the connection.multisubnetfailover
true
(Default) Client attempt to connect to all IPs simultaneously.false
Client attempts to connect to IPs in serial.
pipe
- If set, no Browser query is made and named pipe used will be\\<host>\pipe\<pipe>
protocol
can be set tonp
- For a non-URL DSN, the
server
parameter can be set to the full pipe name like\\host\pipe\sql\query
If no pipe name can be derived from the DSN, connection attempts will first query the SQL Browser service to find the pipe name for the instance.
Custom Dialers can be used to resolve DNS if the Connection's Dialer implements the HostDialer
interface. This is helpful when the dialer is proxying requests to a different, private network and the DNS record is local to the private network.
To force a specific protocol for the connection there two several options:
- Prepend the server name in a DSN with the protocol and a colon, like
np:host
orlpc:host
ortcp:host
- Set the
protocol
parameter to the protocol name
msdsn.ProtocolParsers
can be reordered to prioritize other protocols ahead of tcp
The admin
protocol will not be used for dialing unless the connection string explicitly specifies it. Note SQL Server allows only 1 admin (or DAC) connection to be active at a time.
To connect with kerberos authentication from a Linux server you can use the optional krb5 package. Imported krb alongside the main driver
package main
import (
...
_ "github.com/microsoft/go-mssqldb"
_ "github.com/microsoft/go-mssqldb/integratedauth/krb5"
)
func main() {
...
}
It will become available for use when the connection string parameter "authenticator=krb5" is used.
The package supports authentication via 3 methods.
-
Keytabs - Specify the username, keytab file, the krb5.conf file, and realm.
authenticator=krb5;server=DatabaseServerName;database=DBName;user id=MyUserName;krb5-realm=domain.com;krb5-configfile=/etc/krb5.conf;krb5-keytabfile=~/MyUserName.keytab
-
Credential Cache - Specify the krb5.conf file path and credential cache file path.
authenticator=krb5;server=DatabaseServerName;database=DBName;krb5-configfile=/etc/krb5.conf;krb5-credcachefile=~/MyUserNameCachedCreds
-
Raw credentials - Specity krb5.confg, Username, Password and Realm.
authenticator=krb5;server=DatabaseServerName;database=DBName;user id=MyUserName;password=foo;krb5-realm=comani.com;krb5-configfile=/etc/krb5.conf;
authenticator
- set this tokrb5
to enable kerberos authentication. If this is not present, the default provider would bentlm
for unix andwinsspi
for windows.krb5-configfile
(optional) - path to kerberos configuration file. Defaults to/etc/krb5.conf
. Can also be set usingKRB5_CONFIG
environment variable.krb5-realm
(required with keytab and raw credentials) - Domain name for kerberos authentication. Omit this parameter if the realm is part of the user name likeusername@REALM
.krb5-keytabfile
- path to Keytab file. Can also be set using environment variableKRB5_KTNAME
. If no parameter or environment variable is set, theDefaultClientKeytabName
value from the krb5 config file is used.krb5-credcachefile
- path to Credential cache. Can also be set using environment variableKRBCCNAME
.krb5-dnslookupkdc
- Optional parameter in all contexts. Set to lookup KDCs in DNS. Boolean. Default is true.krb5-udppreferencelimit
- Optional parameter in all contexts. 1 means to always use tcp. MIT krb5 has a default value of 1465, and it prevents user setting more than 32700. Integer. Default is 1.
