System Variables

Table of contents

General system setting variables


This system variable controls a database's default branch, defaulting to the checked out branch when the server started. For a database named mydb, this variable will be named mydb_default_branch. New sessions will connect to this branch by default.


This system variable controls logging levels in the server. Valid values are error, warn, info, debug, or trace. This value overrides whatever was specified on the command line for dolt sql-server or in the config.yaml file.


When set to 1, this system variable causes all branches to be represented as separate databases in show databases, the information_schema tables, and other places where databases are enumerated. Defaults to 0, which means that by default branch-derived databases are not displayed (although they can still be used).

fresh> show databases;
| Database           |
| fresh              |
| information_schema |
| mysql              |
3 rows in set (0.00 sec)

fresh> set @@dolt_show_branch_databases = 1;
fresh> show databases;
| Database           |
| fresh              |
| fresh/b1           |
| fresh/main         |
| information_schema |
| mysql              |
5 rows in set (0.00 sec)


When set to 1, this system variable causes all system tables to be show in show tables and in information_schema.tables. Defaults to 0.


When set to a commit hash, branch name, or tag name, Dolt will map all table data to the schema at the specified commit, branch, or tag. This is useful when you have a query that runs with a specific schema, and you want to run it with data that has a different schema. For example, if you add a Birthdate column to the People table in the most recent commits in your database, you cannot reference that column in queries run against older commits. If you enable schema overriding, and set @@dolt_override_schema to a commit that contains the Birthdate column, you can run the same query with recent commits and with older commits, without having to modify the query for the schema changes in the older commits. Dolt will map the table data to the schema at the specified commit, branch, or tag, and fill in the missing columns with NULL values.

-- check out an older branch that has a different schema
CALL dolt_checkout('olderBranch');

-- running a query that references the Birthdate column will fail
SELECT Name, Birthdate FROM People;
column "Birthdate" could not be found in any table in scope

-- turning on schema overriding allows us to automatically map our data to the schema at the specified commit
SET @@dolt_override_schema = 'main';
SELECT Name, Birthdate FROM People;
| Name      | Birthdate |
| Billy     | NULL      |
| Jimbo     | NULL      |

Note that when this session variable is set, the active Dolt session becomes read-only. To disable schema overriding, simply set this variable to NULL.


When set to 1, this system variable creates a Dolt commit for every SQL transaction commit. Defaults to 0. Commits have a standard commit message ("Transaction commit"), unless @@dolt_transaction_commit_message has been set.


When @@dolt_transaction_commit is enabled, if this system variable is set to a string, it will be used as the message for the automatic Dolt commit. Defaults to NULL, which means automatic Dolt commits will use their standard commit message ("Transaction commit").


When set to 1, Dolt will disable some extensions to MySQL behavior that are intended to increase compatibility with other database engines in the MySQL family. For example, for compatibility with MariaDB, Dolt supports an extension to MySQL's behavior that allows TEXT and BLOB columns to be used in unique indexes without specifying a prefix length. Users who want Dolt to behave exactly like MySQL and not support these extensions can set this system variable to 1. For wider compatibility, this system variable defaults to 0 to enable these extensions by default.


When set to 1, this system variable allows transactions with merge conflicts to be committed. When set to 0, merge conflicts must be resolved before committing a transaction, and attempting to commit a transaction with conflicts fails and rolls back the transaction. Defaults to 0.


When set to 1, this system variable ignores all merge conflicts, constraint violations, and other correctness issues resulting from a merge and allows them to be committed. Defaults to 0.

Replication variables


This system variable should be set on replication primaries to name a remote to replicate to. See Replication.

mysql> select name from dolt_remotes;
| name    |
| remote1 |
| origin  |
mysql> SET @@GLOBAL.dolt_replicate_to_remote = remote1;
mysql> CALL dolt_commit('-am', 'push on write');


This system variable can be set to 1 on replication primaries to make remote pushes asynchronous. This setting can cause commits to complete faster since the push to remote is not synchronous, but it may also increase the remote replication delay. See Replication.

mysql> SET @@GLOBAL.dolt_replicate_to_remote = remote1;
mysql> SET @@GLOBAL.dolt_async_replication = 1;


This system variable is set on read replicas to name a remote to pull from. New data is pulled every time a transaction begins.