For further information on usage:
- https://web.mit.edu/kerberos/krb5-1.12/doc/admin/conf_files/krb5_conf.html
- https://web.mit.edu/kerberos/krb5-1.12/doc/basic/index.html
-
URL: with
sqlserver
scheme. username and password appears before the host. Any instance appears as the first segment in the path. All other options are query parameters. Examples:sqlserver://username:password@host/instance?param1=value¶m2=value
sqlserver://username:password@host:port?param1=value¶m2=value
sqlserver://sa@localhost/SQLExpress?database=master&connection+timeout=30
// `SQLExpress instance.sqlserver://sa:mypass@localhost?database=master&connection+timeout=30
// username=sa, password=mypass.sqlserver://sa:mypass@localhost:1234?database=master&connection+timeout=30
// port 1234 on localhost.sqlserver://sa:my%7Bpass@somehost?connection+timeout=30
// password is "my{pass" A string of this format can be constructed using theURL
type in thenet/url
package.
query := url.Values{} query.Add("app name", "MyAppName") u := &url.URL{ Scheme: "sqlserver", User: url.UserPassword(username, password), Host: fmt.Sprintf("%s:%d", hostname, port), // Path: instance, // if connecting to an instance instead of a port RawQuery: query.Encode(), } db, err := sql.Open("sqlserver", u.String())
sqlserver://username@host/instance?krb5-configfile=path/to/file&krb5-credcachefile=/path/to/cache
sqlserver://username@host/instance?krb5-configfile=path/to/file&krb5-realm=domain.com&krb5-keytabfile=/path/to/keytabfile
-
ADO:
key=value
pairs separated by;
. Values may not contain;
, leading and trailing whitespace is ignored. Examples:server=localhost\\SQLExpress;user id=sa;database=master;app name=MyAppName
server=localhost;user id=sa;database=master;app name=MyAppName
server=localhost;user id=sa;database=master;app name=MyAppName;krb5-configfile=path/to/file;krb5-credcachefile=path/to/cache;authenticator=krb5
server=localhost;user id=sa;database=master;app name=MyAppName;krb5-configfile=path/to/file;krb5-realm=domain.com;krb5-keytabfile=path/to/keytabfile;authenticator=krb5
ADO strings support synonyms for database, app name, user id, and server
- server <= addr, address, network address, data source
- user id <= user, uid
- database <= initial catalog
- app name <= application name
-
ODBC: Prefix with
odbc
,key=value
pairs separated by;
. Allow;
by wrapping values in{}
. Examples:odbc:server=localhost\\SQLExpress;user id=sa;database=master;app name=MyAppName
odbc:server=localhost;user id=sa;database=master;app name=MyAppName
odbc:server=localhost;user id=sa;password={foo;bar}
// Value marked with{}
, password is "foo;bar"odbc:server=localhost;user id=sa;password={foo{bar}
// Value marked with{}
, password is "foo{bar"odbc:server=localhost;user id=sa;password={foobar }
// Value marked with{}
, password is "foobar "odbc:server=localhost;user id=sa;password=foo{bar
// Literal{
, password is "foo{bar"odbc:server=localhost;user id=sa;password=foo}bar
// Literal}
, password is "foo}bar"odbc:server=localhost;user id=sa;password={foo{bar}
// Literal{
, password is "foo{bar"odbc:server=localhost;user id=sa;password={foo}}bar}
// Escaped} with
}}`, password is "foo}bar"odbc:server=localhost;user id=sa;database=master;app name=MyAppName;krb5-configfile=path/to/file;krb5-credcachefile=path/to/cache;authenticator=krb5
odbc:server=localhost;user id=sa;database=master;app name=MyAppName;krb5-configfile=path/to/file;krb5-realm=domain.com;krb5-keytabfile=path/to/keytabfile;authenticator=krb5
Azure Active Directory authentication uses temporary authentication tokens to authenticate.
The mssql
package does not provide an implementation to obtain tokens: instead, import the azuread
package and use driver name azuresql
. This driver uses azidentity to acquire tokens using a variety of credential types.
To reduce friction in local development, ActiveDirectoryDefault
can authenticate as the user signed into the Azure CLI.
Run the following command to sign into the Azure CLI before running your application using the ActiveDirectoryDefault
connection string parameter:
az login
Azure CLI authentication isn't recommended for applications running in Azure. More details are available via the Azure authentication with the Azure Identity module for Go tutorial.
The credential type is determined by the new fedauth
connection string parameter.
fedauth=ActiveDirectoryServicePrincipal
orfedauth=ActiveDirectoryApplication
- authenticates using an Azure Active Directory application client ID and client secret or certificate. Implemented using ClientSecretCredential or CertificateCredentialclientcertpath=<path to certificate file>;password=<certificate password>
orpassword=<client secret>
user id=<application id>[@tenantid]
Note the@tenantid
component can be omitted if the server's tenant is the same as the application's tenant.
fedauth=ActiveDirectoryPassword
- authenticates using a user name and password.user id=username@domain
password=<password>
applicationclientid=<application id>
- This guid identifies an Azure Active Directory enterprise application that the AAD admin has approved for accessing Azure SQL database resources in the tenant. This driver does not have an associated application id of its own.
fedauth=ActiveDirectoryDefault
- authenticates using a chained set of credentials. The chain is built from EnvironmentCredential -> ManagedIdentityCredential->AzureCLICredential. See DefaultAzureCredential docs for instructions on setting up your host environment to use it. Using this option allows you to have the same connection string in a service deployment as on your interactive development machine.fedauth=ActiveDirectoryManagedIdentity
orfedauth=ActiveDirectoryMSI
- authenticates using a system-assigned or user-assigned Azure Managed Identity.user id=<identity id>
- optional id of user-assigned managed identity. If empty, system-assigned managed identity is used.resource id=<resource id>
- optional resource id of user-assigned managed identity. If empty, system-assigned managed identity or user id are used (if both user id and resource id are provided, resource id will be used)
fedauth=ActiveDirectoryInteractive
- authenticates using credentials acquired from an external web browser. Only suitable for use with human interaction.applicationclientid=<application id>
- This guid identifies an Azure Active Directory enterprise application that the AAD admin has approved for accessing Azure SQL database resources in the tenant. This driver does not have an associated application id of its own.
fedauth=ActiveDirectoryDeviceCode
- prints a message to stdout giving the user a URL and code to authenticate. Connection continues after user completes the login separately.fedauth=ActiveDirectoryAzCli
- reuses local authentication the user already performed using Azure CLI.
import (
"database/sql"
"net/url"
// Import the Azure AD driver module (also imports the regular driver package)
"github.com/microsoft/go-mssqldb/azuread"
)
func ConnectWithMSI() (*sql.DB, error) {
return sql.Open(azuread.DriverName, "sqlserver://azuresql.database.windows.net?database=yourdb&fedauth=ActiveDirectoryMSI")
}
To run a stored procedure, set the query text to the procedure name:
var account = "abc"
_, err := db.ExecContext(ctx, "sp_RunMe",
sql.Named("ID", 123),
sql.Named("Account", sql.Out{Dest: &account}),
)
To read output parameters from a stored procedure with resultset, make sure you read all the rows before reading the output parameters:
sqltextcreate := `
CREATE PROCEDURE spwithoutputandrows
@bitparam BIT OUTPUT
AS BEGIN
SET @bitparam = 1
SELECT 'Row 1'
END
`
var bitout int64
rows, err := db.QueryContext(ctx, "spwithoutputandrows", sql.Named("bitparam", sql.Out{Dest: &bitout}))
var strrow string
for rows.Next() {
err = rows.Scan(&strrow)
}
fmt.Printf("bitparam is %d", bitout)
Due to protocol limitations, temporary tables will only be allocated on the connection
as a result of executing a query with zero parameters. The following query
will, due to the use of a parameter, execute in its own session,
and #mytemp
will be de-allocated right away:
conn, err := pool.Conn(ctx)
defer conn.Close()
_, err := conn.ExecContext(ctx, "select @p1 as x into #mytemp", 1)
// at this point #mytemp is already dropped again as the session of the ExecContext is over
To work around this, always explicitly create the local temporary table in a query without any parameters. As a special case, the driver will then be able to execute the query directly on the connection-scoped session. The following example works:
conn, err := pool.Conn(ctx)
// Set us up so that temp table is always cleaned up, since conn.Close()
// merely returns conn to pool, rather than actually closing the connection.
defer func() {
_, _ = conn.ExecContext(ctx, "drop table #mytemp") // always clean up
conn.Close() // merely returns conn to pool
}()
// Since we not pass any parameters below, the query will execute on the scope of
// the connection and succeed in creating the table.