Setting either dolt_replicate_heads or dolt_replicate_all_heads is also required for read replicas. See Replication.

mysql> SET @@GLOBAL.dolt_read_replica_remote = origin;
mysql> SET @@GLOBAL.dolt_replicate_heads = main;


This system variable indicates to pull all branches on a read replica at transaction start. Pair with dolt_read_replica_remote. Use is mutually exclusive with dolt_replicate_heads. See Replication.

mysql> SET @@GLOBAL.dolt_replicate_all_heads = 1;


This system variable specifies which branch heads a read replica will fetch. The wildcard * may be used to match zero or more characters in a branch name and is useful for selecting multiple branch names. Pair with dolt_read_replica_remote. Use is mutually exclusive with dolt_replicate_all_heads. See Replication.

mysql> SET @@GLOBAL.dolt_replicate_heads = main;
mysql> SET @@GLOBAL.dolt_replicate_heads = "main,feature1,feature2";
mysql> SET @@GLOBAL.dolt_replicate_heads = "main,release*";


This system variable indicates that newly created databases should have a remote created according to the URL template supplied. This URL template must include the {database} placeholder. Some examples:

set @@persist.dolt_replication_remote_url_template = 'file:///share/doltRemotes/{database}'; -- file based remote
set @@persist.dolt_replication_remote_url_template = 'aws://dynamo-table:s3-bucket/{database}'; -- AWS remote
set @@persist.dolt_replication_remote_url_template = 'gs://mybucket/remotes/{database}'; -- GCP remote

On a read replica, setting this variable will cause the server to attempt to clone any unknown database used in a query or connection string by constructing a remote URL and cloning from that remote. See Replication.


Set this variable to 1 to cause read replicas to always pull their local copies of remote heads even when they have diverged from the local copy, which can occur in the case of a dolt push -f. A setting of 0 causes read replicas to reject remote head updates that cannot be fast-forward merged into the local copy. Defaults to 1.


Set this variable to 1 to ignore replication errors on a read replica. Replication errors will log a warning rather than causing queries to fail. Defaults to 0.

mysql> SET @@GLOBAL.dolt_skip_replication_errors = 1;

Session metadata variables


Each session defines a system variable that controls the current session head. For a database called mydb, this variable will be called @@mydb_head_ref and be set to the current head.

mydb> select @@mydb_head_ref;
| @@SESSION.mydb_head_ref |
| refs/heads/master       |

You can set this session variable to switch your current head. Use either refs/heads/branchName or just branchName:

SET @@mydb_head_ref = 'feature-branch'

This is equivalent to:

call dolt_checkout('feature-branch')


This system variable reflects the current HEAD commit's hash. For a database called mydb, this variable will be called @@mydb_head. It is read-only.


This system variable reflects the current working root value's hash. For a database called mydb, this variable will be called @@mydb_working. Its value corresponds to the current working hash. Selecting it is useful for diagnostics. It is read-only.


This system variable reflects the current staged root value's hash. For a database called mydb, this variable will be called @@mydb_staged Selecting it is useful for diagnostics. It is read-only.

Persisting System Variables

Dolt supports a limited form of system variable persistence. The same way session variables can be changed with SET, global variables can be persisted to disk with SET PERSIST. Persisted system variables survive restarts, loading back into the global variables namespace on startup.

Dolt supports SET PERSIST and SET PERSIST_ONLY by writing system variables to the local .dolt/config.json. The same result can be achieved with the CLI by appending prefix to keys with the dolt config add --local command. System variables are used as session variables, and the SQL interface is the encouraged access point. Variables that affect server startup, like replication, must be set before instantiation.



SET PERSIST max_connections = 1000;
SET @@PERSIST.max_connections = 1000;


SET PERSIST_ONLY back_log = 1000;
SET @@PERSIST_ONLY.back_log = 1000;


$ dolt sql -q "set @@persist.max_connections = 1000"


Deleting variables with RESET PERSIST is not supported.

Last updated