_, err := conn.ExecContext(ctx, "create table #mytemp ( x int )")
// #mytemp is now available even if you pass parameters
_, err := conn.ExecContext(ctx, "insert into #mytemp (x) values (@p1)", 1)
To get the procedure return status, pass into the parameters a
*mssql.ReturnStatus
. For example:
var rs mssql.ReturnStatus
_, err := db.ExecContext(ctx, "theproc", &rs)
log.Printf("status=%d", rs)
or
var rs mssql.ReturnStatus
_, err := db.QueryContext(ctx, "theproc", &rs)
for rows.Next() {
err = rows.Scan(&val)
}
log.Printf("status=%d", rs)
Limitation: ReturnStatus cannot be retrieved using QueryRow
.
The sqlserver
driver uses normal MS SQL Server syntax and expects parameters in
the sql query to be in the form of either @Name
or @p1
to @pN
(ordinal position).
db.QueryContext(ctx, `select * from t where ID = @ID and Name = @p2;`, sql.Named("ID", 6), "Bob")
To pass specific types to the query parameters, say varchar
or date
types,
you must convert the types to the type before passing in. The following types
are supported:
- string -> nvarchar
- mssql.VarChar -> varchar
- time.Time -> datetimeoffset or datetime (TDS version dependent)
- mssql.DateTime1 -> datetime
- mssql.DateTimeOffset -> datetimeoffset
- "github.com/golang-sql/civil".Date -> date
- "github.com/golang-sql/civil".DateTime -> datetime2
- "github.com/golang-sql/civil".Time -> time
- mssql.TVP -> Table Value Parameter (TDS version dependent)
Using an int
parameter will send a 4 byte value (int) from a 32bit app and an 8 byte value (bigint) from a 64bit app.
To make sure your integer parameter matches the size of the SQL parameter, use the appropriate sized type like int32
or int8
.
// If this is passed directly as a parameter,
// the SQL parameter generated would be nvarchar
name := "Bob"
// If the user_name is defined as varchar,
// it needs to be converted like this:
db.QueryContext(ctx, `select * from t2 where user_name = @p1;`, mssql.VarChar(name))
// Note: Mismatched data types on table and parameter may cause long running queries
The protocol and cryptography details for AE are detailed elsewhere.
To enable AE on a connection, set the ColumnEncryption
value to true on a config or pass columnencryption=true
in the connection string.
Decryption and encryption won't succeed, however, without also including a decryption key provider. To avoid code size impacts on non-AE applications, key providers are not included by default.
Include the local certificate providers:
import (
"github.com/microsoft/go-mssqldb/aecmk/localcert"
)
You can also instantiate a key provider directly in code and hand it to a Connector
instance.
c := mssql.NewConnectorConfig(myconfig)
c.RegisterCekProvider(providerName, MyProviderType{})
If the correct key provider is included in your application, decryption of encrypted cells happens automatically with no extra server round trips.
Encryption of parameters passed to Exec
and Query
variants requires an extra round trip per query to fetch the encryption metadata. If the error returned by a query attempt indicates a type mismatch between the parameter and the destination table, most likely your input type is not a strict match for the SQL Server data type of the destination. You may be using a Go string
when you need to use one of the driver-specific aliases like VarChar
or NVarCharMax
.
*** NOTE *** - Currently char
and varchar
types do not include a collation parameter component so can't be used for inserting encrypted values.
microsoft#129
Key provider configuration is managed separately without any properties in the connection string.
The pfx
provider exposes its instance as the variable PfxKeyProvider
. You can give it passwords for certificates using SetCertificatePassword(pathToCertificate, path)
. Use an empty string or "*"
as the path to use the same password for all certificates.
The MSSQL_CERTIFICATE_STORE
provider exposes its instance as the variable WindowsCertificateStoreKeyProvider
.
Both providers can be constrained to an allowed list of encryption key paths by appending paths to provider.AllowedLocations
.
Import this provider using github.com/microsoft/go-mssqldb/aecmk/akv
Constrain the provider to an allowed list of key vaults by appending vault host strings like "mykeyvault.vault.azure.net" to akv.KeyProvider.AllowedLocations
.
- LastInsertId should
not be used with this driver (or SQL Server) due to how the TDS protocol
works. Please use the OUTPUT Clause
or add a
select ID = convert(bigint, SCOPE_IDENTITY());
to the end of your query (ref SCOPE_IDENTITY). This will ensure you are getting the correct ID and will prevent a network round trip. - NewConnector may be used with OpenDB.
- Connector.SessionInitSQL may be set to set any driver specific session settings after the session has been reset. If empty the session will still be reset but use the database defaults in Go1.10+.
- Can be used with SQL Server 2005 or newer
- Can be used with Microsoft Azure SQL Database
- Can be used on all go supported platforms (e.g. Linux, Mac OS X and Windows)
- Supports new date/time types: date, time, datetime2, datetimeoffset
- Supports string parameters longer than 8000 characters
- Supports encryption using SSL/TLS
- Supports SQL Server and Windows Authentication
- Supports Single-Sign-On on Windows
- Supports connections to AlwaysOn Availability Group listeners, including re-direction to read-only replicas.
- Supports query notifications
- Supports Kerberos Authentication
- Supports handling the
uniqueidentifier
data type with theUniqueIdentifier
andNullUniqueIdentifier
go types - Pluggable Dialer implementations through
msdsn.ProtocolParsers
andmsdsn.ProtocolDialers
- A
namedpipe
package to support connections using named pipes (np:) on Windows - A
sharedmemory
package to support connections using shared memory (lpc:) on Windows - Dedicated Administrator Connection (DAC) is supported using
admin
protocol - Always Encrypted
MSSQL_CERTIFICATE_STORE
provider on Windowspfx
provider on Linux and Windows
go test
is used for testing. A running instance of MSSQL server is required.
Environment variables are used to pass login information.
Example:
env SQLSERVER_DSN=sqlserver://user:pass@hostname/instance?database=test1 go test
AZURESERVER_DSN
environment variable provides the connection string for Azure Active Directory-based authentication. If it's not set the AAD test will be skipped.
These features still exist in the driver, but they are are deprecated.
If you use the driver name "mssql" (rather then "sqlserver") the SQL text will be loosly parsed and an attempt to extract identifiers using one of
- ?
- ?nnn
- :nnn
- $nnn
will be used. This is not recommended with SQL Server.
There is at least one existing won't fix
issue with the query parsing.
Use the native "@Name" parameters instead with the "sqlserver" driver name.
-
SQL Server 2008 and 2008 R2 engine cannot handle login records when SSL encryption is not disabled. To fix SQL Server 2008 R2 issue, install SQL Server 2008 R2 Service Pack 2. To fix SQL Server 2008 issue, install Microsoft SQL Server 2008 Service Pack 3 and Cumulative update package 3 for SQL Server 2008 SP3. More information: http://support.microsoft.com/kb/2653857
-
Bulk copy does not yet support encrypting column values using Always Encrypted. Tracked in #127
This project is a fork of https://github.com/denisenkom/go-mssqldb and welcomes new and previous contributors. For more informaton on contributing to this project, please see Contributing.
For more information on the roadmap for go-mssqldb, project plans are available for viewing and discussion.
This project has adopted the Microsoft Open Source Code of Conduct.
Resources:
- Microsoft Open Source Code of Conduct
- Microsoft Code of Conduct FAQ
- Contact opencode@microsoft.com with questions or concerns